FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 

Something New from Doc Sketchy
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2, 3 ... 13, 14, 15  Next [all]
Author Something New from Doc Sketchy
Dr. Sketch-n-Etch
Hey Bitches,

Late last week, I was set the challenge of coming up with a "simple" design for an interpolating scanner -- something that used a control voltage to morph between one audio input and another. This was something that I struggled with a couple of years ago, and even built one or two things, none of which worked terribly well, nor were very "simple".

Anyway, for some reason this time around the design gods were smiling down upon me, because I was able to come up with a workable design fairly quickly. So, I spent the weekend simulating, tweaking, laying out and building the prototype boards for a four-channel interpolating scanner. It wasn't "simple" enough for the person who gave me the challenge (guess who), so I decided to take this to the public myself, since I think it's pretty cool, and it works great!

If this is something that interests you, please watch the video below to see a brief demonstration:

So, I don't have a panel or anything yet for this -- what you see in the video is all I have so far. However, if there is sufficient interest shown here, then I would consider developing this into a module -- these would be handmade modules, preferably in 5U format, that I would build one-at-a-time right here at Sketchy Labs. I could also sell PCBs, stuffed PCBs, whatever. Also note that these two boards could be condensed into a single board, which would be about 4" x 4".

If you are interested in this, please post here, and/or drop me a PM.

Also, before you ask, please be advised that this is not the Don Tillman circuit, nor is it the circuit that was posted on Muffs a while back. I tried simulating both of those circuits a while ago, and neither seemed to work in simulation. Also, this is not the Haible circuit (at least, I don't think it is -- I've never actually looked at that circuit -- I'm just aware that there is one). I never copy circuits, unless it's absolutely convenient!
Link to the video for iPad users:

Hey this looks very cool!

Quick questions:

Will it pass DC?

Is there anyway to change the crossfade length?

Can it be 6 or even 8 channels?


Dcramer w00t
Dave Peck
Hmm. Neither the embedded video in the first post nor the link in the second post work here.... post #1 is a blank white rectangle. post #2 does to a DOS-y looking web page, not a Vimeo video.
update - embed works ok now.
That looks/sounds cool!
thumbs up

[Edit] Hummm... any way to shift/offset what it is scanning between? Like, I imaging its 1 to 2 to 3 to 4 - any way of 3 to 1 to 4 to 2 etc? Or am I completely misunderstanding the concept hihi
Dave Peck
Very cool!

I think Dcramer's question about crossfade time is important. This function can make a module like this a lot more versatile.

There is a module like this in the Nord Modular G2 which has variable crossfade time, so that it 'switches' between sources at one extreme of this parameter, and crossfades smoothly at the other with no 'dwell' time on each source as we're seeing in the video (example: as soon as source 1 fades out completely and you have only source 2, source 3 immediately begins fading in and source 2 starts to fade out). The behavior demonstrated in the video would fall somewhere about half way between the two extremes of switching behavior and smooth crossfade behavior.

It may be useful to extend the behavior of this crossfade parameter even further, so that the sources overlap even more and you get all four sources at even volume at the max setting. As you begin to move the setting away from max, you begin to hear changes in the levels and a crossfade effect is gradually introduced. This would be really good for creating subtle motion in an audio signal or complex CV source.

So that would be three controls:
- center position,
-crossfade amount (less would be 1-2-1-2, more would be 1-2-3-4-3-2-1-2-3-4),
-crossfade 'slope', from switched to smooth crossfade to all signals mixed.

Dave Peck
.... and it just occurs to me that this looks like it does the same thing as the SSL Segwencer module:

Nothing wrong with that, just wanted you to be aware of this.
Dr. Sketch-n-Etch
OK, thanks for you interest! I'll try to answer your questions.

First, Dcramer:

1) Will it pass DC? Yes. It can be used as a CV morpher.

2) Is there a way to change the crossfade length? I think of this as the crossover voltage. In the demo, it was set to roughly 2.5V because this opens the linearized 2164 VCAs exactly halfway. The plateaus of the morph voltage trapezoids are at about 5V (set by a 5.1V zener diode) which gives unity gain from the VCAs.

