Inductors and Gyrators (Acoustic 360 Project)

Started by Rob Strand, November 13, 2017, 09:57:05 PM

Previous topic - Next topic

rankot

I'll try that all in the morning. In the meantime, this is what Mr Maillet wrote to me when I asked him about transistor choice:

QuoteHi Ranko,

if you look at the overall structure
you'll notice something very "unique"

which everybody seems to miss btw ...

it's not so much about the style of circuit design which is very primitive by today's stds

now ...

THAT very idea can be ported into newer design practices which I did with my "360 MDRN"

same "OVERALL"  idea, radically different circuit implementation ...

best,
~jcm

Well, I'm electronics apprentice, so the only unusual thing I notice is the Volume (gain actually) pot which is connected as variable negative feedback, if I'm right. Am I?
  • SUPPORTER
60 pedals and counting!

Rob Strand

#21
QuoteWell, I'm electronics apprentice, so the only unusual thing I notice is the Volume (gain actually) pot which is connected as variable negative feedback, if I'm right. Am I?
To me most aspects have appeared somewhere before.   There are some uncommon aspects to the design and back when that thing was made they would be even less common.

- The variable gain volume.  Inverting amp with variable feedback. Not so common.
   Lets you get near zero gain.  Can keep noise down, depending is specifics.
- They didn't hold back on the buffers, in particular the one in the middle - that would help keep noise down.
- The mid control circuit like that wasn't so common for guitar/bass amps.

Some possibly 'hidden' behaviours would  be how the transistor stage or the mid control loads down the tone-control.   It looks like they have made an effort to prevent interactions by scaling down the tone control values.

BTW,  is your complaint about noise on the clean signal, the distorted signal or both?
The gain in the distortion section is going to bump-up the noise.  On the good side it's
only got the buffer before so most of the noise will be from itself.

That distortion structure appears in a few pedal of the 60's/early 70's.

On the whole I always thought the guys at Acoustic done a pretty good job for their time.
Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

Clean signal, I've built this without fuzz section (not enough space in 1590BB).
  • SUPPORTER
60 pedals and counting!

rankot

#23
I tried to measure noise using my o-scope. Those are switcher noise before filtering (blue), output noise (yellow) and A-B (pink):



Don't seem to be too related to me, do they? There are some coincident peaks, but they appear at ~100kHz, which is my switching frequency.

This is Q1c (filtered B+ at 25V) compared to unfiltered switcher noise (yellow), also don't seem to be too related. Filtering is done with 1uH inductor and 220p+100u capacitors.


  • SUPPORTER
60 pedals and counting!

rankot

This is the signal on all transistors (except Q2e, which I missed), with grounded input:


  • SUPPORTER
60 pedals and counting!

Eb7+9

Quote from: rankot on February 14, 2020, 05:23:33 PM
This is the signal on all transistors (except Q2e, which I missed), with grounded input ...

Ranko,

May I first suggest you try powering up the circuit using three 9v batteries wired in series, at least for a bit at a time just for noise testing ... this will rule out any psu issues you might be having

I would then consider replacing all your transistors with new ones ... start from the back end and move backwards one device at a time ... test for noise improvement/change after each swap ... you might also want to consider using better noise devices like 2n5210 which I've used in many of my builds going back to early 90's ... I've found them to be reliable and fairly quiet

post voltages if you keep having trouble ...

Rob Strand

I agree with Eb7+9.   Probably making life hard for yourself using the CRO.  It's not always easy to correlate  what you see with what you hear.  The external power supply or batteries will wipe out the PSU altogether.   It's a good idea to disable the SMPS like I mentioned before to be 100% sure the switcher is out of the picture.

FWIW, the small pips on both the PSU and audio could imply the PSU noise is getting into the audio.  It could also be caused by a noisy CRO grounding  point.
Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

I will try to follow advices from both of you, they are very nice! What's CRO?  :-[
  • SUPPORTER
60 pedals and counting!

