Help, my S&H never worked right

[synthetic]

So originally it worked but it would click:

http://electro-music.com/forum/topic-30607.html

I just gave up and lived with it. Now it's not working at all. When I plug it into a filter FM, for example, it sounds like a square wave. It sounds almost the same as the clock output, but a lower amplitude. Everything else on the module works except for this.

Thanks for your time.

[yusynth]

Still have no scope ?

Try and change the BF245 (Q5 JFET) check the diode D3.
BTW, was the JFET provided by Bridechamber a BF245C ?
Also C5 must be a good quality cap (polyester film).

Concerning your previous clicking problem you can try changing the value of C4 from 1nF to 4.7nF.

[synthetic]

There's a scope I can use at work if that helps us troubleshoot. I try what you mention here, thank you.

[synthetic]

[synthetic]

What is C4 supposed to be anyway?

Schematic says 10n
Parts list says 1n
Parts layout says 1n
Board says 10n

I installed a 10n, maybe I'll try a 1n. (I don't have a 4.7 right now.)

[yusynth]

C4 is the cap that sets the sampling time, with 1nF is a very short time, with 10nF it's ten times longer. Depending on this value you can or cannot hear clicks. Therefore try different values in between 1n and 10n until you obtain something that you find satisfactory. I obtained nice results both with 1n or 4.7n .

[synthetic]

I've tried a few different values for that cap. My problem is not clicking, it's that the S&H is not holding. I get random peaks out of the S&H output from the white noise normalled source, but they don't hold. Check out the sound file below. I expect a stepped "random" waveform but instead get this.

http://www.jefflaity.com/music/MP3/yusynth_sh_problem.mp3

[yusynth]

Have connected a jumper to set the LINEAR vs LOG mode (preferably LOG mode)?

If you did it then :
Can you post oscillograms (and give the settings of the scope : V/square and ms/square) of what you get at the white noise output; what you get at the G pin of the BF245C and what you get at the output of the sample and hold.

Here are the possible causes for a holding fault :
- too high white noise signal (>8Vpp)
- a dead BF245C (Q5)
- a dead TL072 (U2)
- a dead diode (D3)

[synthetic]

The jumper is installed. I have replaced all of those parts except for the transistor which is on order. I noticed last night that I used a TLC071 instead of a TL071. So I ordered that part as well to replace it (again).

Is it possible to measure the white noise with a DVM? Are any of the iPad oscilloscope apps useful? If not I can bring it to work, although getting power is always a challenge.

[Benjamin AM]

[Mongo1]

One other thing I'll bring up.
Last night I had to replace a polystyrene film capacitor in my MFOS Keyboard controller. It was part of the sample and hold on that board, and it suddenly decided to stop holding.

Thanks to some kind folks here, I found out that those caps are very susceptible to heat, and can be easily damaged when you solder them in. I replaced the cap, but attached alligator clips between the body of the cap and the solder joint. The clips act as heat sinks and help protect the cap.
Sure enough, that fixed the problem.

If your S&H isn't working properly, you may be experiencing the same problem. And for what it's worth, the board did work for a while before failing.

Just a thought
Gary

[synthetic]

Can you tell me the exact part you used (Mouser part number or whatever) for the polystyrene cap? Thanks.

[Mongo1]

[synthetic]

Well, replacing the op amp and BC245 didn't change anything, so I guess I'll replace the cap. For like the fourth time. WTF.

The project page doesn't specify C4 as a polystyrene cap. All it says is "1n (any value between 1n and 10n will do)." I'm just using a ceramic 10p disc cap for this.

[synthetic]

I switched C4 for a 1n polystyrene cap, no change.

[Mongo1]

C5 is the sampling cap - that's the one I'm talking about.

C4 is the one that may have something to do with the original click problem

Gary

[synthetic]

I replaced C5, no change.

[Mongo1]

Hmmm, I'm out of ideas for the moment. I've got one of those boards on order, so I may be of more help when I get that built. At least we could do some comparisons then.

Maybe Yves can be of more help...

Gary

[synthetic]

Thanks for your suggestions so far.

[Mongo1]

There's a resistor on the schematic marked with ** - I show it as R27 (470K). I think if you lowered that value it would drop the noise level. As an experiment you could tack another large value (470K-1M) across that, and see if it helps. I think if that value gets too small you'll reduce the signal too much, and have a different set of problems.

Gary

[yusynth]

With a DVM use the AC voltmeter function and mesure the output level of the white noise if you obtain something greater than 6V you'll have to reduce the level of the white noise for this do as suggested by Mongo1. This is done by reducing the value of R34 (try 330K or 270K).

[synthetic]

Finally seems to be working. I used a DVM and it looked fine, but I found a friend with an oscillator and he measured the noise at +10V to -10V. So we changed that level and it seems to be working now. Still clicking with every pulse but I can try something new in C4 for that. Thanks.

(For interests' sake I was using a TL074 for the noise as listed in the parts list, not the LM324 listed in the schematic.)

[iopop3]

Finished up my random module yesterday. Had similar problems with the s&h, only saw spikes on the output. After reading this thread, I removed the 470k resistors and added a 500k pot to attenuate the noise. Ended up around 35k, luckily I had some 36k resistors lying around. Now everything works alright.