Sample and hold / "random" voltage LED driver

[nocentelli]

I'm hoping someone can help me out. I'm in the process of putting together a souped-up Uglyface: I have it hooked up to a 4017 based 8-step sequencer, with a pair of LDRs in parallel so both the the envelope and the sequencer can diddle with the frequency simultaneously, with a separate level pot for each (on the respective driving LED).

I am trying put together a circuit for a "random" mode to be switched in place of the sequencer. I've played extensively with the Vanishing Point circuits, which utilise a 4093 to skip the rate of the 4017 around. I have also played with rg's concept of a random LFO, linking several simple inverter oscillators together.

However, the effect i'm really after is like the Maestro FS+H: A regular pulse of random voltages, with the rate set by a pot. I had a play with the Maestro circuit, but it appears to blip a fairly small control current to joggle the OTA filter: I don't know if it is possible to adapt the schematic to produce big voltages that would drive the LED effectively. I'd ideally like to run it off the same LFO that clocks the 4017 (tremulus lune -ish).

Any advice or pointers would be gratefully received. I'm wondering if there are any dedicated S+H chips out there, I haven't come across any utilised for this sort of application.

[Transmogrifox]

You can sample a voltage with a JFET (think active bypass circuit).  You pulse the gate of the JFET with a periodic narrow pulse at the update rate you want.  You "hold" the sample in a capacitor against a high impedance like a MOSFET or another JFET gate.  Probably 2-10nF is appropriate.

Then you either use the typical random noise generation with a reverse-biased BJT or you can make a ramp generator at a high enough frequency (like 30+ kHz) so the jitter and drift on your oscillator makes for a relatively non-deterministic voltage at the time you sample it.

If you need to convert a normal LFO edge to a fast sampling pulse, then you feed it through a capacitor to the base of a BJT and amplify it.  You would be able to get  a short enough sampling pulse every time around to nail something unique off a 30 kHz ramp oscillator.

[digi2t]

This might help;

http://www.diystompboxes.com/smfforum/index.php?topic=110113.0

http://www.diystompboxes.com/smfforum/index.php?topic=110293.0

Works fine in my Infinitphase, and yes, I've thought about the Uglyface angle too, but haven't gotten around to it yet. If you need the vero layout, let me know.

[nocentelli]

You can sample a voltage with a JFET (think active bypass circuit).  You pulse the gate of the JFET with a periodic narrow pulse at the update rate you want.  You "hold" the sample in a capacitor against a high impedance like a MOSFET or another JFET gate.  Probably 2-10nF is appropriate.

Then you either use the typical random noise generation with a reverse-biased BJT...

Ok, this sounds like a description of the Maestro: The "holding" JFET then goes to a pnp bjt, and from there to the control pins of the OTA. The cv output from the bjt collector will not drive the LED to ground, however. I'm thinking perhaps I could just take the signal the from the collector, and use an opamp to increase the voltage so it can drive an LED. I've only used opamps as audio amplifiers before: I guess the "cv output" is DC?

This might help;



Works fine in my Infinitphase, and yes, I've thought about the Uglyface angle too, but haven't gotten around to it yet. If you need the vero layout, let me know.

Thanks! I remember that thread now, I'm going to try this as soon as I get the LF398: I'm guessing you drove the LED direct to ground from the output (pin 5)?

[mth5044]

You could try one of Electric Druids TAPLFO chips. Overkill, but you can hatdwire the wave shape to random and a lot of the other pots, or use trimmers. I had a lot of trouble with the Maestro S/H side, would have gladly gotten a few more paets to make life a little easier!

[nocentelli]

TAPFLO is a cool idea I hadn't considered. It would be great to have the pic LFO clock the sequencer that drives the LED or have it drive the LED direct using S+H, and all with tap tempo control! I look at the schematic for the tap clock - http://www.electricdruid.net/images/lfo/TapClock.gif - and see the clock output is marked 0-5v 25ma, is that likely to be similar to the 4017? Time to look at datasheets.