The crossover voltages can be changed easily so that the trapezoids are fully separated (with silence between the channels) or so that the plateaus all bleed into each other at unity gain. The problem is that changing this changes the entire width of the response relative to the morphing CV. The relationship between the width voltage and the fader's centre morphing voltage is perfectly linear, so I can arrange the circuit to compensate for the effect quite easily. However, it also changes the effect of the incoming CV -- if the widths are changed, the incoming CV will have a larger or smaller effect for the same setting. The maximum effect of incoming CV is set by a fixed resistor, so that may be a bit harder to compensate. I'll tweak it and get back to you.

3) Can it be 6 or even 8 channels? It can be 12 channels, bucko! However, more channels means more circuitry. It would actually be better to have 4-channel units that are chainable. If two 4-channel units were chained, then it could be 6 channels if only channels 2 through 7 were used. I haven't made this one chainable, but it should be fairly easy to achieve. As with the 4-channel unit, the challenge is in how it processes the fader and CV voltages to set the overall width of the channel trapezoids. I'm probably smart enough to figure out how to do this. Again, I'll play around and get back to you.

Now, mckenic:

4) Can the channel order be changed (from 1-2-3-4 to, say, 2-3-1-4) without pulling patch cords? Short answer: No. Longer answer: I could design a channel shuffler based on digital switches that would allow programmed shuffling of the four channels to their 24 different possible arrangements with a push-button or something, but I hadn't thought about it. This would add complexity and cost to the circuit, and would require some sort of LED numerical readout on the panel. It would be cool -- indeed, just that feature without the morphing might be cool as a separate module. I'll give it some thought.

Now, Dave Peck:

5) It requires 3 controls: Centre, Amount, and Slope...? Centre is the fader, Amount is the pot. Slope is what I haven't implemented yet, and the challenge is that is interacts with Centre and Amount. I can easily fix the interaction with Centre (in fact, I've already figured out how to do it), but Amount will be a bigger challenge. I may not be able to decouple them. I may have to put another VCA (or -- horrors -- and OTA) in to change the gain of the Amount CV as the Width voltage changes. Again, I'll have to put my engineer hat back on.

6) How does it differ from the Segwencer? I wasn't actually aware of this module. It differs in several key aspects:

a) It won't cost $299. It'll be less.

b) It doesn't have a "SLOPE" feature, only the "PEAKS" feature. I could add something like the SLOPE feature, but I don't really see the need.

c) Looking at the Segwencer voltage diagrams, it looks as if they are triangular peaks above a certain voltage, which would make it difficult to dial in true unity gain. Mine are true trapezoids, with flat tops at 5V, which means that unity gain for each channel is easy to locate.

d) It's not clear that the Segwencer has a Centre control, so that the Depth CV seems to always emanate from the centre of the channels (between channels 2 and 3). I could be wrong about this, but I don't see a control on the panel that would move the centre. This, if true, severely limits the usefulness of the Segwencer. With mine, it would be easy to make it so that it only scans between channels 1 and 2, or 2 and 3, or 3 and 4, or 1, 2 and 3, or 2, 3 and 4, simply by moving the Centre fader and adjusting the Amount pot.

OK, I think that covers all your questions. Thanks for the "virtual beta testing." I will work on the Slope control (Overlap may be a better name for it) to get it working seamlessly with Centre (fader) and Amount (pot) and get back to you when I have something a little better.

Doc Sketchy
SlayerBadger! SlayerBadger! SlayerBadger!

looks like a lot of fun. love
Thank you for the thorough answer sir! That is very interesting indeed and I understand it would add way more complexity or even an additional module. Thanks for even considering it!

I have 2 of Mathias' SeqSwitches that I use and love - this looks like a very interesting alternative/addition to that paradigm and I'd be interested in one for sure (although I am a peasant and would need something to fit 3u *cough*cough*).

thumbs up
Scanners are useful for all sorts of duties and fun: panning, folding, sequencing, multiplexing ...

