DIYstompboxes.com

Hi,

So I'm messing with cassette tape, trying to get a playback amp working for an eventual tape echo. I saw there's a design on the forum already which I tried to recreate on the breadboard with no success - so, I thought I'd simplify a bit and stick to a simple op-amp design just to get me started making some noise.

I had a LM386 chip in the parts bin and hooked that up as in the following schematic:
(https://www.circuitbasics.com/wp-content/uploads/2015/04/Build-a-Great-Sounding-Audio-Amplifier-with-Bass-Boost-from-the-LM386-Minimal-Amplifier-Schematic.png)

Substituting in the tape head for the audio in jack.

All I get is noise, and occasionally a very weak signal from the head under the noise floor. If I increase the gain by adding a 10uF cap across the gain pins (1 and 8 ) I just get the same, but much louder. So I assume the amplifier is working as it should and something else is the problem. I also thought that schem was missing the virtual ground so I added a voltage divider on the input to see whether that made any change, but to no avail.

I realise this is not optimal for a tape head amp, as I need NAB equalisation and what not, but I'm confused as to why this doesn't work - should this work in theory, or am I being stupid somehow?

PS I've tried several transports and heads, and confirmed they work in their original circuits (I have a handful of portable tape players to tinker with).

Thanks,
-Adam

It should be ok.
Are you using a battery for the 386 power and is the voltage good when connected?
The 386 isn't an opamp so don't worry about virtual ground stuff. It's designed to be that simple.
If you've become used to hooking up opamps like the 741, it's incredibly easy to apply +power to pin 7 by automatic pilot. Check again: +power on pin 6!

If you get a weak signal with lot of noise and you can confirm that the contact with the tape head and cassette tape is "tight", meaning that there is proper tension in the tape transport to keep the tape firmly against the head without alignment issues, then I would suspect a grounding issue.

If there is a DC motor grounded to the transport mechanism and the tape head also grounded to the transport, the (most likely separate ground) of your (assumed breadboard) circuit might cause problems. In my build at first I got huge ground loop noise problems from using wall wart power supplies and in addition powering the guitar amp from mains.

To avoid noise from grounding problems, I found a solution to isolate the DC motor and the tape head grounding from the tape transport mechanism, so that tape head ground is the breadboard ground, and the DC motor is powered from separate battery with its own isolated ground point.

I would also like to know what design from this forum you could not get working? For tape head playback amplifier, (if you want) I could also share a reasonably simple discrete 3 transistor design, which will provide the required 50dB - 60dB (1000x) gain with relatively small background noise level.

Coincidence. I was just wondering whether I could make a small tape echo, and started pulling apart an old tape deck to have a look. Haven't progressed very far with the project though.

the 368 needs a 100uF capacitor from ground to 9v, and possibly one extra 100u from pin 7 to ground

cheers, Iain

Thanks for the replies;

QuoteThe 386 isn't an opamp so don't worry about virtual ground stuff. It's designed to be that simple.
If you've become used to hooking up opamps like the 741, it's incredibly easy to apply +power to pin 7 by automatic pilot. Check again: +power on pin 6!

Ah OK, yea I don't have any experience with the 386 and assumed it was an opamp like the TL072. I'll leave out the virtual ground stuff then. I'll double check the power hook up - I think I copied the pin assignment in that schematic I posted but yea it's possible I've hooked that up wrong, so worth another check, thanks!

QuoteI would also like to know what design from this forum you could not get working? For tape head playback amplifier, (if you want) I could also share a reasonably simple discrete 3 transistor design, which will provide the required 50dB - 60dB (1000x) gain with relatively small background noise level.

Jatalahd, the one I was copying was yours! It's here:
https://www.diystompboxes.com/smfforum/index.php?topic=125858.msg1201030#msg1201030

Really inspiring build. What I want to do is repeat what you did, and then take it a little further by trying to 3D print a replacement transport. I got a 3D printer and now I need a difficult music-related project for it to keep me busy for the next year :P The eventual plan would be to open-source it all so people can order a PCB of the electronics and 3D print the transport parts to make an easily serviceable cassette tape delay without relying on hacking old existing machines (other than sourcing heads).

I actually started with your earlier forum post with the simplified schematics coz it was easy for me to see the individual components. I got the bias oscillator working - well I think it works, I have one of those hobbyist scopes and it shows a nice sine wave, but the P2P voltage seems quite low (less than 1V) and it reports a frequency of only around 50Hz - but that might just be coz it's a bad scope - I figured I'd assume it's fine for now until I get a playback and recording amp working.

I hadn't realised the gain required was as high as 1000x, the current setup only has around 20->200 gain depending on how I setup the LM384, so perhaps this is a bad route anyway? I do have a LA3161 I can try next, which if I can make work would be fine for now, but I want to avoid parts that are no longer manufactured so other people can copy it easily.

I tried building the playback amp from your build a couple times on breadboard, and once on PCB to be sure, but they all failed for some reason. It's almost certainly something I'm doing wrong; I've never had much luck with analogue circuits for some reason - anything digital I build works fine but for some reason my analogue circuits have a high failure rate. I'm a software dev by trade so I find the digital end of things much easier to grasp - this is definitely going outside my comfort zone! In the long term I'd like to make something that has low parts count and uses easy-to-source parts so that there's no worry about transistors being hard to find in the future or dealing with counterfeit parts or what not, but for now I'll take anything that I can get to work!

QuoteIf you get a weak signal with lot of noise and you can confirm that the contact with the tape head and cassette tape is "tight", meaning that there is proper tension in the tape transport to keep the tape firmly against the head without alignment issues, then I would suspect a grounding issue.

So I think the transport/heads are all good in terms of them being in the right positions/tensions - the same setup works fine with the original electronics. I have two transports - one which is almost completely original (I just cut the wires to the head and soldered them to a breakout for connecting to the breadboard). The second setup is from another machine of the same make/model, but I dremel'd off all the case to fit a recording head and the motor is now connected to a STM32 dev board's PWM output via a mosfet driver.

On the grounds - I'll check that when I get home, I hadn't really paid attention to how the transport/motor/head are grounded to one-another, so that's one potential issue to sort! I'll also try running it on a 9V battery to see if that has any affect.

Powersupplies; of the two setups I have, they are like this:
-Setup 1; preamp on breadboard is powered by a 9V T-Rex pedalboard supply. Motor is powered by AA batteries (in original tape player casing).
-Setup 2: Same preamp on breadboard powered by the T-Rex 9V. Motor is powered by an STM32F dev board via mosfet driver - ultimately the power for this comes from the 5V USB from a PC. The motor is supplied at 3.3V - I did worry it would pull too much current for the USB but I tried it and it appears to work so I didn't worry about it  :icon_mrgreen:

Quotethe 368 needs a 100uF capacitor from ground to 9v, and possibly one extra 100u from pin 7 to ground

I hadn't really added any PS filtering and what not, so I'll make sure to add this tonight and see if that helps things, thanks!

Quote
The eventual plan would be to open-source it all so people can order a PCB of the electronics and 3D print the transport parts to make an easily serviceable cassette tape delay without relying on hacking old existing machines (other than sourcing heads).

Sounds like a one year project  ;) ! For background knowledge I would recommend reading the book "Audio Electronics" by John Linsley Hood. The tape recording part of it (1st chapter) is almost fully readable from the Google Books preview:

https://books.google.fi/books?id=EOE1tx3CJ98C&printsec=frontcover

I purchased the full ebook, because that seems to be the latest and most compact introduction to cassette tape recording with full circuit examples available. But after chapter 1, there is no more information on that topic...

Based on the example on that book (Sony playback amplifier) I modified a 9 volt single supply version of a simple (my opinion!) tape head playback amplifier. The topology is the same as in my tape echo circuit schematic, but here is less parts:

(https://i.postimg.cc/RN7yjFZg/playback-amp.png) (https://postimg.cc/RN7yjFZg)

The transistors are: Q1 and Q3 = BC559C ; Q2 BC549C

The frequency response with 4.7k resistor and 0.015uF capacitor included is:

(https://i.postimg.cc/ts7f7zzM/playback-freq.png) (https://postimg.cc/ts7f7zzM)

For linear 60 dB gain, you can leave out the branch with 4.7k resistor and 0.015uF capacitor. This depicted here is more closer to actual cassette player playback amp response, because typically the high-frequencies are boosted when recording. I have narrowed down the bandwidth here using the 3300 pF cap in the input. This is because I want to effectively filter out the bias frequency at this point. Obviously, to drive a loudspeaker, you need a power amplifier after this, I just use my guitar amp (which is actually a bass amp) for that.

It is not necessary to reach the 60 dB, but when trying to match the signal level obtained from a guitar, then the gain should be at least 50 dB.

There is also a good example on using one op-amp as the playback amplifier in the audio electronics book (page 34), but again it uses bipolar supply. Hopefully these examples get you forward.

Quote from: jatalahd on January 18, 2021, 11:50:16 AM...typically the high-frequencies are boosted when recording....

Actually mostly because the wound tape playback head works much better at high frequencies. We start with a complementary roll-off in the playback amp.

The LM389(?) app-sheet has a complete tape play amp plan. If you can't find the chip, it is LM386 plus three 2-penny NPN transistors.  https://rtellason.com/chipdata/lm389.pdf

Also, take a look at the National Semiconductor data sheet for the LM387, a stereo preamp in an 8 pin IC. I've made the NAB circuit in that data sheet and it works well. I have no idea how hard LM387 ICs are to find now...

QuoteBased on the example on that book (Sony playback amplifier) I modified a 9 volt single supply version of a simple (my opinion!) tape head playback amplifier. The topology is the same as in my tape echo circuit schematic, but here is less parts:

I have those parts to hand so I'll get started on breadboarding that tonight and let you know whether I have more luck!

