The Re-Animator

Uncle Krunkus

About six months ago, I suggested to Pehr Wikstrom that the best way to stop a Lunetta from producing digital noise would be to use a common reset to reset all the counters, flip-flops etc. at regular intervals.
That gave me an idea which I played around with in my head for a while, and finally I decided it was worth putting on a breadboard.
I've waited until I got some good results before starting this thread, as I wanted to be sure it was worth documenting. I haven't got a web page to fill with my results from these experiments, so I'm going to describe them in detail in this thread.

The core of the idea is to use a squaring amp to extract a sync pulse from an incoming waveform. Then to use that sync pulse to reset an 8bit counter which is free running from a high frequency master clock. The outputs of the main counter are fed into an R/2R network and the result becomes our new waveform.

Now in this simple form, all you would get is a sawtooth wave which restarts in sync with the incoming waveform. But, with some digital manipulation of the master clock pulses, I realised that it would be possible to change the gradient of the sawtooth wave, dynamically, every time the wave is re-started. This will result in curved waveforms. Another small piece of logic will enable "wrapping" of the wave at top and bottom (triangle) or not, and last, but not least, I've designed a little voltage controlled digital encoder which will allow realtime manipulation of, Initial level (phase), Gradient, and Integral (curviness) of the wave.

I'm also thinking about a pre-scaler which will automatically keep the wavelength resolution about the same, (roughly 256 slices per wavelength) regardless of the incoming frequency.

Before anyone points out that all this could be achieved with 2 chips, a CPLD and some nifty code, I want to point out that this is basically a Lunetta on steroids with just one primary function, and as such is meant to be a starting point for experimentation. It also contains some ideas which could be used separately in a Lunetta for other functions. Most notably, the sync input circuitry (which I'd like to enhance even further so people could plug 'anything' into their CMOS condos) and the voltage controlled digital encoder (which makes it possible to control any function in a Lunetta with a standard CV input) I plan on building it to a point where it becomes a self contained waveshaper. Keep in mind that this could involve as many as a dozen chips, and quite a bit of extra circuitry. I'm doing this 'cos I like playing with CMOS chips, that's also why I've pointed out it's relevance to Lunettas.

Here is a .gif of the schematic as it now stands on the breadboard. It doesn't yet include the integral controller, which will really shift it's value to a new level. Because all of this part works, as it should, I'm confident that the integral controller (which I'm adding to the breadboard now) will hit paydirt. This is also why I've decided to start this thread and (hopefully) get some feedback on ways to improve what's already there.

I'll add some samples of the waves I'm getting out compared to the incoming wave over the next couple of days.

BTW If anyone can think of a more accurate name for this circuit, let me know.

I will update this post of the schematic, as it improves.
Last update - 13/7/09 ver1.2

Re-Animator-(1)-13-7-09.gif

Description:

Filesize:

26.33 KB

Viewed:

565 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Re-Animator-(2)-13-7-09.gif

Description:

Filesize:

71.65 KB

Viewed:

387 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Re-Animator-(3)-13-7-09.gif

Description:

Filesize:

73.83 KB

Viewed:

350 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Scott Stites · Posted: Thu Apr 02, 2009 10:25 am Post subject:

This is very cool, Andy! I'll have to go through my inventory and see if I have any CD4029s.

Cheerios,
Scott

synthmonger · Posted: Thu Apr 02, 2009 10:33 am Post subject:

"..Lunetta on steroids.."

Now that's what I like to hear!

This looks awesome. If you get a chance, could you please post some scope pics?

Uncle Krunkus · Posted: Thu Apr 02, 2009 2:42 pm Post subject:

Okay,
I've already got a problem with the schematic as it stands right now.
I'd been listening and looking at the output straight off the main R/2R top rung, and hadn't put the TL071 in. I put it in last night, and I get part of the wave being replaced with a high frquency hash. It's right where the main counter resets, so I'm thinking that, at that point, the single 9K1 around the TL071 is just too small to stop it from sinking the R/2R in some way and let a load of yuck through. (that's highly technical talk which means I don't know enough theory! Laughing

)
Anyway, I'm gonna try a 100K in series with the input, and a 91K feedback resistor.
If it works okay, I'll fix the schem.
Anyone got any better ideas?