In Euro, Josh Toppobrillo designed the Mixiplexer, a wonderful little device:

I only wish it was eight channel, then one could do some of the extended tricks possible with Grant Richter's Analog Tracking Generator:

They're evidently not very well understood devices. Are we going to discuss Scanners or are you just trying to flog us something, Dave? razz
Very cool!
Dr. Sketch-n-Etch
Navs wrote:
They're evidently not very well understood devices. Are we going to discuss Scanners or are you just trying to flog us something, Dave? razz

Me, flog? Perish the thought! hihi

I'm just now also trying to understand scanners, which is why I built one. Building something is my way of coming to an understanding about it.

Incidentally, this afternoon I have figured out how to get all three of Dave Peck's controls to work seamlessly (I only had to make one relatively small change in my design to get it all to work without hassle) -- I will kludge my existing PCB with the change (it requires an additional opamp, so may be a bit ugly, actually it will require a small daughterboard... indeed, it's going to necessitate a completely new layout -- oh well, it's not like I haven't been sitting in my easy chair (my "nerve centre") too much anyway -- at least when I'm sitting on my ass in front of the TV, I'm actually doing something constructive) and post a new demonstration video, hopefully later tonight. I'll use four VCOs making four harmonious pitches to demonstrate this, I think, and I'll show the scanner voltages on the scope.

But, to seriously address Navs's question: I build this crap for fun. I make a respectable amount of income from licensing my designs to Intellijel, but the stuff I build myself is strictly for my own enjoyment. If I sell any of it, that just gives me an excuse to go out into the garage and make more of it. Even though I "could" live off of Eurorack, making these goofy Doc Sketchy boards is still my hobby, and I'm as addicted to it as ever. Please feed my addiction!
d'oh! as soon as I asked question I got to thinking about chainable units. I think this would be cool as it opens up really interesting control patches! w00t
I think about how I've practiced the 24 different permutations of diatonic 4-note close-voiced arpeggios most of my life on the guitar. A stand alone channel shuffler could take any 4 channel pitches I select on a module (e.g. COA Programmer column, or Meng Qi Voltage Memory outputs 1-4) and provide instant access to any of the (24) 4-note permutations/ sequences.

What would that shuffler interface look like, and how could the shuffler be cv'd to sequence the different sequences/voicings?

Is this trivial to accomplish with the monome and ansible series, or with another complex module? If yes, is that interface in the optimal range for controlling 24 options without excessive complexity or ambiguity?
Dr. Sketch-n-Etch
So, progress update: I completely redesigned the scanner circuit (actually, just one part of it) to do all the things Dave Peck suggested. I decided that I didn't want to kludge my existing board, so I made a new one. However, I've only gotten as far as making the actual PCB and stuffing the jumper wires (which I still have to solder):

Tomorrow I'll stuff this new board and make a new video. Before I go to bed, I'm going to figure out how to chain these together to double the channels. It should be fairly straightforward, I think.

Finally, dbernhardt58, I really don't think that such a 24-possibility shuffler would be very complicated at all. I'm going to think about that tonight and get back to you. It can probably be done with a couple CMOS logic chips and a couple DG40X switches. As far as the interface, it could be something as simple as a single long fader connected to an ADC. This fader would provide a voltage to the ADC, and this voltage could also be provided by CV. I'll probably have something like that designed before I go to bed.
Dr. Sketch-n-Etch wrote:
...Please feed my addiction!

Ok, I'll take an 8 channel version in 3U at 16HP and

Thanks! grin
Dr. Sketch-n-Etch
Well, today I built and tested the new board, and it worked perfectly the first time. I didn't have the energy to make a video, so will have to do that tomorrow.

