MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Transpose Utility Module - who can handle pitch transpose?
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page Previous  1, 2, 3, 4, 5, 6  Next [all]
Author Transpose Utility Module - who can handle pitch transpose?
frankbuss
A new version is attached. I could very much simplify the potentiometer OpAmp circuit for the offset with some tricks, so now in theory it would work with just one quad OpAmp, instead of 2.

But I still used a second quad OpAmp to buffer all inputs, which gives the module a very high input impedance, and I added some capacitors and resistors for ESD protection and as a low-pass filter for very high frequencies (mobile phone noise etc.). Because of the input buffer, I could reduce the resistors for the sum OpAmps (all 10 k and 5 k instead of 100 k), without sacrificing the input impedance at the jacks, which reduces noise, because the OpAmps have bipolar inputs and not FET inputs, and which reduces resistor noise.

I think now it is time to build and test it.
FUUUCCKKKK!!!
pichenettes
Looks good to me, though you might use 10k/20k (E24) instead of 5k/10k in your last stage, or maybe build your 5k with two 10k in parallel, because there aren't many choices of 5k resistors with ultra low tolerances (if a 1.8 mV offset is a nuisance to you, 0.1% or even 0.05% won't be good enough for the resistors).
frankbuss
Right, looks like 5k with 0.05% costs EUR 3.75 at Digikey (0805), but I can 10k with this accuracy for EUR 0.69, so 2 x 10k in parallel is it then. This gets still expensive, EUR 11 for 18 resistors which needs to be accurate, maybe I'll try it with 0.1% first, they cost only half.

How much accuracy do similar modules have? 1 mV sounds very good, but I can't find detailed datasheets for other modules, like accuracy, frequency Bode-plots, linearity curves etc. (I have some industrial electronics design background, where this is standard).
kay_k
thanks Aroom for the shout outs!
and while there is a chance you read this: thanks Olivier for your crazy crazy crazy good shit. AND for making it open source!

the mod for Aroom is done and works. well my very non-musical ears have problems getting 3 voices in tune.

I had to add one Resistor for the unaltered buffered output, and weirdly enough that one was wrong. same roll but 9% too high.
So my first demo video (done in 15 minutes, so don't shoot me for the simple sequencing and detunes) has only three voices.
(but also because I only have three 1V/Oct VCOs and three voltage sequencers in the moment, rest here is MIDI)

https://www.youtube.com/watch?v=U1fhEDcPu20

I hope for a better video from you wink
kay_k
frankbuss wrote:
Right, looks like 5k with 0.05% costs EUR 3.75 at Digikey (0805), but I can 10k with this accuracy for EUR 0.69, so 2 x 10k in parallel is it then. This gets still expensive, EUR 11 for 18 resistors which needs to be accurate, maybe I'll try it with 0.1% first, they cost only half.

How much accuracy do similar modules have? 1 mV sounds very good, but I can't find detailed datasheets for other modules, like accuracy, frequency Bode-plots, linearity curves etc. (I have some industrial electronics design background, where this is standard).


yeah, accurate resistors are not cheap. actually besides the panel the most expensive thing in my module, more than the jacks for instance.
frankbuss
I added again additional protection for all inputs and outputs. The OpAmp has already build-in protection, but better to have extra protection, costs only cents and doesn't have to be populated, if not needed.

I'm still not sure about the width. I guess without the switch and without the additional potentimeter output, it could be possible with 2 HP. But I bought some blank panels, and they are really tiny and have some disadvantages, like I couldn't find a pot with a screw for this width, so this would only be soldered to the PCB, and wouldn't have a nice knob, only a knurled shaft and fiddly to adjust. With 4 HP it is easier to place input and output jacks side by side (3 vertical for input, 3 vertical for output), and I would have space for the extra buffered voltage from the potentiometer, which might be useful when cascading multiple of these modules, or for exact measurements (is this done with modules?). So I think I'll do a prototype for 4 HP.
pieter
Returning to the OP's original list of transpose modules, if you have a Varigate 8+ you can use the CV out to transpose duties: Activate CV1 or CV2 and put it in Track mode. The Root+n sliders then act as a transpose. I believe you can also program this in, but I haven't tested that.

If you want to transpose multiple voices at once you will probably have to employ multiples and unity mixers, and that is bound to take more than 2hp.
aroom
pieter wrote:
Returning to the OP's original list of transpose modules, if you have a Varigate 8+ you can use the CV out to transpose duties: Activate CV1 or CV2 and put it in Track mode. The Root+n sliders then act as a transpose. I believe you can also program this in, but I haven't tested that.