QuoteThe LM389(?) app-sheet has a complete tape play amp plan

QuoteAlso, take a look at the National Semiconductor data sheet for the LM387

I'll see if I can hunt these down - I originally chose the 386 coz it's still available from wholesalers (at least as an SMD part), but it might be a good idea to have more options to try.

Ok so...

Attempted the design you posted, jatalahd. I get the same result - lots of noise and no signal. Tried battery power and filtering in the 386 design - this does lower the noise but still get either very low signal or no signal at all, seemingly at random.

Since I know the transports and heads are fine (as I have several and they all work in their own circuits) I'm trying to figure out what else could be wrong.

Thoughts are that either
A) the circuits are wrong on the breadboard.
B) components I have are faulty.
C) there are errors in the design

A) I'm not sure about this - it seems unlikely i'd get it wrong so many times in a row but it's possible I'm making some simple mistake over and over. Not sure where to go here.
B) I have a few of each part, but they're all from the same supplier. I've come across problems using TH parts before and suspected they're fakes, how likely is this though in this case?
C) seems unlikely the 3 or 4 designs I've tried would all have errors in that prevent the circuit working at all.

Any other ideas?

There have been fake LM386 chips around.
Some notes.
A healthy one will bias close to 1/2 supply voltage at its output pin 5.

The input is easy to work with. For simple experimenting, you may input to pin 2 or 3 (do you care if it's inverted or not?)
The unused input can be tied to 0v. Some say the inverting pin2 input is less noisy - but it can be an easier layout to use pin 2 anyway and ground pin3 next to pin4.

You don't have to have an input volume control. You can go straight in.

There can be a slight DC present on the inputs. It's probably not a good idea to get it on tape heads. Inputting via a capacitor makes things safer. The input impedance of the chip is 50k so any cap from 100nF up can be used.
You may well need the 10uF cap between pins 1 & 8 to boost gain, but if it gets too much, a 5k variable resistor in series with the 10uF can give a useful range of lower gains.

Pin7 allows an extra power filter cap to connect to 0v for the chips preamp section. It comes after an internal 15k resistor from the power input pin6 and there is little point having bigger than 10uF for this. Bigger caps give a slow startup when you apply power.

Pin 6 power should have at least 100uF capacitor to ground.

The low-value resistor and cap across the speaker output (a "Zobel" network) are to damp ringing caused by the speaker coil inductance. You can get away without it, but the ringing, which is usually supersonic, can have audible side effects so definitely worth fitting the network.

This simple amp chip can start to get complicated with all the bells & whistles added  ;D

Quote from: AdamB on January 21, 2021, 03:39:36 PM
- lots of noise and no signal.

Could you please describe how does the noise sound like? Loud "SHHHHHHHH" (hiss)? Loud "RRRRRRRRR" (motorboating)? Loud "WHEEEEEEEE" (whistling)? Or something else?

I have had the motorboating problem a couple of times with these high-gain feedback circuits, it happens because the circuit picks up some interference signal (most likely 50/60 Hz hum) and starts to self-oscillate covering all other sounds underneath. Quite often adding a 100 ohm resistor in series with the battery and a 1000uF cap to ground after the 100 ohm resistor cures that problem (a low-pass filter that filters almost all frequencies from the power supply line). Also the whistling noise refers to circuit self-oscillation.

If it is only loud hiss and no signal, then I would suspect a wiring error at some point.

From the design that I presented, only I can say that I have had that circuit on my breadboard for a month now, I have tested it using signal sources such as an audio signal generator, a tape head and with reduced gain I have used it as a microphone pre-amplifier. I remember that when I had the bias oscillator circuit on the same breadboard, I got some interference from that, but isolating the circuit to its own breadboard has worked well for me. But I understand that saying "it works for me" does not help anything  :(

Maybe the way to go at this point would be to get that LM386 circuit working and identifying where the noise appears. After the noise problem is out of the way, you could try other circuits as well.

Another tip.
If the tape head gives you separate terminals for both ends of its pickup coil, then a twisted pair of wires (or twin & screen cable) can directly connect across pins 2 & 3 of the 386 giving it a balanced drive. Any DC voltage on the pins will be very similar at both ends of the coil so won't do anything. This will also give it some freedom from common ground hum/noise.

Thanks for the notes anotherjim, I'll re-try the 386 design and check with your info as I go to see if that gets me closer! The only issue with the 386 is that it doesn't provide enough gain (it maxes out at 200). But, I could combine two of them (one set to gain 20, and the next one set to gain 50) and that would give me my 1000x gain. So if I can get one of them working without the cazy noise, then I could just bolt a second one on,

QuoteCould you please describe how does the noise sound like? Loud "SHHHHHHHH" (hiss)? Loud "RRRRRRRRR" (motorboating)? Loud "WHEEEEEEEE" (whistling)? Or something else?

With your design just get a quiet buzzing - this might just be background EM it's picking up (as there's no signal coming through it's hard for me to know how loud the noise is, as it's being amplified through my interface's instrument input so there's no frame of reference for me). But, here's a video I took after finishing your circuit on the breadboard;



It's the tape machine with the head connected to a breadboard of the preamp circuit you posted above^ Towards the end of the video I crank the gain on the interface so you can hear the noise clearly over the sound of the tape transport whirring away. It just sounds like typical background EM.

When used with the 386, instead it's a loud Shhhhhhh noise, occasionally I can hear the tape signal underneath quietly but it's not reliable.

QuoteFrom the design that I presented, only I can say that I have had that circuit on my breadboard for a month now, I have tested it using signal sources such as an audio signal generator, a tape head and with reduced gain I have used it as a microphone pre-amplifier. I remember that when I had the bias oscillator circuit on the same breadboard, I got some interference from that, but isolating the circuit to its own breadboard has worked well for me. But I understand that saying "it works for me" does not help anything  :(

Sure, it seems unlikely a problem with your schematics as I've tried 2 of your designs and 1 other using the 386 and all have failed in one way or another - most likely is that I'm making a mistake in breadboarding the designs and next likely is that the parts are in some way faulty.

I'll try both the 386 on one board and re-do your design on another side-by-side, so that way there's 2 chances I can get something working and figure out why I've been having problems.

QuoteAnother tip.
If the tape head gives you separate terminals for both ends of its pickup coil, then a twisted pair of wires (or twin & screen cable) can directly connect across pins 2 & 3 of the 386 giving it a balanced drive. Any DC voltage on the pins will be very similar at both ends of the coil so won't do anything. This will also give it some freedom from common ground hum/noise.

You're saying put the "left channel" of the head into + and the "right channel" into - of the LM386? Twisting them together to ensure they pickup the same interference pattern? Or have I misunderstood?

Thank you for the video. Regarding the design I shared, it sounds like an error in the connections. In the video I only hear normal background hum, which implies that the signal path is broken somewhere (or my ears are damaged).

I could not trace all your connections properly from the video, but please check the polarity of the second electrolytic cap. The minus side (-) should be towards the 100 ohm resistor. I could not see this properly from the video (my eyes are not that good either).

I also made another breadboard reference build to show you my layout. I can confirm that this build is working. The transistor order is the same as in the schematic from left to right. In the image below, the PNP transistors BC559C are facing south (downward) and the NPN transistor BC549C in the middle is facing north (upward). The image should have a decent resolution (if the service shows the original size) so you can zoom in to see the details better, also the polarity of the electrolytic caps.

(https://i.postimg.cc/PCndpRWK/IMG-20210122-213928.jpg) (https://postimg.cc/PCndpRWK)

The black/green/yellow wires disappearing to the top left are the tape head connection wires. The emitter of Q3 biases at 4.8 volts in this build, it can be measured for example from the left side of the output capacitor.

Hopefully this helps ...

Quote from: AdamB on January 22, 2021, 12:05:30 PM

QuoteAnother tip.
If the tape head gives you separate terminals for both ends of its pickup coil, then a twisted pair of wires (or twin & screen cable) can directly connect across pins 2 & 3 of the 386 giving it a balanced drive. Any DC voltage on the pins will be very similar at both ends of the coil so won't do anything. This will also give it some freedom from common ground hum/noise.

You're saying put the "left channel" of the head into + and the "right channel" into - of the LM386? Twisting them together to ensure they pickup the same interference pattern? Or have I misunderstood?

No, the idea needs each head channel to have 2 pins for both ends of each coil. So there are 4 pins total on the back of a stereo head. It would apply to either left or right channels, not both. You can't do it if there are only 3 pins for a common ground for both coils. Some heads have a little PCB over the pins that commons the grounds and I wouldn't recommend unsoldering the PCB from the head if yours has one.
What I'm suggesting is to treat it like a dynamic microphone with balanced XLR connection. That too feeds a +input and -input of a preamp and the grounded screen only provides screening. The twisted pair does do some noise reduction. The balanced connection also gives x2 extra amplitude before any amplification.

There are quite a few different tape preamp chips that have been sold on ready built PCB's. They seem to mostly be on auction sites now.

QuoteI also made another breadboard reference build to show you my layout. I can confirm that this build is working. The transistor order is the same as in the schematic from left to right. In the image below, the PNP transistors BC559C are facing south (downward) and the NPN transistor BC549C in the middle is facing north (upward). The image should have a decent resolution (if the service shows the original size) so you can zoom in to see the details better, also the polarity of the electrolytic caps.

Thanks for this, a massive help. I'm rebuilding my breadboard to be a replica of your setup, if I have trouble I'll post a picture and we can play spot the difference!

QuoteNo, the idea needs each head channel to have 2 pins for both ends of each coil. So there are 4 pins total on the back of a stereo head. It would apply to either left or right channels, not both. You can't do it if there are only 3 pins for a common ground for both coils. Some heads have a little PCB over the pins that commons the grounds and I wouldn't recommend unsoldering the PCB from the head if yours has one.
What I'm suggesting is to treat it like a dynamic microphone with balanced XLR connection. That too feeds a +input and -input of a preamp and the grounded screen only provides screening. The twisted pair does do some noise reduction. The balanced connection also gives x2 extra amplitude before any amplification.

