Softpot Ribbon Controller

[mark_olson]

Damn, Scott.

Since I'm up to my neck in unfinished projects I've been trying hard to ignore this thread, but those samples; wow!

This is by far the most stimulating thing I've come across since, well..., the Klee.

I know a lot of Mennonites would have us believe God lives in South-central Kansas - I'm beginning to think they may be right, in Mt. Hope, to be precise.

Amazing.

Mark

[Scott Stites]

Well, she doesn't live here, but I hear she'll be in your neck of the woods September 12, at the Lied Center.

Cheerio,
Scott

[Dan Lavin]

All:

I found this video on You Tube of Scott's escapades with his new toy:
http://www.youtube.com/watch?v=fLtQsfYceXs

[Coriolis]

Harrumph! I like The Nice and ELP, but THAT video was pretty lame
I hate when people go to a concert just to talk all the way through it, but that "chase" would've had me discussing the weather all the way through...

I prefer the real Scott, err..Keith..uhm...

C

[State Machine]

[Scott Stites]

[Scott Stites]

Yipes - musta miscounted. Count is 14 ICs with the changes.

Here's what I'm shooting for - the changes clean things up a bit, leave no unused gates, and make the schemo a little easier to grok.

U1 CD4017
U2 CD40106
U3 CD40106
U4 CD4081
U5 CD4013
U6 7556
U7 LF398
U8 LF398
U9 LF398
U10 LF398
U11 LF398
U12 LM358
U13 TL074
U14 TL072

Last night I added the inverters to the 7556 inputs and put in the 4081. Haven't done the re-do on the trigger or the CD40106 clock, but they should work fine.

Cheerios,
Scott

---

appendage_rev028.PNG
Description:	Rev 0.28
Filesize:	183.16 KB
Viewed:	401 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[fonik]

[Scott Stites]

My - circular - hadn't even thought of that!

Here's an idea for an Appendage based XY controller.

---

appendage_xy.PNG
Description:	Appendage XY Controller Application
Filesize:	21.39 KB
Viewed:	276 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[mono-poly]

Woow i've been sleeping under a stone i guess.
Great stuff you guys are doing here@

[crashlander42]

Anyone find a good supplier for these? I'm searching around on my normal suppliers sites and not finding them. I've searched Google but come up with very little besides the manufacturers website.

I miss obvious things though, and this could be one of them.

[Scott Stites]

Hey Crashlander,

SparkFun is where I'm getting them - they have the best selection at least on this side of the pond AFAIK:

http://www.sparkfun.com/commerce/advanced_search_result.php?keywords=softpot&search_section=products

I've finished the changes shown in the last posted rough schematic. BTW, there are a couple of things to note: the timing resistor in the clock is 22K, not 33K. It's pretty much a 5 kHz clock (period is a hair over 200 uS). The comparator signal rise time cap is 10 nF, which is shown on the last posted schemo, but was at some other value on the first schemo. Both of these timing elements are fairly critical. You could probably go a lot higher in value on either on them without noticing much of a difference, but you wouldn't want to go much lower.

Cheerios,
Scott

[Uncle Krunkus]

Hey Scott,
If you didn't actually have a softpot, this circuit would still work with a chain of small value resistors and a stylus wouldn't it?

[Scott Stites]

I suppose it could - the glitch discriminator might be able to handle the bouncey-bounce of the sylus. The amount of slide resolution you would have would be dependent upon the number of resistors, of course. Obviously you wouldn't have the resolution of a continuous strip like in the SoftPot (it's around 20K for a 500mm strip, so how many molecules would that work out to? ).

That would induce steppiness, which would affect the slide and bend voltages. A way around that would be to introduce a lag time, but that's a rocky road, judging from my experience with my first attempts at finding a time constant that accomodated the glitches. The trick (to retain the slide and bend voltage glide effect), would be to introduce the glide after the first sample is taken - that way you start out at zero volts differential without the lag creating an offset between the two S&Hs.