The new board has three controls:

Centre (the fader, which serves as a manual morph control, and also positions the centre of the CV with respect to the four channels)

CV Depth which attenuates the CV input (and effectively limits how much of the "fader width" the CV will cover)

Width. This one is a little bit difficult to explain. The way this thing works is that a control voltage ramp is shifted with four voltages and fed to opamps which amplify it by a factor of 5. The channel-controlling voltage trapezoids have sides which are defined by this gain ratio. Where these trapezoids cross through zero volts is determined by the fixed voltages which are controlled by the Width pot. Hence, this pot determines the width of each trapezoid relative to the width of the entire field over a voltage ramp. If the fixed voltages (and hence, the Width setting) are large, then the trapezoids are well separated and there will be silence between each channel. If the voltages are small, then the trapezoids will overlap. At a certain Width setting, the voltages will overlap at 2.5V, which gives 50% gain in the VCAs. If the Width setting is turned fully CCW, then the voltage gaps are zero, and all the trapezoids overlap. At this setting, four different VCOs will play a four-note chord.

There is one trimmer on the board. This trimmer adjusts the centre of the fader. The way I have adjusted it is as follows: I set the CV to zero, the Width to a fairly small value (such that the trapezoids overlap at some moderate voltage), and put the fader on its centre detent. Then I look at the Channel 2 and Channel 3 voltages on a scope, and adjust the trimmer until they are the same. It's a very easy and quick adjustment to make, and could also be made with a DVM (or, better still, two DVMs).

One problem with this design is that, if the Width pot is turned too high, the 1st and 4th channels "fall off" of the fader. This is because the voltage gaps become so large that the fader cannot generate a voltage large enough to reach those channels. (Also, if the fader is centred, then the CV may not be large enough to reach those end channels, even at full strength.) I have limited the voltages to the fader by putting fixed resistors on either end, but this has limited the range too much. However, if those resistors are removed (and the fader is allowed to swing between +15 and -15V), then all four voices are crammed near the centre of the fader when the Width pot is set to lower settings. This means that there really needs to be a fourth control:

Range. This control will simply be an attenuator on the fader's wiper. That way, the range of the fader can be widened at high Width settings, and narrowed at low Width settings. I'll add that control tomorrow -- luckily, this is not a PCB kludge, but simply a "panel" change.

This is actually a lot of controls, and they are not all necessarily very intuitive. I'm going to have to play with this thing a bit to get a real feel for it (and that's why I didn't make a video today -- I want to "practice" with this thing a bit before I commit my "performance" to video).

Until tomorrow, then...

Oh, and one more thing: I discussed this module with Danjel, and he confirmed that he has no interest in making this an Intellijel product. Hence, there will be no eurorack version of this for a while. I will only be making 5U versions of this. I'm actually considering creating my own "brand" or eurorack to make things (at my own financial risk, of course) that Dan isn't interested in (and perhaps to make some things which have been discontinued at Intellijel), but I haven't acted on this impulse yet. I don't want to handmake eurorack modules, though, in any case, so sorry, Navs.
supercool. I would like a pcb and or a schematic, if the latter is something you're into sharing - I'm learning the basics so am finding making my own layout from schematics a useful tool for absorbing general principles and specific topologies - happy to pay for and not share, use or abuse.
waah noooooo.... Eurorack needs cool utilities! Someone needs to look at doing it in 3U cry
Dr. Sketch-n-Etch
Merry Christmas! I made a new video. I had to upload it to Youtube because my Vimeo account is full for this week.

I put the scanner boards behind an old panel I had laying around (which I soaked in acetone to get the old graphics off). I figured out how to get the range of the fader (which is now a pot) to stay constant as the Width control is changed. However, I would like to do the same thing with the CV but I haven't figured out how just yet (and it may not be possible). Also, you'll probably notice that the range of the Fader pot seems a bit narrow (with all four voices confined to about 50% of the pot's rotation). This is because I sized some resistors on the board assuming the fader I was using, which was 50k, but the pot is 100k, so the limiting resistors I used are a little too big. In any case, it is a fairly comfortable range for the pot, so I think I'll just leave them in.