The heads in the machines I have do have 4 pins, but the grounds are bridged together. However, I have a bunch of other heads I grabbed from ebay and also from out of car cassette adapters - some of these have the 4 pins separated so should be able to give your idea a go with one of those.

OK I copied your breadboard layout exactly, jatalahd. And it works!

https://www.dropbox.com/s/15mhpqbpp8r0jkt/PXL_20210204_202057707.jpg?raw=1 (https://www.dropbox.com/s/15mhpqbpp8r0jkt/PXL_20210204_202057707.jpg?raw=1)

I get a decent signal out of it, must of been me getting something wrong translating the schematic to breadboard. Next up I need to build the recording amp and check the bias oscillator I have on breadboard already works - once I have that going I'll try converting it all to a PCB layout and I'll post the gerbers and BOM for that here once verified. I figure even without the 3D printing stuff that might be useful for people to use as a drop-in replacement for making their own cassette delays.

In other news, my 3D printer is built and ready to start printing. I feel that end of things is going to be much harder - I'm gonna try printing my first calibration cube in the next couple of days, wish me luck!

I think for the first iteration I will print the enclosure, reels, motor mount and some transport parts and use pre-made gears (bitsbox has a bag of gears of various sizes I picked up to play around with). I think the gears are going to be the hardest part because the teeth are small and it probably needs printing in nylon, so I'll get on to that once I've got some printing experience and proven the rest of it can work.

Edit: changed image to link coz it was massive lol.

Nice that you got it working! Meanwhile I integrated that playback amplifier with some minor component value changes to replace the old one in my tape echo prototype and made a v2.0 schematic:

(https://i.postimg.cc/qtpDL1KD/complete-1600x1200.png) (https://postimg.cc/qtpDL1KD)

I added also a buffer after the playback amp, because one needs to take care of attenuating the signal in the reverse direction. I also added more filtering to the playback amp and one more filter cap to the recording amp. The problem is that when the playback head and record head are near each other, the record bias signal is induced to the play head (also the wires must be separate!). This induced bias is then amplifier in the playback amplifier and in the worst case takes the output to the rails, muting the original signal from the tape.

I am interested to see your 3D printing results, so please keep us posted on your progress on the project. If you have some problems with the bias oscillator or the recording amplifier, I am happy to help. The recording amplifier can be made simpler, but since I wanted to add the high-freq boost, and the need to filter the bias frequency, and my limited skills, I cannot figure out a better one than the existing is.

Thanks for the full schem. Will try my best to breadboard bits so that I understand what's going on, and then I'll PCB-it-up!

The buffer, is that the section around Q5?

QuoteThis induced bias is then amplifier in the playback amplifier and in the worst case takes the output to the rails

So I guess this goes for the erase head, too? I guess that can go on the other side of the transport, far away from the other 2 heads. When I make up the PCB I'll be sure to route the connections to opposite sides of the boards to make it easy to keep the head wiring away from each other.

If I'm understanding the schematics right, the erase head would replace the inductor L2, right?

-Adam

Quote
The buffer, is that the section around Q5?

Yes, Q5 and R14 makes the buffer, then R17 + R18 is the feedback level control. Making R17 smaller, you will attenuate less on each echo repeat, making it too small will cause too much feedback and that makes the echo boost up into oscillation. If the feedback path is investigated from the guitar input buffer towards the playback amplifier output, you want there to be attenuation, so that the guitar signal does not interfere with the output. This is taken care of in the schematic, where the attenuation is in the ballpark of 1/100.

It might also work without the buffer, the output impedance of the playback amplifier without buffer is around 1k, which is not that bad, but I like to have it lower. With buffer it is less than 100 ohms. if you want to experiment leaving the buffer out, then just neglect Q5 and R14.

Quote
So I guess this goes for the erase head, too? I guess that can go on the other side of the transport, far away from the other 2 heads. When I make up the PCB I'll be sure to route the connections to opposite sides of the boards to make it easy to keep the head wiring away from each other.

Yes, erase head will radiate the bias frequency as well so keep it far away from play head. This is what I experienced when measuring the circuit with the oscilloscope and moving the heads around. This problem does not exist on a normal tape recorder, where the play and record are two separate functions and the bias signal will not stick to tape when recording.

Quote
If I'm understanding the schematics right, the erase head would replace the inductor L2, right?

Yes, correct. I have not managed to find any erase heads anywhere, so I have not had the change to measure how the signal behaves with that.  Actually, after I finished my build, I found this guy on youtube, who had used the same bias circuit with a real erase head. He got the bias frequency almost to 100kHz. He also incorporated the bias trap between the recording amp and bias oscillator, but I am not sure did he tune it correctly. In my schematic, the use of bias trap is avoided using the same reverse path attenuation method that I explained on the playback amplifier. This is beneficial in the sense that it can be built to work without tuning the bias trap with an oscilloscope, but somehow I feel that I don't get the same recording level (volume) as I would get when recording with a normal tape deck, without distorting the signal that is.

So here is the video of the guy making his tape recorder with GOOD bias :)



If you have more questions, just post them here and I (or someone else) will try to answer them :)

Oh and there was one mistake in my new schematic regarding connection of Q4 base junction. I edited the post #20 above and added a link to corrected image. Now it should be ok. If you saved the previous image, please delete that and take this new one from post #20.

OK thanks for all the help, I'll be sure to update on here with progress - and there will definitely be questions. The recording/bias part seems a bit more involved so no doubt I'll have trouble with that, too!

I managed to get my first print done - it came out better than I expected it to. I printed the cliche benchy that everyone does. Really happy with the results for a first print, pic linked here:
https://www.dropbox.com/s/vx9b8lo7vupstfu/PXL_20210205_160736889.jpg?raw=1 (https://www.dropbox.com/s/vx9b8lo7vupstfu/PXL_20210205_160736889.jpg?raw=1)

There's the tinniest amount of stringing in there, but yea pretty happy with that. Next I'll try to print a simple mock up of a motor mount or a tape reel to see how accurately I can make the simpler parts and what degree of post processing might be needed. I'm not very experienced with CAD so I think there's gonna be a lot of printing small pieces until I get the dimensions etc. right.

OK had a first crack at getting the bias/recording amp working and.... no luck. I hear the whirring of the motor and some general RF noise, and a very faint dry guitar signal - which in itself is interesting as the dry isn't hooked up so we should only be hearing the delayed signal.

We know that my playback breadboard is working fine (I left that as-is), so it's either something with my breadboard of the recording/bias or something about the recording head/transport.

Here's a picture of the setup:
https://www.dropbox.com/s/9hzak4wuz1q5bkt/fullsetup.jpg?raw=1 (https://www.dropbox.com/s/9hzak4wuz1q5bkt/fullsetup.jpg?raw=1)

And here's a picture of the recording amp (with bias oscillator on a separate little breadboard above):
https://www.dropbox.com/s/c5va2d77qn7qsvj/recording_bias_breadboards.jpg?raw=1 (https://www.dropbox.com/s/c5va2d77qn7qsvj/recording_bias_breadboards.jpg?raw=1)

I changed the input buffer to a TL072 unity gain setup because I didn't have enough transistors to go around, I shouldn't think that causes any issues though. I then fed that straight into the recording/bias circuit from your schematic, leaving off the connection between the output of the playback amp and the input of the recording amp (for zero feedback).

One obvious issue is that the recording head is loose - I'm playing some guitar one-handed while I hold the record head over the tape with my other hand - not ideal but the shims wouldn't hold it in place so I figure this would be fine for now just to show if the circuit is working. I'd expect it to not give me a fantastic signal but I'd expect to at least hear something. I also connected both L and R coils for the recording head together so that it writes to both channels. The playback head is just shimmed and hot-glued into the blank tape, it has a tape guide on the head and a soft pad to press the tape onto so I figure this is likely fine.

Anything else obvious I'm doing wrong here? Jatalahd, do you have any voltage readings I can check over on the recording amp?

Maybe a stupid question, but be sure the heads are the right way up. May be so long since this tech was common but cassette only uses half the width of the tape. It looks like you have swapped heads for recording/playback as separate tests.
Also, your breadboarding is much neater than mine. That could be where you're going wrong  :icon_rolleyes:

Being serious, can you pickup audio on the recording head (and if you have a scope can you also see the bias signal).

QuoteMaybe a stupid question, but be sure the heads are the right way up.

Good question - I think so. One of the reasons I shimmed and glued in the playback head was so that I could turn it around. The recording head has this long PCB attached to it, so it would only go in the opposite way around to how the play head was mounted on the transport. So yea I just took the playhead out, turned it around and glued it in :P

I did try quickly turning the rec head around and trying to get it in to position and it didn't help (though it's hard to get it in position this way).

My breadboarding's only so neat because I'm copying the way Jatalahd did his breadboard :P

QuoteBeing serious, can you pickup audio on the recording head (and if you have a scope can you also see the bias signal).

I can pickup audio on the recording head (if I stick the recording head wires into the playback amps input and then press the head up to a running cassette with some audio on).So hopefully that means the recording head is all good.

I do have a scope, but it's a cheap hobbyist one and it doesn't seem to go up high enough in frequency to pickup the 50Khz signal from the oscillator. It does show a sine signal at some settings but reads it's frequency out as around 50Hz (I guess it's aliasing?).

In theory, if you put the heads together face to face, you should hear the recording head signal back through the playback head - like a crude transformer coupling. I would be interested to know if that works.

You should be able to see which side of the tape the heads match. Since the deck never knows which side of the tape it's looking at, as long as the heads are all aligned with the same side of the tape, it should be good.

