Sample and Hold problem

[john trials]

see photo for illustration of my problem. You can see that the input signal (blue triangle wave) is VERY slow (2.5 seconds per division), yet the sample (yellow signal trace) can't keep up with it when the input is decreasing.

When the input gets faster (higher frequency) the samples really can't keep up, and the samples are always positive, since it seems like the sample capacitor cannot discharge fast enough.

schematic: https://www.yusynth.net/Modular/Commun/NOISE/NoiseSH-sch.jpg

My S&H works well only if the input signal is increasing in voltage. It doesn't sample correctly when the input signal is decreasing in voltage.

I have disconnected the power to the noise portion of the circuit, and the S&H problem still exists, so the noise is not causing the problem.

It seems like the "hold capacitor" (C5?) is not able to discharge fast enough to track a decreasing voltage. Yet it seems to me that if I add a resistor to allow C5 to discharge, then the HOLD function won't work well (the held signal will discharge, and decrease in voltage).

Any thoughts?

---

sample and hold problem.jpg
Description:
Filesize:	210.02 KB
Viewed:	381 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[blue hell]

Hmm .. I'm not too good at FET's but .. juging it's data sheet ..

pin11 of U2 should be negative ..

If it is .. I'd either replace U2 (as it apparently won't go negative on it's output (and your scope should show that too then) .. or the FET .. as that thing should be symmetrical in operation .. it should be able to get charge out of the holding capacitor (C5) just as well as it can get it in.

And when that U2 pin 11 is not negative something is wrong with the power or the power lines I'd think.

And yeah ... a resistor will not make it better.

[john trials]

U2 is powered +/-15v, and the output can go negative. In the photo, the sampled output does go negative.

I like your comment that the FET should be able to "get the charge out of the holding capacitor C5". Does the charge dissipate through R14?

I also forgot to mention that I could not get a BF245B FET, so I used a J201, which Yves (YuSynth) mentioned will work (in a different forum post).

Further testing: with very small input signals (under 1Vp-p), the S&H works well. Unfortunately, this is not very usable in real synth situations, since that is a pretty small Control Voltage.

Further testing: the S&H tracks the 5Vp-p input signal well if the clock speed is fast, but again, this is not always useful in a synth, either. Most times, I want a slow clocking.

The S&H does not track the noise well at all. It's almost like the noise just continues to charge C5, and C5 never gets a chance to dissipate any charge.

[blue hell]

No, R14 is just a bias for the FET, the opamp should set the level on C5 when the FET conducts.

Setting the level means it should be able to source and to sink current, and it's either that which is failing (the sinking bit), or the FET doesn't conduct in reverse ..which per data sheet it should do.

Hmm maybe it's gate does not go high enough .. which should be done by R14 when U2d 's output goes high ... diode reversed or broken?

Maybe look ar pin 14 U2d's output and after the diode.

[john trials]

I was wondering if the FET was allowing current to flow one way and not the other, which might explain why the S&H doesn't track well on negative slopes. So I swapped the FET drain and source leads, and that didn't change a thing, so the FET is working properly (source and drain should be interchangeable as far as I know).

So I measured the gate signal. Two photos attached. The yellow trace is U2-pin 14 (the gate signal produced by the internal clock...it sits at -15V until a rising edge comes along on U2-pin12, and then pin14 goes to +15v for about 350us). The blue trace is the node at R14, D3, and the FET gate. It sits at -15v also, but the voltage when gate is high varies!

The FET gate signal (blue) changes in amplitude when "high". It ranges from about +4v to -4V. It tracks the S&H input voltage fairly closely. In my test, the S&H input is a slow triangle signal (shown in the previously attached photo where the S&H isn't tracking well on the falling slope). The FET gate signal peaks fairly closely to the triangle wave voltage during the "high" gate signal.

Is this what the FET gate signal is supposed to look like? Is the diode D3 doing its job correctly? I am not skilled enough at circuit analysis to know if the signal at the FET gate terminal is correct in my circuit.

---

S&H gate high.jpg
Description:
Filesize:	105.83 KB
Viewed:	336 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

S&H gate low.jpg
Description:
Filesize:	113.93 KB
Viewed:	309 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[blue hell]

Yes, when U2d.pin14 goes high that signal is blocked by the diode, allowing for R14 to let the gat go to the S&H input voltage, which makes the FET conduct.

350 µs seems like plenty enough time for the FET to go into conductance mode, as the flattening of the blue line shows (the 2M2 will have to cgharge the FET input capacitance and also the input capacitance of your probe, so without a probe it will be faster even).

The scope traces also show U2d.pin14 to work ok.

Sooo .. next question is .. what happens at U2a.pin1 ? hmm .. the blue signal actually tellsus that, it seems to be fine too.

Leaves only that FET to be suspect IMO.

Let me look at the data sheet for the J201, had look at the BF one only, ok ... looks like that one should just work too.

[john trials]

Blue Hell, thanks so much for the help and info. I really appreciate it.

Since I couldn't get any BF245 FETs, I tried to get some equivalents (although I don't exactly know what parameters are most important for the FET in this circuit). Yves mentioned the J201 in another forum post, and searching the Internet, I also found BS170 and 2N5485 mentioned, but I don't remember where I found those FET equivalents (so they may not be equivalent...I don't know if the source of that information is trustworthy).

So I have the J201, BS170, and 2N5485 FETs. I can try the last two FETs in the circuit soon to see if they will work better than the J201.

[gabbagabi]

sorry if it is a dumb question but,
did you have tried to make C5 (100nF) smaller? like 10nF or 1nF?

like here:
http://www.sdiy.org/ensign/1.jpg

may it worth also try to add a resistor to GND from PIN14 (U2D)?

[john trials]

I soldered in a 2N5485 FET and the sample and hold works beautifully, even with slow inputs varying over large voltage ranges!!!

I did not try the BS170, but hopefully if someone else has trouble getting the correct FET, they can find a 2N5485.

Gabba, thanks for the suggestions. No suggestion is dumb.

Blue Hell, thanks for the tips and motivation to keep trying. I am isolated here and I don't know any electronics guys locally, so having guys like you help me is fantastic.

I recently fixed a flaw in my YuSynth Comparator, too (using an op amp as a comparator creates trouble, but adding positive feedback introduces a bit of hysteresis, and fixed the problem). It'd mess up at slow speeds. I am so happy to have both of those problem modules fixed!

[blue hell]

great you got it working