Uncle Krunkus · Posted: Thu Apr 02, 2009 3:06 pm Post subject:

Jesus!
That worked!
I'm starting to get the scary feeling that I actually know what I'm doing!!! Laughing

Just don't have the descriptions down pat I s'pose. Rolling Eyes

Okay, 100K on the input and a 91K feedback, and I added a 33pF across the feedback as well to stop the top end a bit, and I also added a couple of decoupling caps (100nF) on the TL071. That reminds me, I haven't put any decoupling caps on the schem.

Hey, while I'm here, does anyone know how to incorporate a 14Khz lowpass into this output TL071? Maybe just put a bigger cap in the feedback path?

Uncle Krunkus · Posted: Thu Apr 02, 2009 7:06 pm Post subject:

Schem updated.
I changed the 91K to a 33K to bring the levels down to a max of 5Vp-p.
I added the 33pF around the 071 to stop the HF stuff.
I also biased the non-inverting input to drop the wave around the 0V line.
I haven't done the decoupling caps yet.
Gonna go see if I can get the gradient control done.

Uncle Krunkus · Posted: Thu Apr 02, 2009 9:05 pm Post subject:

Great news!
The integrator is working! Cool

At the moment, I've got spirelli pasta shaped waveforms in sync with an incoming 1khz!!
I just need to slow down the clock on the gradient counter a bit, and I'll post some shots of the scope.
Very Happy

loss1234 · Posted: Thu Apr 02, 2009 9:33 pm Post subject:

wow

what a week for intersting circuits!!

I must say, if thing works like you explained, i can see a million uses for it.
I too LOVE playing with CMOS and for some odd reason would rather work with CMOS than a 2 chip solution.

i think this could be great for ALL kinds of things. i wish i didnt have to work tomorrow as i want to build this PRONTO but i KNOW what i will be doing all weekend!

when will we be seeing the Gradient and Wrap circuits?
thanks so much

loss1234 · Posted: Fri Apr 03, 2009 4:19 am Post subject:

let me ask...without having the integral controller yet, should i bother trying to build this AS IS or should i wait?

thanks

seraph · Posted: Fri Apr 03, 2009 5:06 am Post subject:

Scott Stites · Posted: Fri Apr 03, 2009 6:34 am Post subject:

Cool! Carlo built the circuit and has posted his first scope shot!

Laughing

Uncle Krunkus · Posted: Fri Apr 03, 2009 10:28 am Post subject:

Yeah!
Well, actually, if you used a few of those to build sawtooth shapes, that's what it's like! Very Happy

Loss, at this stage, I'd wait and see if we can improve it even further, but you're welcome to start experimenting. It's a hairy breadboard though. Shocked

seraph · Posted: Fri Apr 03, 2009 10:29 am Post subject:

Uncle Krunkus · Posted: Fri Apr 03, 2009 2:59 pm Post subject:

This is a copy of an e-mail I just sent Scott.
We have been talking about the possibility that the front end of the ReAnimator (enhanced to guarantee a clean sawtooth from any input (something I planned on anyway)) could drive his sub-octave wave folding circuit.
I'm posting it here, as I think it includes some useful info,

loss1234 · Posted: Fri Apr 03, 2009 4:34 pm Post subject:

this is actually what I was struggling with all week in my experiments with Scotts Wavething....I was using a guitar fed through a lot of circuitry to arrive at a sawtooth and i was looking for another way to turn ANY signal into a sawtooth....a Solid sawtooth that i could then use to drive the wavething..

I dont know enough about your circuit to know what it currently will do or wont do so it is a bit hard for me to completely relate to the email you sent scott as it refers (i think) to parts of your circuit which arent in the schematic you posted

but let me say i am definitely up for testing any of this....

bit sad to find today that i only have 2 4029's....damn

now i am going to have wait at least a week....boo

can i ask how you are going to change this to guarantee a clean sawtooth?

looking forward (and often) to any developments with this post

Uncle Krunkus · Posted: Fri Apr 03, 2009 7:19 pm Post subject:

loss1234 · Posted: Fri Apr 03, 2009 11:07 pm Post subject:

ok

can you explain the prescaler a bit more? i have never worked with one before

also...since i cant see everything you havent drawn yet, can you give me a warning about what parts i should order to start testing this thing?

i always LOVE getting NEW cmos chips to learn about!!

thanks

loss1234 · Posted: Fri Apr 03, 2009 11:11 pm Post subject:

and no i dont have any 4585s

how many of those will i need?. i dont see them on your schematic

thanks!