As to fixing the heads on the deck. You will normall see a fixed screw on one side of the head and a spring-loaded adjustment screw. The adjustment should allow you to tilt the head one way or the other. This makes a huge difference to sound quality. The heads usually have an elongated or slotted screw hole on one side for the adjustable one.
Look up "tape head azimuth setting". Some of the write-ups can get very technical and call for special equipment.

The easy way is to use a recorded tape you know sounds ok and do the adjustment while playing it. Then change to blank tape and record to it while listening to the playback. Adjust the recording head tilt to find the best playback sound.

QuoteIn theory, if you put the heads together face to face, you should hear the recording head signal back through the playback head - like a crude transformer coupling. I would be interested to know if that works.

That doesn't work, so suggests something bad in my recording amp right?

QuoteYou should be able to see which side of the tape the heads match. Since the deck never knows which side of the tape it's looking at, as long as the heads are all aligned with the same side of the tape, it should be good.

On the one in the pics above I can't see the head because it's shimmed/glued into the tape lol. But when I did the pressing-heads-together test I used a different unit where I can see which way round the head is.

QuoteAs to fixing the heads on the deck. You will normall see a fixed screw on one side of the head and a spring-loaded adjustment screw. The adjustment should allow you to tilt the head one way or the other. This makes a huge difference to sound quality. The heads usually have an elongated or slotted screw hole on one side for the adjustable one.
Look up "tape head azimuth setting". Some of the write-ups can get very technical and call for special equipment.

For now I'm not too worried about quality, as I'm going for a low-fi kinda vibe anyhow, but certainly once I've gotten the first unit working I could look at how to include adjustment for azimuth to optimise that.

While I work on getting this recording amp going, I've been messing with CAD and printing stuff. Here's a quick test print of a reel and spindle. If I work on individual parts and then figure out how to integrate it all and get the quality up I think that's a good strategy.

That is good progress that you have the whole setup laid out. It needs to be noted that each circuit block (playback, record and bias) can be tested as separate entities, before hooking them together.

Unfortunately it is getting close to my bed time, so I can only give quick instructions.

1. As said, the record amplifier can be tested separately. Just connect guitar to the input and output to guitar amp. If signal comes out amplified through the recording amp, then it is working sufficiently (at this point). You basically need only about 12 dB (x4) gain, maybe a simple unity gain buffer is also enough at this point.

2. You have wiring errors in your bias oscillator. That is why it won't work. I re-built a relatively clean breadboard layout for you, it is the one on the bottom part of the board in the image (rows A to E). The top part is the leftovers from my earlier proto:

(https://i.postimg.cc/5Yt2LJDS/y.jpg) (https://postimg.cc/5Yt2LJDS)

The red one is the 1uF cap, the blue ones are the 100 nF caps, the gray one is clearly 10 nF, and the last green one is 100 pF cap. I built it with a 1 k resistor without the bias pot, for clarity. With this setup you get a very decent output of 41 Volts (rms):

(https://i.postimg.cc/BX248tXG/x.jpg) (https://postimg.cc/BX248tXG)

You don't need a scope to measure the output voltage. A normal multimeter with AC measuring works fine. It directly shows the rms voltage, but because of limited probe impedance it will show only 25 to 30 volts rms for the output.

Once you get the bias oscillator running (and some signal out of the recording amp) let's hook them together. I have been still testing my own circuit and as I mentioned about the bias trap, I included that to my proto and it made things better. The recording amp seems to load the bias oscillator signal down, but even with an untuned bias trap, I mostly got rid of the loading.

But lets get back to this later. Time to sleep ...

Thanks for all this Jatalahd, really appreciated.

So I tried isolating the record amp and going straight into the guitar amp with it, that didn't work. I know how to build a simple 4x gain opamp stage with that tl072 so I'll do that for now and revisit the record amp once the system is working.

So next thing to do for me is fix the oscillator - I'll copy your layout for that. I don't suppose you could you point out the error on my breadboard, I can't see it and it'd help me to learn if I can figure out where I'm going wrong with copying over the schematics.

Thanks,
-Adam

Ah I think I see the problem with my oscillator breadboard - the missing connection between the inductor and the 1uF cap right? I'll rebuild it anyhow to be sure!

Yes, the inductor connection was missing and also I think you were missing the ground connection on the emitter of the npn transistor. It is easy to make mistakes like this, that is why I always triple-check my breadboard builds before testing them. And usually after that there still might be issues ... So just taking the time and being patient would be the way to go.

Op-amp gain stage is good for the recording amp, it has a low output impedance, so when connecting the output of the record amp with a 10k resistor to the junction of the record head and bias oscillator output, you will get a proper reverse signal path attenuation when looking from the bias oscillator towards the record amplifier output.

Hopefully you will get the circuit working soon.

I would still like to discuss about adding the bias trap, but that can be done later.

I haven't had chance to work on the electronics more yet, however made some progress on the transport end of things. I've decided to try and make it work using standard cassettes (ones ideally converted into short tape loops). I figure this makes it easier to construct, though it means I'm stuck with the 3-in-a-line head positioning that will fit into the top of the cassette. But anyhoo...

BEHOLD! A 3D printed motor mount and flywheel assembly!



I notice that the rpm even without the gearing from an ordinary transport is actually not all that fast - it's probably about the top speed you'd want a tape delay running at. What this leads me to thinking is that as a first try I might just mount the spindle on the flywheel and be done with it - not sure what that'll do to the motors stability when running at lower speeds as I guess it'll have less torque but if I can keep the 3D printed parts simple it makes it easier for other people to put together.

Here it is with the spindle mounted on the other end of the flywheel shaft:
https://www.dropbox.com/s/39iwkg9y7d3f050/simple_assembly.jpg?raw=1 (https://www.dropbox.com/s/39iwkg9y7d3f050/simple_assembly.jpg?raw=1)

And boom, here's it driving a tape reel:


I've been doing some design work on how this might connect to a printable transport mount, it looks something like this currently:
https://www.dropbox.com/s/fnr3rxtzvz4h0ol/transport%20mount.png?raw=1 (https://www.dropbox.com/s/fnr3rxtzvz4h0ol/transport%20mount.png?raw=1)

I'll try this and see how it lines up - though I think it might be a better idea to rework it so that it can be mounted inside a 1590BB enclosure, with a couple holes drilled to pass the spindles through the top of the enclosure, so that the tape is attached to the top outside of the enclosure. That'd look neat.

A frustrating thing about this project is that you can only click like once.
Keep going. This is popcorn munching stuff, and I don't like popcorn.

Looks very nice and professional. The higher the tape speed is, the better high frequency response will be (up to certain limit of course). But I am a bit afraid that turning the tape reel at the middle directly from the motor drive will cause pretty big wow and flutter problems, even if the motor would be using a speed stabilization circuit.

When the "traditional" pinch roller mechanism is used in tape transport, the very small variation in the motor speed is not so noticeable, since the radius of the capstan is so small and the motion is converted into linear pull of the tape due to the pinch roller with larger radius. In your current implementation, the large radius of the tape reel driven from the center will "multiply" the variations in the motor speed when considering variations in the angular velocity at the outer edge of the tape reel, which directly affects the linear motion of the tape itself.

Not sure if my writing above made any sense, but definitely looking forward to see the progress of this project.

QuoteKeep going. This is popcorn munching stuff, and I don't like popcorn

I'll definitely give it a good try!

QuoteBut I am a bit afraid that turning the tape reel at the middle directly from the motor drive will cause pretty big wow and flutter problems

Indeed - the original plan was to create a completely custom transport that could use a pinch roller setup and 3d printed reels, but I decided for now that using a standard(ish) cassette allows for interesting possibilities (such as different "prepared" cassettes with damaged tape and the like for making whacky sounds that can easily be swapped out) as well as making construction easier.

The problem with cassettes though is that if I want an erase head then there's nowhere to stick a pinch roller.

I'm a little worried that it won't have the torque to pull the tape through once all 3 heads are in place, but we'll see I suppose. If it has a lot of wow and flutter that's not necessarily a problem - I'm going for a lofi warts-and-all tape sound. One of the reasons I'm using a microcontroller to drive the motor via PWM is so that I can modulate the motor speed in interesting ways. I'll likely add a "broken" knob that controls how erratic the motor speed is.

More more progress on the transport, here's the head engage/release mechanism.



It works using two 3d printed plungers (engage and release) that are mounted on metal springs and sandwiched between 3 layers of 3d printed plastic. The head just screws directly onto the engage plunger.

Here's a CAD picture showing a cutaway of how it works:
https://www.dropbox.com/s/v4stkpn9yt9svu5/tape%20head%20mechanism.png?raw=1

The head is a couple millimetres too far off the tape when engaged, so I need to shorten that plunger and re-print. It also jams sometimes, so I've ordered some PTFE lubricant to put on it to hopefully resolve that. I might also beef it up and make all the parts a bit thicker just to make it feel a bit more sturdy as the PLA I'm printing in is a bit too flexible - though I'll likely switch to printing in PETG once the design is finalised to add extra rigidity.

There's only one head (playback) mounted at the moment, but once I've had chance to test it working playing back a tape via the playback amp, I'll change the design to mount all 3 heads on that same plunger. It might need a couple extra "runners" left and right for the plunger to make it more stable when engaged but that shouldn't be too much of a problem.

Very clever stuff.

From the drawing, it looks like a 45deg angle between the movement faces. As only the head needs to push the release slider across and not the other way around, a shallower angle in favour of the head slide could be better?

The engage/release mechanism looks a bit complicated, and suitable springs might be hard to find in different parts of the globe if you are planning to release this as a DIY project. I would suggest using a simple screw mechanism, which would make the design more versatile considering different head heights and optimal head mounting depth with respect to the cassette/tape. As the tape heads also have different proportions all around, all heads might not align properly in the middle of the tape, so some fine tuning option should be considered there.