[digi2t]

Thanks! I remember that thread now, I'm going to try this as soon as I get the LF398: I'm guessing you drove the LED direct to ground from the output (pin 5)?

Yeah. Just one addition though, a 1M trimmer at the output to adjust the intensity. There was a bit too much pop running it flat out.

Here is the vero;

[nocentelli]

A 1 megohm trimmer in series with an LED seems quite high value.

[digi2t]

Well, quite honestly, I wasn't quite sure at what point it would be OK, so I figured a 1M would be more than adequate to do the job. In the end, I found that I'm actually somewhere in the middle of the range, but YMMV. In the Infinitphase case, the S/H output is tapped in before the opamps driving the LED's, so the required signal has to be really toned down to get the range right. According to Mike at Prophesysound (without going into any details), their S/H add-on goes direct to the LED's, but when I tried that, I found that it induced a lot of popping into the audio, which I really didn't like. I decided to go before the opamps, attenuate the output, and use the Smooth pot to vary the intensity instead. I find it's better, allowing me to dial the S/H down to a very subtle attack.

In the Uglyface case, you probably won't need the last trimmer, but I would breadboard the S/H, and test though. The nice thing is that it's fairly low parts count, and other than having to audition for the right transistor, easy to make work right.

[digi2t]

Sorry, I just realized that the schematic I posted isn't the right one. Here is the one the vero is based on;



Just add the 1M trimmer at the output, if you need it. If not, pin 5 direct out. The noise section was what was suggested to me by Mike. I believe that he said that this was what is actually used in the Infinitphase. It works great. Just remember to use sip sockets for the noise transistor, because you might have to audition a bunch to find one you fancy. Mine ended up being a 2N5401, but it could be anything really.

I haven't tried this at +/-9v though, but I assume it should work since the 398 is rated from +/-5v to +/-18v. You might have to tweak some values for 9v operation, like R99/R100 maybe. You'll need a 1044 (or similar) supply as well.

[nocentelli]

Thanks. Waiting on the LF398 for further progress. I notice the the noise generator transistor+opamp section looks similar in structure to the Maestro, except for the noise transistor pins: the original maestro uses the emitter to connect to the opamp, with the collector connected to -9v and the base left floating. I altered this to have the base at -9v and the collector floating (at aquatuar's suggestion) and it works great. Now this schematic has yet another arrangement... some experimenting to do yet.

[PRR]

> the noise transistor pins

The main thing is to get the Emitter-Base junction reverse-biased enough so it breaks-down, acts like a 7V Zener.

If you leave the Base open, then current flows through the Base-Collector junction as a forward-biased diode. This has no real effect as a hiss-source (slightly better tempco as a Voltage Reference), eXcept the breakdown voltage is ~~0.5V higher.

On a 15V supply, 7V or 7.5V in the Zener is no real difference.

On a 9V supply, some difference (R100=470K may give a more robust hiss).

On a saggy-battery "9V" supply you sure want the B-E connection, not the C-E with B left open, for the 0.5V less drop.

Note that the S/H input and output are +/-12V (or +/-9V). If you run an LED output to ground, half the samples will be negative and 10%-50% of the positive samples will be too small to light the LED.

Running the LED up from the negative rail will make more pulses right-way big-enuff to light the LED.

I do not see ANY output load spec on the LF398? (Infinite short, but no clue what current that will be.) We don't even know if it has the guts to pull an LED. My suspicion (from the technology of that era) is that it has 5mA-10mA.

For +/-9V supply, 2V-3V LED (not White), maybe output to 1K-2K to LED to V- would work fine.

[Freppo]

Is there a specific reason to use a CD4093 here? Can I just use a CD40106 insted?
With both cd4093 inputs connected it's configured just as an inverter right? Or am I mistaken?

/ Freppo

[noisette]

Is there a specific reason to use a CD4093 here? Can I just use a CD40106 insted?
With both cd4093 inputs connected it's configured just as an inverter right? Or am I mistaken?

/ Freppo

Hi Freppo!
No. Yes. Yes. No.

[Freppo]

Hi Freppo!
No. Yes. Yes. No.

Thanks