Rob Strand

Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

Quote from: Rob Strand on February 15, 2020, 01:19:50 AM
FWIW, the small pips on both the PSU and audio could imply the PSU noise is getting into the audio.  It could also be caused by a noisy CRO grounding  point.
I've connected CRO ground to input or output jack, if I remember well.

But could it be the problem with noisy pots? It seems that noise is starting to go wild at Q4 collector, which is connected to Treble/Bass tone stack. But to stop guessing, I will try with another power supply or 3 batteries.
  • SUPPORTER
60 pedals and counting!

rankot

Hm, I was a little bit lazy to desolder TLC555, so I tried another approach - simply tried to run SMPS at higher frequency. Instead of 120kHz, I tried 450kHz and the most of the noise is gone now. So it is definitely SMPS problem, but I must find out how to fix it. Maybe just leave it as is - but run it at 1.2MHz, for example.
  • SUPPORTER
60 pedals and counting!

Rob Strand

QuoteI tried another approach- simply tried to run SMPS at higher frequency.
Good idea.

Quotemost of the noise is gone now. So it is definitely SMPS problem, but I must find out how to fix it.
Noise can get in through the power-rails or coupled electromagnetically into the audio (or both).
Assuming it's coming through the power rails, a simple RC filter between the switcher output and the circuit
is good thing to try and not too hard to retrofit.

If it's coupling electromagnetically then it's not so easy.  You can only use a better inductor, better layout, or some judicious shielding between the switcher and the audio circuits.
Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

#32
Well, I already have LC filter which seems to kill all the noise on power supply - 1uH + 100uF || 220pF. Power is very noisy before that, and after that almost nothing (see one of previous posts), so it is probably being coupled somewhere else. I may try to add another capacitor pair at the end of B+ line, near Q4 and see what happens then.
  • SUPPORTER
60 pedals and counting!

Rob Strand

#33
My apologies I've somehow edited your post.  I've tried to restore it
Quote
    Well, I already have LC filter which seems to kill all the noise on power supply - 1uH + 100uF || 220pF. Power is very noisy before that, and after that almost nothing (see one of previous posts), so it is probably being coupled somewhere else. I may try to add another capacitor pair at the end of B+ line, near Q4 and see what happens then.

OK I see it now.   Is that inductor 1uH or 100uH?   The layout shows 100uH.

As far as the filtering goes it might be worth adding 1k in series with that inductor as a test.

However,  I can see two problems:

The first is a layout problem.   The ground track to the analog circuits is tapping off the noise input track.  Ideally this ground track should ideally connect to the -ve terminal of your 100uF filter cap.  But the input socket is already exposed to the noisy ground ...

The second issue is the power is switched using the input socket, as is normally done for effect pedals.   The only problem is with switch-mode supplies is they pulse the input current and those pulses flow down the input socket ground and noise gets into the audio.   You 220uF input cap helps but might not be enough.

I don't know if these are the cause in your case but they are very suspicious.

One way forward would be to remove the 2A fuse and wire the ground wire from the DC jack to the -ve terminal of the 220uF input cap.   (This may also help the analog power tap-off issue but not 100%.)

If the ground through the input jack turns out to be the problem the correct solution is to use a MOSFET switch to switch the power and use the input jack to switch the MOSFET.   That way the noise ground line does not pass through the input socket.   The common way to wire this is a p-channel mosfet switching the *+ve* rail and the gate being switched by the input socket.
[EDIT:  so the power socket ground gets wired to the -ve terminal of the 220uF input cap and the input socket ground wires to a 'quiet' ground point.]

There's a few schematics on the web about this.  There was a thread where Marcus-Munky has this issue with an angry switch-mode on a bass with a lithium battery.  I think there was also something on RG's site, perhaps on one of the schematics.  I can't find the link or the thread at the moment.

Keep in mind you might  find every one of the above points adds to the noise.  When you remove the main cause the others could still be adding a little and it's not until you remove all the causes that the circuit is truly quiet as it can be.
Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

Rob, filter inductor is 1uH. I will try to add some series resistance.

I have to 220uH caps. The first one is preceding SMPS, another one is between diode and inductor in SMPS. Shall I connect ground from DC jack directly to the later?