Also I would suggest using separate mounting mechanism for all heads, or at least a fine tuning option for each. For optimal performance you could consider how to include proper adjustment for the azimuth on that engage plunger.

QuoteFrom the drawing, it looks like a 45deg angle between the movement faces. As only the head needs to push the release slider across and not the other way around, a shallower angle in favour of the head slide could be better?

Yea that's a good idea, I'll play with that. So far, I've found its only the release plunger that jams, and only on the way out (the spring doesn't have enough force to overcome the friction with the bottom and top layers it's sandwiched between). I've made the plunger a tad thinner to give it some breathing room and also sanding it a little (as I'm not currently post-processing my prints at all) will probably help, to give it a smooth surface. And the lubricant will also help hopefully.

QuoteI would suggest using a simple screw mechanism

That's an idea - something I'll think about. My issue with screw mechanisms is that they are a pain to print with. If you print a screw mechanism that uses metal screws but also needs to be adjusted/re-used, then you have to use threaded inserts which need a good deal of plastic around them to make them stable. This increases the size of the part quite a bit around where the insert fits. If you don't use inserts the screws just strip the thread in the plastic very quickly. I think it may also be the case that the screws are only fully secure when they reach the end of their travel. Alternatively, if we printed a screw in plastic, they tend to need to be quite big and suffer from wear quite a lot - I'd probably have to switch to something slippery and tough (but hard to print with) like nylon.

I'd also worry about how quickly you could undo the screws/re-fasten them to switch tapes - seems like it wouldn't be something you could do between songs at a gig.

For finding springs - we might be able to 3D print a "spring" mechanism - similar to how snap-to-close enclosures work; just a thin U shaped sprung bit of plastic. One of the tape transports I took apart used this type of spring on a release plunger.

I am a little concerned about other people being able to just print it off and mount their parts on (heads especially) - as you say heads have different mountings and dimensions. I've seen people release designs where the user can fill in some values to customize the print before they do it, so maybe we can do this (I think it's called parametric modelling). But yea I've no idea how to do that yet - I'll figure that out once my own one works :P

I would like to add some way to adjust the head azimuth, but I think for the first version we just need to get it working, then we can start getting fancy once v1 is done. I've no experience with CAD so it's quite hard going designing mechanisms that will print and go together properly! I'd rather keep it simple for now - perfection is the enemy of progress.

Sometimes I see sliding parts where they make most of it too narrow and have a few lobes sticking out that keep it centred. Presumably, that reduces sliding friction a lot.

QuoteSometimes I see sliding parts where they make most of it too narrow and have a few lobes sticking out that keep it centred. Presumably, that reduces sliding friction a lot.

Yea I've definitely seen that approach in other transports I've disassembled. The problem is making it printable - coz to reduce friction as much as possible you'd want the lobes sticking out top and bottom, which when you print would mean that the bulk of the print is off of the print bed (an overhang), which will either fail to print (coz the extruder is printing into the air - there's nothing for the plastic to bond to) or you need to add supports to the print which I'm trying to avoid. But definitely room for improvement in there - I could make at least 1 side have less friction by keeping the base of the plunger flat and then putting little lobes on the top surface.

You do have two planes to work in. Bumps on the wide horizontal sliding surfaces and nodes on the narrow vertical surfaces. An advantage is you can make it with tight tolerances and correct any tightness with a flick of a file.

Another latch type comes to mind. The head slide can have a built-in springy finger. When it slides in, the fingertip springs into a recess and locks it. The release pushes the fingertip out of the recess. The release doesn't actually need a spring, just preventing it from falling out. The head slide still needs a spring.

Hello, I've decided to try build my own tape echo also. I've used the EP3 preamp and playback circuit and have tried the bias oscillator provided by jatalahd, however I can't seem to get it to work. I've tested all parts of my circuit, it seems the tape playback, preamp and recorder amplifier all work but as soon as I turn on the bias oscillator there is a very loud noise that drowns out everything else. I'm regulating down 24V to power the oscillator and recorder amplifier and it seems to be getting hot even with a heat sink, as are the transistors in the bias oscillator.

I directly copied the breadboard layout so I'm not entirely sure what's gone wrong  :-[ ! It seems that the oscillator is working as I'm getting a decently high voltage at the output but the noise is unbearable, if anyone has any ideas it'd be much appreciated!

Ronan.

Quote from: RLawlor on March 25, 2021, 04:00:48 PM
If anyone has any ideas it'd be much appreciated!

In my first breadboard prototype builds, I had some problems with loud motorboating low-frequency oscillation, but the reason was not the bias oscillator, only the high-gain playback amplifier was going crazy. But in EP3 playback amplifier the gain is quite modest, so cannot directly see that as a problem. For me the problem was eventually in the power source filtering and physical connections (jump wire lengths, distance between wires) in the breadboard. After soldering the build to veroboard, I have not had any problems, but I have only used battery power in my build. I would assume a mains-connected PSU would cause noise problems.

Without more detailed information about the problem, I can just give some general advice to isolate it (assuming that the noise is heard in the output of the playback amplifier):

- enhance the psu noise filtering in the playback amplifier to cut out lower frequencies (bigger filtering cap)
- add a 3.3 nF cap at the base of Q1 in the playback amplifier, to see if any difference in noise
- use a star grounding scheme, and connect Vdc+ further away from the ground connection
- keep bias oscillator components and wires far away from rest of the circuit
- test the circuit without the playback head connected, if noise goes away, then the head is picking up the noise from ground. Try the same also with the record head.

Also, as you noticed, running the bias oscillator with 24V is not a good idea. I would suggest 12V maximum.

Quick update on where I'm at.

I found that driving the tape with just the take up reel didn't really work. There was too much friction with the 3 heads for the motor to pull the tape through unless the motor was at a high RPM - which meant the useable delay times would be quite limited to short times - and it just wasn't very reliable.

So, for now I've ditched the cassette. The problem is that as Jatalahd suggested, I need a pinch roller to tension the tape, give traction and pull the tape through the heads - but I can't do that with a cassette and have an erase head due to lack of room.

So this is the latest prototype (which still doesn't work lol):
https://www.dropbox.com/s/mhzjl34t01o24n8/PXL_20210330_082838824.jpg?raw=1

The metal shaft (top left) that presses against the roller is powered, as well as the reel. This is achieved using a double-pulley system. I will at some point probably need to play around with different pulley size ratios (I imagine the reel needs to run slightly faster than the roller in order to tension the tape) but for now I don't need it to be perfect, I just need it to work!

Current issues are;
-The tape doesn't run against all 3 heads. I think I need to go with a pad or roller between the head and the tape.
-Tape alignment to the heads is off, I need to tweak the mountings a little.
-The pinch roller works for a short while, but then the tape slips up or down the roller until it isn't between the roller and the metal shaft anymore. I suppose I need some sort of guide to keep it in the right place.

But no matter, we'll figure it out!

Quote from: AdamB on March 30, 2021, 04:38:44 AM
I need a pinch roller to tension the tape, give traction and pull the tape through the heads - but I can't do that with a cassette and have an erase head due to lack of room.

If you use a combined erase&record head, then you would have space for pinch roller in the cassette. I have seen those combined heads available in AliExpress (have not ordered them though, so cannot know are they actually usable). Also even when using the cassette, since you are building a custom made machine, you could try to place the pinch roller to the side of the cassette and make a hole at the side of the cassette to take the tape out so that it travels around the pinch roller mechanism. Then you would have room for 3 heads... Or the erase head could be placed to the side of the cassette, leaving room for the pinch roller at the "normal" position.

Quote from: AdamB on March 30, 2021, 04:38:44 AM
-The pinch roller works for a short while, but then the tape slips up or down the roller until it isn't between the roller and the metal shaft anymore. I suppose I need some sort of guide to keep it in the right place.

This might turn out to be tricky to get it working properly. I had this effect using the "official" tape player mechanism, in my case it was related to head misalignment, which in some cases took the tape slightly out of the center line of the tape path and eventually caused the tape slipping out under the pinch roller.

Hi...I accept you've cleaned the head (with a cotton bud and propan-2-ol, not a cleaning tape). If not do as such.

Have you demagnetised the head? I once had a reel-to-reel video recorder that gave a low abundancy signal from the tape that imrproved terrifically when I did that.

QuoteIf you use a combined erase&record head, then you would have space for pinch roller in the cassette.

Thanks for the tip, that certainly sounds like it needs investigating! I'll see if I can get hold of one. I have switched design to a non-cassette version, but I have all the old revisions of my designs saved so I can switch back to cassettes very easily. I'll see what I can find, coz I'd definitely prefer the cassette style design.

So here's where I'm currently at with the transport:


I need to try it out with the variable speed control from the Arduino, going to hook that up again tonight and see how it fairs. So far putting the little doodad over the roller at the top left seems to be enough to stop the tape wandering off the roller (I think it was actually the roller itself travelling up the shaft, so now it's held nicely in place).

Also needs some way to keep the reel on the spindle, so far gravity has been doing that job for me. Will think something up.

Quotevariable speed control from the Arduino

I used a 555 PWM circuit which worked quite well. Good thing about it is that you can change the PWM by just changing a cap, although if you wanted anything fancy with the motor speed the Arduino would work better! I found that a low frequency tended to improve the torque of the motor and helped move the tape more smoothly through my mechanism.

If you abandon the cassette you have something like Roland Space Echo where the capstan/pinch-roller does all the tape transport and felt pads keep head contact. The endless tape itself just bunches in a box area with a transparent cover.

I think the best tape transport I've seen was the old Data cartridge introduced in the 70's. The outside of both spools was driven by a continuous band and the tension keeps constant as one spool increases in diameter, the other decreases. The band is driven by a hub in contact with a rubber drive "puck" on the motor. The recording side of the tape only contacts the heads.