If you want to transpose multiple voices at once you will probably have to employ multiples and unity mixers, and that is bound to take more than 2hp.


I have a VG8 and a VB. and you can't program neither control with CV the key change in Track mode. it's a good feature, but differ from what I'm looking for.

I will receive now a 2hp module who can handle pitch transpose for 3 voices. So it's not impossible.
Paranormal Patroler
aroom wrote:
Check him out, he got some good modules going on :
https://www.modulargrid.net/e/other-unknown-wk3-midi-thru


Yeah, I asked that he also make a 2HP MIDI merge since he did the MIDI Thru. Would be awesome to have that functionality on a eurorack case.
pieter
aroom wrote:
you can't program neither control with CV the key change in Track mode


What do you mean by "neither control"? I use the sliders to transpose a voice all the time. I typically have the notes coming from a different source, though, not the same CV out as the one I use for the transpose. The A-110 has a separate pitch CV in, which I use for the transpose in (similar trick to using the FM input on most VCOs).
aroom
frankbuss wrote:
I'm still not sure about the width. I guess without the switch and without the additional potentimeter output, it could be possible with 2 HP. But I bought some blank panels, and they are really tiny and have some disadvantages, like I couldn't find a pot with a screw for this width, so this would only be soldered to the PCB, and wouldn't have a nice knob, only a knurled shaft and fiddly to adjust. With 4 HP it is easier to place input and output jacks side by side (3 vertical for input, 3 vertical for output), and I would have space for the extra buffered voltage from the potentiometer, which might be useful when cascading multiple of these modules, or for exact measurements (is this done with modules?). So I think I'll do a prototype for 4 HP.


Personally it's a big deal going from 2hp to 4hp BUT it depends on the functions available.

If you go 4hp, you should consider adding some features. Like a potentiometer per voices. Did you consider that the pot offset should be quantised to octaves or semi-tones?

You could also consider adding a second transpose intput.


Your did mentioned switches in your design. What are they for?
aroom
dbl damn smartphone
aroom
pieter wrote:
aroom wrote:
you can't program neither control with CV the key change in Track mode


What do you mean by "neither control"? I use the sliders to transpose a voice all the time. I typically have the notes coming from a different source, though, not the same CV out as the one I use for the transpose. The A-110 has a separate pitch CV in, which I use for the transpose in (similar trick to using the FM input on most VCOs).


I meant you can't control this function with an external CV signal.
frankbuss
aroom wrote:

Your did mentioned switches in your design. What are they for?


The switch is for setting the pot offset to zero, because my assumption was that this is difficult to do manually, and sometimes you might want to do this temporary, but then back to the pot settings.

A quantised pot offset sounds difficult (I would use a microcontroller and a DAC for it), but right, might be very useful. Maybe switchable, too, if you need finer control for other things.
aroom
frankbuss wrote:
aroom wrote:

Your did mentioned switches in your design. What are they for?


The switch is for setting the pot offset to zero, because my assumption was that this is difficult to do manually, and sometimes you might want to do this temporary, but then back to the pot settings.

A quantised pot offset sounds difficult (I would use a microcontroller and a DAC for it), but right, might be very useful. Maybe switchable, too, if you need finer control for other things.


my guess for a 4hp module of this kind would be :

inputs :

T transpose master CV signal
T2 idem T - to add lfo for tremolo for instance

POT CV quantized to octave or third /quinte offset would be normalized to T2

1 CV signal from a sequencer
2 idem 1
3 idem 1

outputs :

T copy of T input

1 T+T2+1 inputs
2 T+T2+2 inputs
3 T+T2+3 inputs

with all T output, you'll have a 3 poles switch to select between what you want to add to the inputs :

nothing / T / T+T2



that would make it worth the 4hp form factor wink
gimber
I've been wishing for an add-one-to-many module for a couple years and am so glad to see this thread and the modules that seem to be coming out of it. I'm running four links right now to have four transposable channels but am really only using the middle section of each, so pulling this off in 2 or 4 hp (even if it's 3 instead of 4 channels) would be incredible.
aroom
gimber wrote:
I've been wishing for an add-one-to-many module for a couple years and am so glad to see this thread and the modules that seem to be coming out of it. I'm running four links right now to have four transposable channels but am really only using the middle section of each, so pulling this off in 2 or 4 hp (even if it's 3 instead of 4 channels) would be incredible.


yeah, I'm so grateful that frankbuss, kay_k and pichenettes got involved.