Uncle Krunkus · Posted: Fri Apr 03, 2009 11:26 pm Post subject:

I just uploaded the second sheet. Not finished yet, but getting there.

loss1234 · Posted: Fri Apr 03, 2009 11:46 pm Post subject:

woo hoo

Uncle Krunkus · Posted: Sat Apr 04, 2009 2:12 am Post subject:

I just fixed that second sheet, so it better reflects what's on the breadboard now.
Inhibit on the 4520 held high,
Jams on the 4029 held low,
and clocked from the clock/16

I just have to wait for my 4029s, and then I'll do a 4bit encoder for the gradient controller.

Next after that,.... I think it's time to hammer out the pre-scaler.

Chip count is at 11.
I'm estimating the complete unit to use about 20. Cool

Uncle Krunkus · Posted: Tue Jul 07, 2009 12:00 am Post subject:

Scott Stites and I have been puzzled about a certain effect I've seen with the encoder section, and I thought I'd put it out there for others to consider.
Attached is the basic 8bit encoder clock.
The R/2R sends a voltage to the comparator which sets the count up or down. Which means the counter is forced to "track" the voltage which is coming out of the pot.
The problem arises when you sweep the pot up or down at a fast rate. Instead of tracking the pot, the counter "locks up" or down depending on which way you're turning it.
I've tried a number of possible solutions, but it's still a mystery.
Anyone out there got any ideas?

Level-Encoder-3-7-09.gif

Description:

Filesize:

28.66 KB

Viewed:

300 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

Inventor · Posted: Tue Jul 07, 2009 12:29 am Post subject:

Well, I can only guess but I'd imagine that when you sweep the pot quickly you're giving the counter a noisy output. Note that sometimes chips are not implemented the way we think they are. If the up/down input has transitions near the moment of the clock edge, all heck could break loose. I'd suggest putting a 0.1uF cap from the pot's wiper to ground.

I made a controller with cheap scratchy thumbwheel pots and I found that adding caps smoothed them out nicely.

Also if you get counter noise as in the counter oscillating up and down around the desired value, then you might have too much hysteresis. As it stands, your amount of hysteresis is a function of the potentiometer wiper setting. If you can buffer the pot with a voltage follower and then send that into the comparator with a resistor, then you will have a constant amount of hysteresis and you can tailor that resistor value so the hysteresis is smaller than one bit resolution.

Just a thought.

Les

p.s. one more thing: I think you want the hysteresis on the R2R ladder, not on the pot. It might suit you to reverse the comparator inputs and add an extra inverter at the output of the comparator.

Uncle Krunkus · Posted: Tue Jul 07, 2009 2:40 pm Post subject:

Thanks Les,
Yes, I tried a few different values of cap from the wiper to ground, but it didn't help. The problem seems more systemic than that.
I also added a CV summer and perfect diode and fed the pot voltage through that, but it still locks up.
One thing I haven't tried is the hysterisis on the ladder. I might try that next.
I also have to update that schem with the new things I've added. I'll try to get that done today.

Inventor · Posted: Tue Jul 07, 2009 3:14 pm Post subject:

Yes, that last one might be it. With 256 steps on a 5V part your bit resolution is only 19 mV. At mid-position of the wiper, you have 57mV of hysteresis. So when the counter counts up over the threshold, you get about four bits worth of change in the reference value. So it's like trying to hit a moving target.

At best this reduces the 8 bit successive approximation function to 6 bits of resolution. But it gets worse than that as you describe, Andy, in the situation where both the reference and the counter are changing.

In fact, now that I think of it, the hysteresis has it's own time constant. A 1 MOhm hysteresis resistor that sees only 30 pF of capacitance will form an RC filter with a time constant equal to one clock cycle. You've likely got 30 pF or more due to wiring and stuff, so it's possible that the time constant of the hysteresis is introducing asynchronous signals which are going straight into the counter's up/down input without being digitally buffered.

So if I could guess again, I'd say: try removing the hysteresis resistor! But that's just a quickie possibility to try. When that doesn't work you can get back to your plan of putting hysteresis on the ladder.

Interesting puzzle....

Les