Concept of Aliaser / Bit Crusher Improvements

Started by Mr. Lime, November 06, 2019, 04:55:14 AM

Previous topic - Next topic

Mr. Lime

Gathering together all my impressions of aliaser circuits as guitar pedals, I want to sum them up and discuss a little in this thread.

1) Oscillator bleedthrough
This could be avoided by an envelope controlled gate that powers the oscillator.

2) PWM could be done in various ways
Imo the most elegant solution is feeding a triangle carrier oscillator into a comparator. This offers constant amplitude of the output sqaurewave and allows us to tweak the oscillator circuit for a wide range.
The other input could be used as voltage divider or as another LFO modulating the duty cycle. Different wave shapes may lead to interesting random PWM.

3) S&H / aliasing switching
There are different switches that can be used but logic CMOS switches seem to be the best. They are fast, cheap and the off-resistance is very high compared to JFETs.
This improves the lowpass situation formed by the alaising 100n cap.

4) Clean blend control
Not too sure about that one but it might be a nice addition to mix some clean signal.


My first reference would be Paul Nelson's Wolfbagger circuit.
Parasit Studio seems to improved the envelope gate in the Sonic Reducer circuit.

https://www.parasitstudio.se/uploads/2/4/4/9/2449159/sonic_reducer_2_doc.pdf

Now some questions before I start a circuit based on those points.

3) Can a CD4053 be used instead of a 4066?
I don't like the fact that 3 of 4 switches are unused and I might try to take use of the 2 SPDT switches of the 4053.

One interesting switch is the 4007. Is there a point of taking this one instead? I read it's behaviour is more like a JFET so it's not limited to logic states. Think of triangles modulating the switch.
I guess the disadvantage of low resistance in "off" position is shared with the JFET?


I would appreciate some thoughts and experience sharing.

Thanks a lot!
Thanks for help

Fancy Lime

I don't have the time to go into details right now but I have thought about this hard a good long while ago and it never went anywhere. Would love to see you succeed where I failed. I'll try to share some results as soon as I can.

Cheers,
Andy
My dry, sweaty foot had become the source of one of the most disturbing cases of chemical-based crime within my home country.

A cider a day keeps the lobster away, bucko!

anotherjim

The 4007 transistors can go to very high (as good as disconnected) off resistance. Unlike discrete Mosfets, the bulk diode is routed to a supply pin, so doesn't conduct in the signal path provided the signal doesn't swing outside the supply voltage. However, just like JFET's, there is the problem of signal modulation of the Vgs condition which will limit signal swing. Of course, that can be lived with if you design it right although you will probably have to provide a bias trimmer. The 4000 Mosfet's are far more consistent than Jfet's, but still not identical for Vgs, Ron and linearity.
Unless you absolutely need some linear control of the switch, the full transmission gate CMOS devices are far easier to work with.
IIRC, the 4016 switch has some ability to have the switch action slowed slightly compared to others. Apparently, because the control input has less buffering and so more closely connected to the internal gates.


ElectricDruid

I've started trying to avoid JFETs because they're getting to be hard to find. CMOS switches are an obvious alternative for a job like this. And yes, why not use the 4053? You could use the two unused sections to do bypass switching, maybe? Or is that pointless if you can just switch the thing on and have the signal go straight through (only two buffers in the path)?

Jim is right about slowing the transitions on the switches. The 4053, 4051, etc have Schmitt trigger control inputs, so slowing the edges down doesn't do anything useful. The 4066/4016 are simpler. Still, for a sample/hold type application it probably doesn't matter.

I'm not sure about the PWM. Am I right in thinking this effect is actually a "follow and hold" not a "sample and hold"? So the pulse time determines how long the signal is followed for before it is held? In which case, some sort of modulation of the pulse width might be interesting. Doing that simply is the challenge there. You'd need a sample-rate triangle wave oscillator, an LFO, and a comparator. Gets to be quite a lot of circuit to do a job that could be done by a 555! (or a 7555, if we want any chance of it being quiet).