This wouldn't be as difficult as it may sound. When pressure is first applied, and the delayed comparator signal and glitch discrimination agree that the voltage is stable, both initial note and slide S&Hs fire simultaneously. Immediately (just a few microseconds after that) the gate and trigger fire. This same signal could latch a CMOS switch open (or shut or whatever) that applied the lag capacitor. Your glide would still be dependent on the lag time constant, but it would glide, and you would have your bend and initial note voltages as well.

Perhaps you could make a sort of fretboard and use some guitar string. Each fret would tap into the resistor chain, and the guitar string would be the center tap. This would get rid of the stylus. In fact, an Appendage style bass might work out, though bending notes would feel unnatural (it wouldn't be a side-to-side motion). You could combine fretted/frettless bass by initiating a switch that kicked in the glide when you wanted to slide up or down. It would be a fairly easy thing to create a signal at each fret interval to create a trigger when that fret was accessed by the string. This would actually be easire to play at standard musical intervals than an unquantized ribbon. It might be a cool way to control your ASM.

Maybe, instead, one of those video tape ribbons would work out with all of this. I dunno.

Cheerios,
Scott

[Uncle Krunkus]

Wow, I hadn't even thought about the fretted resistors effect. If each of the strings was electrically isolated (tricky, but not insurmountable) they could act as independant wipers on the same voltage divider. That would work wouldn't it? That way, you could feed CV's to 1,2,3 or 4 VCOs at the same time. Or maybe use the E and A strings as your note values on 2 VCOs, with the D string controlling the VCF and the G sent to the VCA. Even without a lag to simulate slide action, it could be a great synth/guitar interface in it's own right. Especially considering all the scaling and differential control options you've got happening. You could even play it upside down!! Now that would spin the punters out! (No, I don't mean your family Jan. )

[germaniac]

It seems cavorting with those snakes was both a foreshadowing and an inspiration.

Characteristically brilliant work Scott.

[Scott Stites]

Thanks Germaniac!

Germy and I are usually on-board with each other when either of us gets a hare-brained idea or see's a nipple-hardening part for sale somewhere. We talk about some crazy sh*t from time to time.

I've improved the voltage acquisition section. I began to wonder about my claims to latency, realizing the 4017 would have to cycle twice on any note-on event. I checked it out on the scope, and sure enough it was actually varying between 5 and 6 ms. I trebled the clock frequency and tightened up the voltage check window, and now latency varies between 1.5 and 2 ms, depending upon the SoftPots mood when the finger hits the ribbon. I'm spending the weekend softly tapping and releasing (the squirrely region of the SoftPot) to make sure it is perfectly reliable. So far it's been working frikkin' great!

Here's a sample of the improved circuit, though I don't move fast enough to take advantage of the decreased latency (and to be truthful, I'm not certain one could tell a difference between 2 and 5 ms, but it was a point of pride).

It's the Appendage in Spacy mode. It's controlling three oscillators through the 2040 filter through a VCA. It's gating and triggering an EG which is controlling the cutoff of the filter and the VCA. The ADSR is set up for a very slow attack, and a not as slow release. This allows me to control the amplitude by how long the pressure is applied to the Appendage.

The initial voltage output is controlling the three VCOs; the filter cutoff, and the pulsewidth of one of the VCOs is controlled by the bend voltage. As a standard feature, the bend voltage can be mixed with the initial voltage - this allows any range of initial voltage to be controlled by any range of bend voltage. The sample starts out with only one of the VCOs, (the one that has the PW controlled by the Appendage) playing. As the sample continues, I adjust the amount of bend voltage. I then turn the bend voltage all the way down so that only the filter and pulsewidth are modulated by the bend voltage, while the initial voltage sets the pitch of the VCO. As the finger is slid up and down the Appendage, the filter cutoff and PW of the VCO are modulated.

At 2:48 I mix in the other two VCOs and start putting bend back in. The output of the VCA is passed through a digital delay and reverb (my Lexicon MX200) for extra juicy spaciness. This was all recorded in one take with no overdub or editing.

I'd love to hear this thing batting around a few Buchla VCOs and a few Low Pass Gates.