I'm reasonably happy with this thing now. I'd be interested to hear what y'all think (especially Dave Peck, whose suggestions I tried to implement). Cheers!

Finally, dbernhardt58, I really don't think that such a 24-possibility shuffler would be very complicated at all. I'm going to think about that tonight and get back to you. It can probably be done with a couple CMOS logic chips and a couple DG40X switches. As far as the interface, it could be something as simple as a single long fader connected to an ADC. This fader would provide a voltage to the ADC, and this voltage could also be provided by CV. I'll probably have something like that designed before I go to bed.

A single fader or knob is one way to get a choice of 24 options. Math and music theory is probably full of many approaches to sequencing and pitch sets/ selection - none of which I know anything about.

But for me, musically the first pass is: three knobs with cv: the first knob/cv to select the first pitch from all 4 choices; the second knob/cv to select the second pitch from the remaining 3 choices; then the third knob/cv to select the third pitch from the remaining 2 choices; obviously the fourth and last pitch is the one left over. Seem like it would be lots of fun to send cv's to these three points and see how the 4-pitch (or whatever signal) pattern is permutated. This mirrors the way I commonly think about playing arpeggios on the guitar.

The next iteration is a slight paradigm shift: maybe you don't want the first knob/cv (with a choice of 4 pitches) to select the first pitch in the sequence; maybe you want a choice of which of the 4 positions in the sequence that first knob/cv will control. And so-on for the remaining two knob/cv's.

To me this is an interesting thought exercise that is part of my learning about the synth. It lead, of course, to another paradigm shift: I realized that I probably can get a similar level of control by using the RYO VC Seq as pitch sources, and VV COA3 Programmer outputs to drive the VC Seq's CV/CLk-in; when you send a voltage instead of clock pulse, the VC Seq output can be selected specifically from the 8 choices available.
Nice module Doc! It's just missing some flashy light eh? hihi

I think it would be fun to run some S&H, Lfos and envelopes into the thing and then send them flying out to mash up some FMd VCOs and filters and whatnot.

Kind of like an arpeggio of CVs that could then be mult'ed and patched out sending the modular through the wall and out into space!

woah woah

I don't know if the module will do that, but that's what I was thinking..

How would one classify this module? Is this a new utility? Are there other modules that do something like this? Maybe they do this in a different approach?
hmmm..... hmmm.....

Well, it got me thinking and, that in itself in awesome.

I had to look up interpolating… And now the phrase "interpolating scanner" is tripping me out…

If the three inputs had individual outputs that fired when the scanner scanned them. Then of course you already have the mixed signal output. All the knobs could be CV-able…. Maybe a mix switch that could mix any one of the inputs with any other input only or a combination of inputs… A clock in that is CV-able...

You could call it a galactic scanner…. hihi

JUST SPITBALLING over here….. Not suggesting anything! Just the morning metal acrobatics… Pay no heed

Thanks for sharing. Keep up the good work bud!

thumbs up
yes..... nice progress. applause
i clearly need a seq. of some sort.

looking fwd. to one of these. thumbs up
Dave Peck
Dr. Sketch-n-Etch wrote:

I'm reasonably happy with this thing now. I'd be interested to hear what y'all think (especially Dave Peck, whose suggestions I tried to implement). Cheers!

Hey, you're quick! Yes, this thing is looking pretty interesting. Some comments:

I don't know how useful it would be to have those large 'silent gaps' between the stages when you turn up the 'range' control as shown in the video. I was thinking it may be more useful for the range control to operate like so:

1. At full counterclockwise, it sends a mix of all four inputs to the output, as we hear in the video.
2. At 12 o'clock-ish, it 'narrows' the crossfade region of each input so you get an even crossfade from 1 to 2, 2 to 3, 3 to 4 as you turn the top knob or as controlled by a CV input. It is always outputing either one input or a crossfade of two adjacent inputs. We hear this in the video, but the range control is barely moved away from full counterclockwise when it's producing this effect.
3. At full clockwise, it changes from even crossfading between adjacent inputs to cross-SWITCHING between adjacent inputs, so it becomes a four-way electronic switch. And yet it still produces a constant output of one of the four inputs with no silent gaps.