I think I've got everything working, however the bias oscillator only gives 50VPP when unloaded and as soon as I connect the record amplifier it drops to around 8VPP. Anyone have any ideas on how to stop the oscillator getting loaded down so much? Any help would be much appreciated!

Maybe insert an emitter follower NPN buffer before R33 & R34?

Quote from: RLawlor on April 06, 2021, 08:30:39 AM
I think I've got everything working, however the bias oscillator only gives 50VPP when unloaded and as soon as I connect the record amplifier it drops to around 8VPP. Anyone have any ideas on how to stop the oscillator getting loaded down so much? Any help would be much appreciated!

Yes, I had made more improvements on my prototype schematic regarding this. The answer is to use the bias trap between the oscillator and the recording amplifier. Here is my latest revision schematic, notice the bias trap on the middle of the right side:

(https://i.postimg.cc/VrV3fBxw/complete-1600x1200.png) (https://postimg.cc/VrV3fBxw)

With those values it is not automatically tuned correctly, but will improve the situation nicely. If possible try to tune it to maximum oscillator signal at the tape head. In my own proto, I used two parallel caps, 390n + 330n to make a 720n cap in parallel with the 10 mH inductor, which gave me a maximum of 44 V rms bias voltage (about 100 vpp) at the record head even when loaded with the recording amplifier. Note that the inductor should be around 10 mH, preferably more to give better notch filtering. I simulated using different sized inductors and came to the conclusion that 10 mH to 33 mH inductor would be optimal.

By the way, the unloaded bias oscillator voltage of 50 vpp is quite small compared to what I am getting. If the quality factor if the inductor in the bias oscillator is poor (too much series resistance compared to the inductance, that is QL = L/RL) it will reduce the amplitude. Also, the ratio of the capacitors in the oscillator is important. If it is something else than 10n, 200n, 1u, then it will affect the output amplitude for sure. It is a bit tricky unfortunately. Also measuring with low-impedance probes will load the oscillator and show a lower reading on the voltage.

Hopefully this helps.

Thank you very much for the help! I substituted the inductor for the erase head in my circuit and as such replaced the 10nF capacitor with a 22nF one. Maybe I'd be better off driving one half of the record head and the erase head rather than both sides of the record head so I can use a 'better' inductor in the oscillator. As for the bias trap I thought about that but didn't realise a high value inductor was preferable as I'm only using 330uH currently. I'll be sure to try something bigger and see what I can do!

OK. So. Hooking up my transport to the arduino speed controller worked fine - I had to re-print the pulleys at a slightly bigger ratio to get the torque needed for really low speeds but it's working really well now and is really stable.

Playback from the transport works fine (I cut the tape from an old cassette to make the loop and it happily plays back what was on the loop already) - however still having some trouble with recording.

So, here's what my hobbyist scope recons is coming off of the bias osc:

This one is with the erase head in place of L2 (note that if I turn down R32 to lower the PP voltage it forms more of a sine shape, so the weirdness might be it going out of the 20Vpp range of the scope)
https://www.dropbox.com/s/zvbpjjxi1bb9z0p/Scopey1.jpg?raw=1 (https://www.dropbox.com/s/zvbpjjxi1bb9z0p/Scopey1.jpg?raw=1)

And this one is with a 820uH inductor for L2
https://www.dropbox.com/s/gv6kzisx4crg6vq/Scopey2.jpg?raw=1 (https://www.dropbox.com/s/gv6kzisx4crg6vq/Scopey2.jpg?raw=1)

I tested my recording preamp (by just connecting an audio jack to the output) and it passes signal. Hooking the oscillator and recording amp together I don't get anything recording onto the tape. I think the erase is working, coz it's nuked what was already on the tape.

I do hear something from the oscillator through the output - kinda a low pitched 'farting' sorta noise which goes away when I either turn R32 ( to lower the PP voltage of the osc) or disconnect the osc completely.

Where do I go from here? I assume either:
-There's a contact problem with the tape on the heads (though the tape is travelling through the heads guides and is pulled reasonably firmly against the heads)
-The recording amp is not providing enough gain
-Oscillator isn't tuned right
-Oscillator signal is just nuking everything (bias trap needed)

One thing with all this; if the oscillator isn't tuned just right, will it just not work at all? Or should it be "working", just at low quality? It's hard to know how to debug this thing!

-Adam

Turns out that if I turn the gain on the interface right up that you can actually make out a really faint, very distorted echo. Check out this video


There's a huge amount of noise - some of it might be from the Arduino, but I suspect not all of it.

-Adam

Nice progress!

Based on your oscilloscope screenshots, your scope is good enough to measure the output of the bias oscillator. That is good and will help the debugging a lot. The second trace with the "820 uH" inductor is quite close what it supposed to look, but there are a few issues there still.

   1. There is DC offset in the measured signal
   2. The Vrms value is a bit too low
   3. With 820 uH inductor you should get about 56 or 57 kHz, now it shows 47 kHz. With 1 mH inductor you would expect to get 50 kHz.

So as for the oscillator, you should be able to at least get rid of the DC offset. And at this point it would be good to measure the circuit so that the recording amp and the oscillator are both connected to the record head and measure the bias oscillator signal at the head. You need to have DC blocking capacitors on the output of the record amp and at the output of the bias oscillator.

Secondly you need to aim for 16 - 20 Vrms bias oscillator signal (clean sine wave), measured at the record head when both recording amp and the bias oscillator are connected to it. Bias trap will help here. If you have some old tape recorders around, you could try to find a suitable tunable bias trap inductor from those, then you could also snatch a suitable capacitor also. You need to match the 1/(2*pi)*sqrt(1/LC) of the bias trap to the frequency of the bias oscillator and with the scope, tune the bias trap so that you will get maximum Vrms bias signal at the tape head.  After this you can adjust the recording amp volume to suit your needs.

Also, if possible, I would test the recording on a "traditional" tape transport with a C-cassette first. Just connect the recording amp and the oscillator to the existing head in the transport. Then you can play it back on another cassette player to verify that you get a good enough recording.

To me it it looks like that your transport mechanism is spinning the tape too fast and quite often the tape wobbles off the surface of the tape head. This too will affect the quality of the recording, because the record head most likely cannot get proper contact to the tape to bias it properly. So I would not use that for preliminary tests of the recording circuit. The buzzing noise that you are hearing probably mostly comes from the DC motor, I have that problem too that the motor induces some noise especially when it runs fast, but not in that magnitude that it would be disturbing.

So as a conclusion, my advice would be to focus on getting a clean recording on a normal tape mechanism to finish the electronics part and after that focus on testing the transport part. I know that it might be frustrating to just try out different component values to optimize the bias oscillator, but try to snatch some components from existing tape players, so that you would not need to buy them as new. As I mentioned before, your oscilloscope will be a good help in troubleshooting.

Thanks,

On the DC offset - I can't seem to kill it with decoupling capacitors, have you any advice on that? I tried both poly film and electrolytic (I figured poly film would be best as it's an AC signal?) but yea no change.

Good plan on putting my transport aside for electronics development - I have a tape transport here that is just a tape head and motor for cassettes (I'm using it to build a cassette synth but can easily connect it to this circuitry for testing). I'll use that to just blindly record to the tape and then try to play it back and see what the fidelity is like. I'll keep developing my transport alongside but stick to the off-the-shelf transport for testing for the time being.

I don't have any gear that already has a tuneable inductor, so I've ordered a set of various sizes of inductors - I'll try and just match the closest I can. Still trying to figure my way through the maths for it!

Again, thanks for the help

Polyfilm caps are fine in this case and ceramic too, but if any cap does not work, then there might be something funny with the scope. I checked the specs on that scope, it has only 1Mohm probes, so it will show a bit lower reading on the amplitude. Also the max input is 50 Vpk, what ever that means, but that explains the faulty trace on the erase head case, it provides too high peak voltage to your scope.

If you use the pot in the oscillator to attenuate the bias with the erase head connected, do you still get the dc offset?. If not, then better to use the erase head as the inductor.

But if you have caps on both sides of the tape head, you can be quite sure that there is no dc offset there. Also note that on the oscillator side, some caps need high voltage rating due to the large voltage swing.

It is good if you can test the recording on normal cassette transport, it might work better already without any changes to the circuit. Report back how it turns out.

OK yea I think the DC offset is just weirdness from the scope. If I attenuate down the bias signal then the DC offset disappears. I've put a couple of 100n film caps around the head anyway to be sure.

Parts for the bias trap turned up - that helped, however for some reason it only works if I add an emitter follower to the output of the recording amp (between the recording amp output and the bias trap). I went with 10mH inductor and 330pF cap for now as that seems to give the best result from the parts I have at hand. I'm now getting around 24V rms sine at the head with everything connected, at 64Khz.

So, I went ahead with that and recorded to another tape machine to see what would happen - first just connected the head to the record amp and played some guitar in. Then rewound the tape, connected the head to the playback amp and turned on the motor - it plays back quite nicely. It's very noisy and thin but it clearly wrote the sound to the tape which is a step in the right direction. Here's a video:



Trying the same thing with my transport I get nothing at all at the moment - just occasional clicks or grungy noise. So must be something about the tape alignment/pressure on the heads or something about the tape heads I'm using on that transport. I tried shimming the heads to raise/lower them relative to the tape but that doesn't seem to make much difference. Next thing I'm gonna try is redesign the transport top-left corner to fit in another tape guide to try and keep the tape straight and add in some posts to drop those little head pressure-pads that are built into cassettes into place to keep the tape against the head.

I'll report back once I get there!

Edit: On the emitter follower buffer - I did already have a Tl072 unity gain buffer in there but for some reason that caused the bias oscillator to get loaded even with the trap in. Not sure what's going on there but we can look into this stuff to try and get optimal quality once I get the transport to work now.