Cheerios,
Scott

---

Appendage_Meander.mp3
Description:	Wandering Fingers.......	Download (listen)
Filename:	Appendage_Meander.mp3
Filesize:	7.95 MB
Downloaded:	1158 Time(s)

---

[Uncle Krunkus]

Do you think there would be enough interest to warrant another EM-PCB run? I think there would be. We have to get the MPS out the door first though. I wonder if Mosc is ready for licking stamps again?

[bugfight]

[Danno Gee Ray]

MPS...MPS...MPS!

Oops, wrong thread.

[Scott Stites]

I'd certainly like to see it benefit em. But, well, what DGR said.....

[Scott Stites]

I've got the clock at 20 kHz. If it goes much higher than that, the resolution on the voltage window won't be enough to discriminate between valid and invalid voltages (the window is +/- 70 mV). BTW, latency is 2 to 3 mS (scope was at 2 mS sweep rather than 1 mS - whoops). Anyway, for all practical purposes, that's instantaneous in this app.

With these parameters, the rise time delay is no longer required - the discriminator is handling it just fine with no rise time delay of the comparator. That means, if my thinking is correct, that'll free up two Schmitt inverters. Which means I think I have the leeway to implement a function I think will be quite useful.

The resolution of the SoftPot is much higher than the thickness of the human finger. When slide or bend is set to a high level, one has to really employ some technique to keep the resistance from changing a whole lot when removing pressure from the ribbon after sliding up or down and holding the note. With quick slides or notes that end at the end of a slide, this is no problem. If one slides up or down then releases, it takes a very careful turn of the finger to keep a slight downshift from happening, due to the surface area reducing before the comparator goes low. If one plays with the very tip of the finger, that helps.

My idea is to implement a momentary hold switch - this would be optional to the build, but the idea is if one wants to hold the exact pitch upon release without having to think about it too much, just press "hold" and do any damn thing you want with the finger - the pitch will hold at that point.

The fact that, at high bend levels, finger pressure can be critical is something to consider when using smaller SoftPots for modulation control. If the Softpot is just a few inches long, and one is using it for bend, there is a lot of resolution there in that small space. In this case, the Appendage will have the width control that I mentioned much earlier in this thread and implemented successfully on the first breadboard.

"Width" allows one to set a "dead zone" in the bend voltage at a particular physical width. This "dead zone" is wherever you start out - top, bottom, middle. That way, you can press the ribbon without worrying about too much pressure bending the voltage before you want it to - this would be very helpful with control applications using very small soft pots. You basically would set the width to the size of your finger tip. This also helps finding "return to zero" much more easy. This could be either a panel mount pot, or trimmed at the board level if you never see yourself changing it. For instance, you may wish to use a body part other than your finger for controlling things......

Each board will also allow a break voltage to allow a "zero" space at the "bottom" of the ribbon for slide voltage. This is more or less specifically for controlling VCAs in multiple Appendage applications. You wouldn't have to hold your tongue just right to get zero volts at that last mm to completely close the VCA.

Cheerios,
Scott

[frijitz]

[Scott Stites]

I've been wanting avoid a force sensor for a few reasons. The first reason is that I don't know if I could lay one straight enough to get the SoftPot and the sensor lined up all that well ( ). Another reason is that I think it would complicate the build a bit, and the third reason is I've only found one source for a force sensor long enough to do it, and it ain't cheap at that. It certainly would make figuring out how to defeat it easier.

I've already put in a hold switch, which works very nicely. Another thing I've tried today is putting a lag into the feed from the first S&H to the bend sample and hold - any negative going signal is lagged. That actually works pretty well, though it does put a bit of a slide in hammer ons. It also tends to put a small offset in the bend voltage. This also puts a slide in a transition from a high note to a low note (hit a note high on the ribbon then hit a note low on the ribbon).

I'm using a Schottky diode to feed the rising edge across the lag resistor, but there can be enough of a differential between the two sample and holds (bend and inital) so as to put that small difference in there, which translates as voltage. My thinking is I can avoid everything but the hammer-on slide (which isn't too objectionable) by putting a FET in to switch the lag on after the first sample is taken, since that control signal is already there.

Cheers,
Scott

[frijitz]