I really don't need input jack acting as a switch, since I don't use batteries here, so I can try to skip that if it will reduce noise.

I settled on 33k R1 / 1k R2 / 100p Ct for 555, which gives 412kHz SMPS. It seems to have less noise than before, at least to my ears, but I will measure it to check what's going on.
  • SUPPORTER
60 pedals and counting!

rankot

Tried all those ground points and noise seems to remain the same.  >:(

Will try to add some resistance into filter.
  • SUPPORTER
60 pedals and counting!

Rob Strand

QuoteShall I connect ground from DC jack directly to the later?
The correct connection for in DC input ground is to the -ve terminal on the 220uF input cap.

Quote
Tried all those ground points and noise seems to remain the same.  >:(

Will try to add some resistance into filter.
It's a matter of working through the options. 

FWIW, you could increase the value of the 2u2 output cap on the switch-mode.

Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

I am OK with the noise level right now, it seems that increasing switching frequency fixed it. I have another problem now - preamp is distorting the signal, and it is pretty obvious on transients, I must turn the volume down to 1/2 to avoid that. And if I turn it to 3/4 or more, it is distorting almost always.

I tried to inject sine wave 100mV p-p into input, and this is what I get at Q4 base (it is clipped a little, but it is normal, due to nature of my signal generator) - so I can say it's not distorted there.



But when I look at Q5 base, I get this - like the bottom of the wave is flipped at some point.



Q6 base looks like this:



What can be the reason for this? Is it possible that simple treble/bass tone stack can distort the signal like this???
  • SUPPORTER
60 pedals and counting!

Rob Strand

QuoteWhat can be the reason for this? Is it possible that simple treble/bass tone stack can distort the signal like this???

The inverted signal on Q6 comes from the fact Q5 inverts the signal.  The cause is Q5.

As to what is the cause.  I'd look at the biasing of Q5.   Off-hand I don't know if there are any voltage lists for that circuit.

What might be happening is Q5 is overloading and that could cause cap on the base of Q5, and perhaps caps in the tone control, to charge-up.   However, the fact weirdness occurs on the negate swing makes me think there's a bit more going on.   Perhaps Q5 is overloaded on the positive input cycle into Q5, and that charges the input cap via the base-emitter diode.  There will be a DC voltage on the input cap with the base side being negative.  When the input to Q5 goes negative the DC voltage on the input cap cause the base to swing more negatively and breaks down the BE junction of Q5.

*BUT* the breakdown theory doesn't hold up because the voltages on your Q5 waveforms are too small.   Are those CRO voltages correct?  only 100mV?  Are you using a x10 probe?

Check the DC voltages on Q5.   Perhaps check and/or replace Q5.
Send:     . .- .-. - .... / - --- / --. --- .-. -

rankot

#39
No, probe is at x1. I have the voltage chart from the service manual, it doesn't show waveforms in different spots, but for 300Hz 100mV p-p input (My input is ±100mV. Is it OK, or I need to inject ±50mV?), Q5 base shall be at 400mV and Q4 base at 1.2V. Q6 base shall be at 800mV.

If I read this CRO well, I have 1.2V at Q4, 400mV at Q5 (but bottom is flipped) and 600mV at Q6. So Q5 is definitely a suspect, just have no idea what's going on...

I have also measured DC with shorted input, and this is what I get, but have in mind that my B+ is 23.7V, not 25:


 
   
   
   
   
   
   
   
Q...Collector........Emitter..........Base.............
125.0 (25.0)19.7 (19.5)18.8 (20.1)
210.3 (11.0)0.95 (0.9)1.6 (1.5)
324.9 (25.0)9.7 (10.3)10.3 (11.0)
47.0 (7.0)3.1 (3.4)3.8 (4.0)
512.6 (12.0)5.7 (6.3)6.2 (7.0)
624.9 (25.0)11.9 (11.3)12.6 (12.0)

Values in parenthesis are from the service manual. Please note that my B+ is 23.7V (not 25), so I scaled all my readings with the factor of 1.055.
  • SUPPORTER
60 pedals and counting!