Ok the bias should be good now. Definitely we need to improve the quality when you are ready for it. For that I would need to see clear pictures of your circuit, mainly how the recording side is connected now to the tape head and bias oscillator junction. I could not make it out from the video.

I had plans to design a simpler recording amp, it would consist of two transistors and the latter would be the emitter follower driving the signal to the head. I could work on that in the mean time and see how that comes out. Definitely we want the quality to be so that a clean sound guitar sounds just as clean on the echoes.

Hi Adam,

Just tested a simpler recording amplifier on the breadboard and noticed some issues...

The noise you are getting is most likely coming from the recording part of the circuit (or how it reacts with the rest of the circuit). I tested a single transistor gain element with emitter follower buffer (with the bias oscillator and playback amplifier connected) and got the same noise issues you have showed on your videos. The noise covers mostly everything and all I got was one distorted repeat as echo. Just like you did. After facing the issues, I re-built the recording amp which is on my schematic to the breadboard, and the noise is cleared. The circuit sounds the same as the one I have soldered to veroboard and hidden inside the metal box.

So, unfortunately, currently it seems that the circuit must be built exactly as drawn in the schematic  :icon_redface: . I did not try how it sounds with a PSU, just used a battery and that 1000 uF capacitor parallel to it. If I take the filtering capacitor out, the noise is just like a jet plane would rocket itself to the sky... the fine line between noise and silence ...

Oh and another thing, I tested placing an electrolytic cap after the recording amplifier and got 2 volts DC offset to the tape head. Then I replaced it with ceramic (or poly film) cap that I normally use (1 uF or 2.2 uF) and the DC offset was gone. So strange things going on with the caps as well. My advice would be to use non-polarized poly film or ceramic caps where ever possible.

I will send you a breadboard photograph to help build the recording amp from my schematic tomorrow evening (24 hours from now). Currently the board is too messy to show anything clearly and I need to rest a little ...

Sorry for the inconvenience, your transport might work just fine. The circuit noise issue needs to be solved first.

OK great. I did get much better recordings with the cassette transport so I suspect maybe it's a combination of electronics noise and bad transport. This is the latest version of the transport (now with those little felt pads to press tape to the heads, tweaked roller and added tape guides)



The quality is real low. I'll try rebuilding the recording amp as you suggest and then revisit once the noise is sorted.

In the previous post I promised to send pictures of building the recording amp (the one from the schematic) to the breadboard. I took 2 pictures, one from the input side to the inductor and the other from the inductor onwards. I use two 820uH inductors in series. Here is the input buffer circuit, where the guitar is connected, and the recording amp high-impedance buffer using JFET:

(https://i.postimg.cc/hXLYX875/input.jpg) (https://postimg.cc/hXLYX875)

The red and black wires on the left go to the input jack. The red and blue wires on the middle indicate the placement of the 10k pot. The 2nd transistor from the left is J112 JFET, so the pinout is according to that component (gate on the left). For BF245C, the pinout is reversed with respect to the gate pin.

Then the rest of the circuit:

(https://i.postimg.cc/H8tJJhdL/rest.jpg) (https://postimg.cc/H8tJJhdL)

There are a lot of components, all connections are not that clear, but hopefully that helps. Again, the red and blue wires are reservation for the 10k pot to adjust the recording attenuation. The rightmost components are the bias trap inductor and cap, connected directly to the tape head stud.

I tested also using a 9 volt DC power supply (takes mains voltage in, puts 9V DC out). So I needed to add some more filtering for the noise. It is very surprising that the recording amp seems to be the cuplrit for adding most of the noise to the circuit. By adding a 150 ohm resistor and 100uF cap to the collector of Q8, the circuit remains stable and relatively quiet when using a PSU. Also, there might be noise issues on the breadboard even when using a battery, so the same resistor + cap combo should fix that as well. The 150 ohm resistor is shown in the second picture, but the capacitor is not connected, because it would have blocked the visibility of other components.

For this reason I made yet another addition to the original schematic and here is the latest version with that resistor + cap combo added:

(https://i.postimg.cc/RqzzGTcX/complete-1600x1200.png) (https://postimg.cc/RqzzGTcX)

Now that I know most of the supply filtering needs to be targeted to the recording amp, I might still try to find a simpler circuit to replace the recording amplifier. But for the time being, that is the best I can come up with. The original recording amp is good because it has a feedback loop to stabilize it and lots of places to add filtering to the bias oscillator signal.

In addition to what I wrote above, you could test the isolated circuit noise so that the playback amp, recording amp and bias oscillator are connected together and all powered by the same power supply and tape heads connected to the circuit. Then just take the output to a guitar amp and listen to the noise. The noise should be in such a level that it can be compared somehow to other pedals, a little bit higher of course, but not much.

OK, so. Latest update is that I got myself a Bush tape recorder - handily the head in it is connected via PCB header, which makes it super easy to switch out which head it's connected to. It also has an aux in - so it's super easy to connect a guitar up to and record to tape. But, you can also connect a guitar up to record and then plug a different head (on a different transport) in and use the recording amplifier/bias from the bush which we know works. I've also tested playing back using the bush transport but outputting it's head to the playback amp on the breadboard and again that works fine.

So I know this electronics setup is all good - but still I get terrible results with the 3d printed transport. There's two main problems;
-Very weak signal and lots of noise
-Extreme wow

I've spent some time trying to solve the wow problem but it's hard to tell whether things are better or worse when the signal is so low. So I'm concentrating on getting a better write/read level out.

To do that, I've redesigned the head mounting on the transport to allow for height and azimuth adjustment; basically the heads are now mounted on screws that pass through a couple of springs that allow us to adjust both read and write heads independently. Here's a picture:
(https://www.dropbox.com/s/y6zf6o74jzyjr5l/PXL_20210426_133604198.jpg?raw=1)

Another great thing about this is that I left enough space between the main body of the transport and the heads so that you can also put in there head "adapters". So if you have heads of different size/shape/mounting positions you can just make a small adpater that connects the head to the transport, but still allows for azimuth/height adjustment. So if this next step doesn't work out my next step will be to replace the heads to be sure it's not just bad heads (being as I've never used them in anything but this machine who knows if they're any good!).

I'm still waiting on the last few parts to print, and then I'm going to fire it up and see whether we can tweak the head positions to give a better output.

Quote from: AdamB on April 26, 2021, 10:53:19 AM
use the recording amplifier/bias from the bush which we know works.

Well, this is good that you now have a reliable rec amp + bias. But then I assume that you could not get the rec + bias from my schematic working  :icon_sad:

The azimuth/height adjustment is a must have feature, so good to have that there now. Different heads give different output levels, depending on their impedance and head gap structure. I have tested now 3 different head types and one of them gave clearly lower output, it seemed that they did not make proper contact with the tape. Therefore switching the heads would be good to try out.

QuoteBut then I assume that you could not get the rec + bias from my schematic working

I haven't tried your recording amp yet. I know your bias works, because I used it with my own simple TL072 4x gain buffer and that worked fine when using an existing tape transport to record to tape.

So right now my priority is figuring out why my transport doesn't work, coz it seems like that's the weakest link currently. So for now I'm using off the shelf electronics that I know work perfectly, so as to make it simpler to debug the transport.

Once I prove out that the transport can work, then it's back to trying out your recording amp on the breadboard!

I tried the new transport last night and if possible I've made it perform worse lol. I get no output at all with the new head adjustments in - not quite sure how that's happened. So I'm gonna just assume these heads are bad and I've made some adapters to put in heads I took out of tape machines that I know worked - gonna try that tonight. Currently looks a-like this:
(https://www.dropbox.com/s/w7z5a8m07u55wgh/PXL_20210427_110519303.jpg?raw=1)

Fingers crossed this time I get something decent out of it!

Nope, big old fail.

Not sure where to take this transport design from here. Perhaps a redesign of the pinch roller to use a sprung mechanism that holds the wheel against the steel rod, I've seen that kind of thing in a few machines. Time to do some thinking.

 :icon_cry:

It would be best to design the mechanism in this phase so that it would be easy to debug. I would return to supporting standard cassette format and also not worry about fitting in the erase head at this point. Just have the record head and play head connected or just one of them, test recording and playback individually on the transport, as to cross check --> Record to cassette on your transport, play back on normal cassette deck, record to cassette on normal cassette deck and play back on your transport. That way you could isolate the problem better. For example, if your recording made on the transport plays well on normal cassette deck, but not on the transport, then the problem is most likely in the play head alignment. I know that you already made this kind of cross check, but maybe it should be more systematic...

Also the thing you mention always about the noise. As I noticed later on in my breadboard, large noise problems in the circuit causes the echoes to sound very faint and distorted. So the noise issues need to be cleared as well.

Right, getting there. This is with an off the shelf transport (I purposely chose a machine that's easily available off of Amazon for around 25 quid), 3D printed chassis and mounts for the extra heads. The little tape guides are stolen from the insides of a cassette.

So not the fully 3D printed masterpiece I want, but still easily recreated DIY.



Electronics are a TL072 recording amp (fixed gain of ~4x) with Jatalahd's bias oscillator + bias trap. The playback amp is from the original tape machine (the jacks and power on the original machine are very easy to desolder, so I figure might as well reuse what's there for now).

Motor is controlled via the Pi Pico with squishy keys for now, via a mosfet motor driver on breadboard.

So next steps; It's still got lots of flutter where the tape action isn't completely smooth, but it's at least starting to be useable in a musical way. I'm going to PCB-up the Pi pico motor control, the recording amp and oscillator, and add on an output mixer to mix dry and repeats (and allow for feedback for multiple repeats) and rebuild this with a more complete enclosure. That'll be version 1, and I'll run that on my vocal board for a while to see how it holds up. I'll post a new demo when I get that far, and perhaps release the V1 BOM/STLs/Gerbers.