But it occurs to me that this may be really difficult to implement on one continuous range control... SO -

You could keep the 'range control' design as is, with the range control going from sending a mix of all four inputs to the output, to crossfading between adjacent inputs, to fading between inputs with silent gaps between them, AND - add a toggle switch to completely bypass whatever the range control is doing and change the output to a CV SWITCH type output. So when you activate this switch, the CV amount control still determines whether you are scanning a short range, and hearing a switched output of two adjacent inputs, or a wider range and hearing a switched output of three or all four inputs.

This could be implemented a few different ways, like by boosting the level of the internal VCA control signals and clipping them severely to convert smooth fades to almost instantaneous on/off, or by using simple logic compare circuits to turn each region's VCA on & off as the CV passes specific threshold values,etc.

This way, you have all of the crossfade options of the design as is, including the silent gaps if you want, but you also have a way to scan between, and then output, only one input at a time with a switched output instead of a crossfade output, with a constant signal present at the output and no silent gaps.

I think this could make it a lot more versatile for both audio and CV processing purposes. Waddaya think?
Dr. Sketch-n-Etch
Dave Peck, I think those are interesting suggestions. I'm going to explain how this scanner works, so that you understand what the controls are actually doing, and what their limitations are.

So, the core circuit consists of four identical pairs of opamps. They take an incoming ramp of voltage (CV), and convert it into a trapezoid with a rising side that is 5 times steeper than the incoming ramp, a flat top at 5V of fixed width, and a falling side that is also 5 times steeper than the incoming ramp (but negative). The "floor" voltage when the trapezoids are "off" is about -0.6V (one diode drop below zero). The four trapezoids are always identical in terms of their width relative to the incoming voltage. These trapezoids are used to turn four identical linearized 2164 (Irwin) VCAs on and off, where 0V is off and 5V is unity gain. One can process whatever one wants with these four VCAs.

The distance between the four trapezoids is determined by the Width control (the red knob in the video). This control is just a pot connected to the +15V rail and ground. The wiper goes to a buffer, and the buffer output goes to an inverter to generate the negative voltage. The positive and negative voltages go to a string of resistors strung between the two voltages to generate a symmetric set of four voltages which are equidistant and centred on 0V. These four voltages are the reference voltages for the trapezoids. Hence, if these voltage differences are large relative to the (fixed) widths of the trapezoids, they will be spaced apart (with silent gaps). If the Width control is turned fully CCW, then the voltages are all zero, and the trapezoids all have the same reference voltage, and therefore coincide.

Making the voltage references symmetric about 0V was the key to fixing the fader control, which is also fed with the positive and negative rails. Hence, the centre of the fader is always at the centre of the trapezoid field.

So, if the Width voltage is high (say, 10V or higher, such that the trapezoids are, say, 3 or 4V apart), then the entire voltage field of all four trapezoids may be, say, 12 to 16V wide. That means that the incoming CV, which is probably a triangle wave with a peak-to-peak range of 10V, is not wide enough to access the full width of the trapezoids, and voices 1 and 4 may not be heard. Of course, I can increase the gain of the incoming CV so that it can always reach the outer voices, even at the highest Width setting. Alternatively, I can limit the Width setting to some value where there is a significant gap between the voices, but the CV at unity gain can still access all voices from the centre position.

So, as you can see, there really isn't that much that can be changed here. One thing I was thinking of was to replace the Width control with a rotary switch that selected a limited number (4 or 6, say) of fixed widths that were meaningful -- say, 0 (all four voices sounding simultaneously), 50%-gain-at-crossover, nearly separate, separate, and quite separate. If it were a double-pole switch, then I could also use it to select gain resistors for the CV so that the CV always covered the same width of field. What do you think about that?
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2, 3 ... 13, 14, 15  Next [all]
Page 1 of 15
Powered by phpBB © phpBB Group