I shall be interested to see what you come up with. I've been kicking similar ideas around myself, but my scheme was going to be simply Input buffer->PIC->Output. Since the thing is supposed to be lo-fi, the limitations of the PIC's 10-bit ADC  don't matter and changing the sample rate and resolution then become a question of firmware ( R.G.'s "ASMOP"). A wholly digital approach seems fair enough for a project like this!

Tom

Mr. Lime

#4
QuoteThe 4007 transistors can go to very high (as good as disconnected) off resistance. Unlike discrete Mosfets, the bulk diode is routed to a supply pin, so doesn't conduct in the signal path provided the signal doesn't swing outside the supply voltage. However, just like JFET's, there is the problem of signal modulation of the Vgs condition which will limit signal swing. Of course, that can be lived with if you design it right although you will probably have to provide a bias trimmer. The 4000 Mosfet's are far more consistent than Jfet's, but still not identical for Vgs, Ron and linearity.
Unless you absolutely need some linear control of the switch, the full transmission gate CMOS devices are far easier to work with.
IIRC, the 4016 switch has some ability to have the switch action slowed slightly compared to others. Apparently, because the control input has less buffering and so more closely connected to the internal gates.

So the 4007 is a nice JFET alternative.
I probably stick to the 4053 and use the two unused SPDT for triangle/squarewave selection and LFO / manual control selection of the PWM.

QuoteI've started trying to avoid JFETs because they're getting to be hard to find. CMOS switches are an obvious alternative for a job like this. And yes, why not use the 4053? You could use the two unused sections to do bypass switching, maybe? Or is that pointless if you can just switch the thing on and have the signal go straight through (only two buffers in the path)?

Jim is right about slowing the transitions on the switches. The 4053, 4051, etc have Schmitt trigger control inputs, so slowing the edges down doesn't do anything useful. The 4066/4016 are simpler. Still, for a sample/hold type application it probably doesn't matter.

I'm not sure about the PWM. Am I right in thinking this effect is actually a "follow and hold" not a "sample and hold"? So the pulse time determines how long the signal is followed for before it is held? In which case, some sort of modulation of the pulse width might be interesting. Doing that simply is the challenge there. You'd need a sample-rate triangle wave oscillator, an LFO, and a comparator. Gets to be quite a lot of circuit to do a job that could be done by a 555! (or a 7555, if we want any chance of it being quiet).

I shall be interested to see what you come up with. I've been kicking similar ideas around myself, but my scheme was going to be simply Input buffer->PIC->Output. Since the thing is supposed to be lo-fi, the limitations of the PIC's 10-bit ADC  don't matter and changing the sample rate and resolution then become a question of firmware ( R.G.'s "ASMOP"). A wholly digital approach seems fair enough for a project like this!

Tom

I've got the first draft done. Yeah it's quite a lot of circuit for the oscillators. The comparator forms the duty cylce either with the LFO or the "Resolution" pot. Hope I got the envelope follower and it's comparator connected right to the oscillators to keep things quiet when the guitar isn't played.

Sure a microcontroller would be perfect for this application but I'm far away from being able to program them myself and a ready to buy solution isn't available as far as I know..
In the next weeks I will try some very simple arduino projects with ATTiny85 controller..


https://imgur.com/9ZLMG1a
Thanks for help

Mr. Lime

I adopted the oscillator from Paul Nelson's 5 Knob Aliaser mod but think there's a mistake with those values and the targeted range of 31Hz - 2.2kHz

Anyway I did some math to come up with a wide range like ~15Hz - 30kHz.
I hope I got it right. This I think is the most reasonable range for this application.
Are there any problems that can occure with those values? I think of amplitude failure or maybe ticking..?

30k should be right above the audio range so the switching of the CD4053 should not be heard anymore.
Only the PWM of the LFO sweeping should be noticeable giving a kind of tremolo effect.
This idea is from this thread: https://www.diystompboxes.com/smfforum/index.php?topic=90416.0



Thanks for help