I can't wait to test the module. It will be perfect to transpose voices, but I'm sure it can be much more than that.
kay_k
So, when Aroom likes it and I find a few others I might build it right into the PCB. So it will be either a solder jumper on the back or an individual module.
I also plan to make a 1RU module featuring 2 WK1. (because I made myself a case with a 1RU row)
guestt
This is brilliant - thank you everyone smile

applause

Possibly redundant given how this thread has progressed, but to help complete the list Rabid Elephant's knobs has the ability to do two at the flip of a switch, it's very cool!

https://www.rabidelephant.com/products/knobs
frankbuss
[quote="aroom"][quote="frankbuss"]
aroom wrote:

POT CV quantized to octave or third /quinte offset would be normalized to T2


How does the normalization work?

Sure, I could do this, but I wonder if someone would buy it. My simple version would maybe cost EUR 20 for parts. With the additional in/outs, switches and quantization and the additional required OpAmps, microcontroller, DAC, ADC etc., it might cost EUR 50 for parts. But would be fun to do it Rockin' Banana!

So this would be the specification:

inputs :

T1 transpose master CV signal
T2 idem T1 - to add lfo for tremolo for instance, or for cascading multiple of this module, by feeding the pot output from other modules to it

CV in 1: CV signal from a sequencer
CV in 2
CV in 3
CV in 4

outputs :

pot CV output
copy of T1 input
copy of T2 input

out 1: CV in 1 + transpose voltage
out 2: CV in 2 + transpose voltage
out 3: CV in 3 + transpose voltage
out 4: CV in 4 + transpose voltage

transpose voltage: selectable with a 3 pole switch: nothing / T1 / T1+T2

other elements:

POT: CV offset

switch: select pot offset mode: fine linear, quantized: semitone, whole tone, third, quinte or octave (6 position switch). Might be a good idea to use a digital rotary encoder instead of a pot? in the "fine linear" position it could have some kind of acceleration: if you turn it slowly, it is really fine, but if you turn it faster, it changes exponentially faster.

Don't know, does this still fit in a 4 HP module? And would this still be in the spirit of modular systems? My assumption was that they shouldn't do too much with hardwired internal connections, but just one thing but this right, because then it is easier to combine them and more flexible. So for example there could be a module just with the quantized pot outputs (and switches to select the mode), maybe 4 pots and 4 outputs. This would be useful on its own for all kind of scenarios. And then simply a module which adds 2 CV signals, maybe with 5 channels (so two inputs and one output for each channel). Then another module which can split 1 input to 7 outputs (I think there are already many such spitter modules, best would be active, not passive, to avoid too much load from the inputs of connected modules). Then you can patch it all as you want seriously, i just don't get it
aroom
frankbuss

I imagined that the CV generated by the POT would be summed to anything on the T2 input. like in the A-185-2 module

I have no idea how Doepfer managed to add 1 V per stage on their POT. maybe it's not a pot. maybe it's not that complicated and doesn't involve a micro controller. Sorry can't help you with the electronic aspect. I guess it's incremented and adds or removes 1 V on each positions.

At the end, this POT feature is not the biggest asset of the module, IMO. for me it doesn't need to have all that quantization, semi tones, third, quinte features. Maybe it doesn't even need to be at all.


The switch feature is maybe more relevant, due to the 4 hp form factor. I'm just throwing some ideas on the thread at this point, just to see what could be done differently than what Kay is already proposing in 2hp.

I guess that before going crazy on a fixed architecture, it's good to exchange some ideas (I hope more people will come and drop some good idea as well), but the most relevant part would be to try some prototypes and see what the most useful at the end.


frankbuss wrote:
But would be fun to do it ! Rockin' Banana!


Exactly w00t

Regarding the spirit of modular, I guess that it's personal if someone want's to go in the One Module One Function way or One Module a lot of Function way. It also depends on how small you wanna keep your rig.

So you are right, almost every modules needed to achieve that are already out. Just check the list I made on the first post. It's just a matter of HP : a small specialized module.


And like I said in an early post, this whole idea / concept was based on how to transpose more than one voice in 2hp but could be used for anything else. Times will tell.