Nice progress!

Small update; now with output mixer and feedback that allows for multiple repeats.



There is an issue with the repeats beginning to sort of oscillate that I've yet to figure out - anyone have an idea what's happening here? Here's a video to show the issue in isolation:



As the repeats continue you hear it sort of focuses in on the high end and then start to oscillate as it dies away.

Quote from: AdamB on May 18, 2021, 08:34:17 AM
As the repeats continue you hear it sort of focuses in on the high end and then start to oscillate as it dies away.

Well, I would not say it oscillates, but just emphasizes the high end on every repeat. Does that happen if you slow down the tape speed? Because the faster the tape moves past the tape head, the high end becomes more and more prominent. If you need/want to keep the tape speed high, you could try to balance the frequency response by amplifying bass in recording and adding a simple low pass filter at the end of the playback signal chain. 

OK so hit on a problem; I've made up the first pass PCB and started to test with that but having trouble with the bias oscillator.

I tested the oscillator first in isolation (mounted all its components first and I included test pads on the PCB to check it). When powered by a mains 9V supply this worked fine, exactly the same behaviour as on the breadboard.

However, now that I have everything else mounted - Arduino, motor controller, relay switching, recording amp, playback amp (from the tape machines PCB) and output mixer - the bias oscillator no longer works; it oscillates correctly for a few moments and stops, then works for a few moments and then stops again and so on.

What's most interesting though, is that if I switch from mains power to a 9V battery the oscillator works exactly as expected again - it just seems to hate the mains supply now. Obviously, I can't just use batteries because the arduino and the motor will drain that too fast to be practical, so I need to figure out what's happening here.

One thing I also notice is that if I power the circuit from the mains, but lower the voltage to around 5V then the oscillator stabilises, albeit with a PP voltage and frequency that is too low to be useable (and also the Arduino can no longer power up as it needs >7V to function).

First best guess is that the oscillator is very picky about the quality of the supply voltage? I notice Jatalahd that you have a massive 1000uF filtering cap on your power supply - is that for this same reason?

Edit: Running on battery I've managed to successfully test the whole delay and it "works" (it's a bit low-fi but I think that's mainly the transport which I need to work on more). I had to hack the power supply for the playback amp - I was using the 3v3 supply from the arduino but that didn't fly - I assume the arduino can't supply enough current for it. Powering that separately everything works OK. If I switch to mains power it "works" but the repeats are super distorted where the oscillator isn't functioning right as detailed above. So if I can sort the oscillator-on-mains-power problem I can move on to revision 2 of the PCBs.

Thanks!
-Adam

Quote from: AdamB on June 21, 2021, 09:52:34 AM
First best guess is that the oscillator is very picky about the quality of the supply voltage? I notice Jatalahd that you have a massive 1000uF filtering cap on your power supply - is that for this same reason?

Well, in my circuit the 1000uF cap was used for DC filtering and reservoir for the whole circuit, mainly to avoid some motorboating noise arising from somewhere...

Lately I have started to build a home-made cassette recorder/player. I found some proper erase heads from aliexpress and currently testing those for recording music on cassette. Later I finally found also the traditional tape transport mechanisms from aliexpress, but it seems again I need to wait for several months for them to arrive...

But anyway, related to that, I built the same oscillator using the erase head as the inductor and bounced to the same issue you are facing. The oscillator did not want to keep running. For me the simple solution was to place a 100 uF electrolytic across the power pins close to the oscillator. I  can't say for sure it will solve your problem. I built my own mains power supply giving 12 volts unregulated DC and the oscillator worked fine with that, there was some excessive low-frequency ripple, but it kept running smoothly. So it should not be picky about "quality", but it requires "quantity" (of current). Also noticed that 12 volts is the absolute maximum voltage to run the oscillator, the transistors started to overheat at that voltage.

Please note that the oscillator takes up to 0.1 - 0.2 amps of current. Also note that most likely the motor will take close to 0.1 amps as well. So your mains adapter needs to supply enough current for the circuits. And you might need to add filtering to the motor's DC line, as it produces quite nasty electromagnetic noise.

Hopefully you get it working.

WIP enclosure design is coming along.

(https://www.dropbox.com/s/n8a86y6dx43xpqb/PXL_20210713_110726550.jpg?raw=1)

I need to redesign the transport mount to take account of holding the reels in place (I already designed little caps that go over the tape guides to hold them in place, they're not fitted in this picture). I imagine it'll take the form of some narrow arm that goes over the reel and fits into the center of the reel where it meets the spindle. Might ruin the look of the reels a little but I'm not sure how else to handle that problem other than glueing the reels in place - which means the machine can never be serviced. Another option might be to mount the glass on the inside of the hole in the top of the enclosure so that the bottom face of the glass can be mounted just a fraction above the reels, but the tolerances on that are probably a bit tight to make work with an FDM printer.

Should have the latest PCB built up in the next few days so I can report back on where I'm at with that.

OK so I managed to test the latest transport and electronics and it's *mostly* a success;

Well; what I've learnt is that the thing that kills the oscillator is running the motor. As soon as the motor kicks on, the oscillator starts to fail, and as the motor spins up to a higher RPM and draws more current the oscillator stops running entirely. So. It appears on the surface of it that the power supply can't supply enough current.

I've concluded that it's not the arduino or associated motor controller circuitry, as cutting the wires to the motor and just sticking them straight onto the +5V output of the regulator has the same behaviour.

Now - what I can do is power the motor from a separate power supply and then everything works and it sounds pretty good - the latest transport is a lot smoother and yea it works well. All the controls work as expected save for me wiring some of the pots backwards, doh! and some pot values need tweaking; but the control board where the pots are mounted is a separate PCB connected to the main board by a ribbon cable, so it's easy to experiment with different controls now.

So now I need to figure out how to get the motor working off of the same power supply as everything else. I've tested the following which all have similar results;
9V 1A supply
12V 1.5A supply (what's interesting with this one is that I have to plug in the motor *after* I switch on everything else, or the motor kicking on seems to "trip" the power supply on the arduino and cuts it's power - I'm assuming this is back EMF from the motor starting up?).
9V 3A (this is one of those variable supplies so I'm suspicious about it actually being able to deliver 27W of power).

All have the same issue with oscillator dying when the motor starts. So my question is, is this just a matter of finding a supply that can supply enough current or is there possibly something more fundamental about the setup of my power supply section in my circuit that is causing this?

Back emf in a motor opposes the applied voltage so reduces current! Apart from the instant voltage is first applied, it doesn't happen until it rotates and increases with speed. That said, I'd always fit a reverse protection diode across a DC motor that doesn't need to reverse direction. If the drive control is solid-state, it definitely needs the protection diode, whether or not it's the cause of this problem (though it may be included in the controller).
If the mechanical load on the motor is high, it starts slow and back-emf isn't really happening and the current will be down to whatever winding resistance the motor has. If you have PWM control, you can have a start-up delay and slow ramp-up. Also, an Arduino powers up with I/O as inputs until the setup code makes any of them outputs, so the external circuits should cater for that.
There should also be a small ceramic cap (10nF to 100nF) immediately across the motor terminals to cut RF noise. The motors can should also be connected to the power 0v. The OCD fitment is 3 caps. One across the terminals and one from each terminal to the can/ground.

OK cool, the schematic for the motor/oscillator/power is here for reference:
https://www.dropbox.com/s/uxnuheik7usoejt/motor_and_bias.png?raw=1 (https://www.dropbox.com/s/uxnuheik7usoejt/motor_and_bias.png?raw=1)

So with some more testing I can confirm that for some reason putting a big old 1000uF cap across the terminals of the motor seems to sort the problem; I do have a 100nF ceramic cap in there already but I guess it needs MOOOOREEEE. 1000uF is probably overkill, I just had them on the bench so stuck one in to see what would happen, I should probably experiment with values to see what works best.

Yes, that motor controller is the right way to do it with the pull-down resistor on the MOSFET gate.
The big cap is acting as a current reservoir and after boot up, it ought to be fully charged and able to start the motor when switched on, however, if the motor stalls, the cap will discharge and the motor winding resistance will set the current ("stall current") which might get you back to PSU dropout. Not necessarily a bad thing if it stops something from burning out and a good argument for a regulator just for the motor.
You can measure the motor resistance and see how low it can be (10-30ohm), but many tape motors have built-in speed regulators that prevent you.

OK latest PCBs are in and... early indications are that it's sorted. I fixed the problem by whacking 2 more massive 1000uF caps in - one around the motor poles and one close to the +9V supply of the transistor in the oscillator. Now with a 1A 9V supply all works well. I've yet to connect up the tape output because I ran out of parts, but I'm hoping this is it now. Which means a little bit of work to the 3D printed parts and it's done - I should have a basic demo in the next few days.

Latest demo:



Guts:
(https://www.dropbox.com/s/7n14ajk6e3marwj/Guts.jpg?raw=1)

The controls are on a separate PCB (shown at the top hidden under the main board). It connects to the main board via 24pin ribbon cable (the two IDC box headers). The tape unit and playback amp are hidden underneath the main board (you can just make out 3 pulleys poking out from behind the top of the main board). I decided not to put a footswitch on this - it's all 3D printed plastic so I didn't think it'd be very sturdy for standing on. Instead, there's a "Remote" jack that you can plug in any momentary (or latching with a software tweak) footswitch - the arduino reads the state of the switch and uses a relay to switch the effect in and out. It's true bypass, and the entire audio signal path is analog with the Arduino just controlling switching, lighting (LEDs on the control board) and the motor speed.

Next steps; I gotta do some more cad work to pretty the enclosure up and design a backplate. I also need to account for the 3 absolutely huge 2200uF caps. Then I need to work on choosing better pot values to get reasonable ranges for the brightness of the repeats and levels and what not. But getting there!