By the way, your idea to have an output and input to chain more than one of this module is really smart.
frankbuss
Looks like the quanitzer circuit is simple, for example in a Doepfer A156:

http://www.doepfer.de/service/A156_Abgleich.pdf

I can't find a schemaitc, but you can see a TLC5620, a good 8 bit quad DAC for EUR 4, and a PIC16C73, an old and outdated microcontroller (which has integrated ADCs). They might do some tricks like oversampling or stacking the DACs, I would simply use a cheap but better 12 bit DAC.
frankbuss
I did a test on a breadboard for the sum and quantize circuit, and to test the ADR03 reference. The reference voltage was spot on at 2.500 V, says my multimeter and even after buffering with the OPA4180 etc. with 1% resistor from my parts bin it was still 1 mV exact. That's a good start.

Then I used the MCP4822 DAC and an Arduino to test the quantization. The measured values are not the best, errors compared to the calculated values range from -1 mV to +7 mV. But then one semitone is 83 mV, so should be still good enough and the additional ADC of the Arduino might result in a larger recorded error than what it actually is. This is the Arduino script:

https://gist.github.com/FrankBuss/387382d840e14c295ce4148931bc42d9

And this the debug output, recorded with Minicom and displayed with a speadsheet program (Libre Office, I'm working mostly on Linux, y-axis in millivolts) :



Really nice to test such things with the Arduino, because it is easy to program ideas and then record and analyze data, either audio or numbers with the serial port stream.

So how does it sound in a real setup? I tested it with a sequencer, and my only module I have so far, a Ginkosynthese Grains, with the PWM saw Arduino sketch with VCA/VCF (configured with VCF). The sequencer CV output is summed to a CV, which is controlled by a linear pot, and to a CV, which is controlled by the Arduino and the DAC, which quantizes another pot, and then the summed signal is used for the pitch of the Grains VCO. This is how it sounds:



As you can hear, there are some glitches when changing the quantized pitch, but I think I can solve this with a digital hysteresis in the software, because at the edges between two steps it oscillates sometimes. The noise etc. might be because of the breadboard setup and the not exactly hifi quality of the Grains VCO.
frankbuss
Another idea for the architecture, maybe crazy, really difficult to design a universal and useful module nuts



It is 8 HP, wasting a lot of precious space, but on the other hand it is very flexible and integrates 3 modules in one. Function should be obvious, because there is no internal routing, everything can be patched externally as you like. All functions discussed in this thread should be possible with patching. And I added a switch for the potentiometer min/max range, to allow finer adjustments, if you just want one octave, but on the max level, you can adjust it to +/-5 octaves.

I have some questions about the quantize function. Is it useful to add scale functions, like the A156 has (http://www.doepfer.de/a100_man/A156_man.pdf)? But then it would be absolute, note C starting a 0 V. Is this right? With the last semitone test it is kind of absolute, too, because it scales the measured voltage to 1/12 steps, but of course it doesn't matter, if the note at 1 V represents C or D or whatever. Then I could use 2 switches, as in the A156 unit. I don't know if the 3rd switch is useful, probably you don't want to transpose +6 or +7 often, and then a rotary encoder with 8 positions for some useful applications might be better, including maybe an full octave position and a "no change" position? Everything special could be done with the semitone position (when you control it with a pot).

I'm not sure about the exact dimensions of the elements, I measured it with my Grains module which I have here. Attached is the Inkscape (free vector graphics program) SVG file, if you want to move elements around or if you have other ideas.
SlyFrank
frankbuss wrote:
Another idea for the architecture, maybe crazy, really difficult to design a universal and useful module nuts



It is 8 HP


8 HP is fine with me for such a useful module. I'd buy it in a minute whether pre-built or DIY. This thread is very timely as I am recently looking to be able to transpose up to 4 separate sequences at the same time with 1-4 constant offset voltages (it's kinda doable now, but it's very clunky). At least 4 channels of 2 inputs each and an output each would be great. Someone earlier mentioned 3 sequences being transposed would be OK, but for me, 4 would be ideal...and awesome w00t

Edit: Also, the quantizer on-board is nice, but imo opinion not essential, so if it saves cost/HP then I would omit it. I am most interested in transposing sequences not in increments of 100 cents (microtonal scales), and even in those times when I am using a 12-TET scale, my voltage offsets will be coming from a source that can produce those exact values. Others may disagree...
MUFF WIGGLER Forum Index -> Eurorack Modules Goto page Previous  1, 2, 3, 4, 5, 6  Next [all]
Page 2 of 6
Powered by phpBB © phpBB Group