White Noise + Sample & Hold module

here’s mine Kosmo format of a S&H from René Schmitz
(with LF398 chip)

“YASH” S&H :

sample and hold

White Noise :

link : https://www.schmitzbits.de/sah.html


Ha, I discovered this post after replying to @devicex in the Mail Day Thread

@Dud Yep in essence I build the same thing (as I wrote here earlier:
Mail Day Thread ) You may want to have a look at the way I connected the capacitors to the LF, I used IC sockets, so that I can easily experiment with other types.

Here is an updated version of the schematic:

and here is a picture of my implementation:

As you can see in the schematic I left out the oscillator from the original schematic so my version depends on an outside source for triggering. Initially I had also not included the gating bit (the 4093 logic) to the circuit, I thought I could do this by using an external trigger, but the LF398 needs a very short pulse (a few micro seconds) which I had no source for, so I added this eventually which is why it is on a separate strip board.

So far I’ve used it with rather frequent triggers, and not noticed any drop in the output voltage between those which could affect the circuit that follows the S&H. I tried it with an oscillator frequency for a while given that the human ear is very sensitive to pitch changes and could not detect any whining / sinking of the pitch, so to speak.

Your and probably my mileage may vary with long periods between the pulses. So the IC socket approach may turn out to come in handy when the need arises to change the capacitors / experiment with other type capacitors.

Because I wanted to see when the output voltage is either positive or negative I’ve added 2 transistors that will each light an LED. Of course this is a rather course way of indicating the output voltage but it is a FUN way !

Luckily nowadays it is not so much a problem than it used to be to talk about these matters, but I just like lots and lots of flashing lights in my modules!

The attenuator potmeter drawn in the schematic near the Vout leads I ended up not adding to the module, I decided to make a separate general purpose attenuator module for that.

Best practice with CMOS chips: if you are going to build this circuit, do not forget to give the inputs of the 4093 gates that are not in use a fixed value, e.g. ground them. Otherwise the unused ports can lead to oscillations which may affect this or other circuits.


while going over my notes I remembered this:

I had to add a 1k resistor to the output of the LF398, otherwise I could not use it with some of my synths. Without it the S&H seemed to loose its output voltage very quickly, which should not depend on its load and which is odd as it is supposed to have a low output resistance. And although I do not understand what was happening here, a 1kOhm resistor resolved the issue.


What devices did you connect it to? As you say, it has a very low output impedance, but the specifications use a 10k load and the short circuit currents are only a few milliamps at room temperature, so the source/sink capacity is probably rather limited. Still sounds a bit strange that adding 1k would save the day, unless the thing you connected it to had a very low input impedance…

I connected it to a few inputs of a Behringer Neutron, Model D and a Crave. With some the response was as I had imagined, with some it was as described earlier. I’m afraid I didn’t write down which inputs were problematic. The 1k resistor resolved the matter.

I will keep an eye on its behaviour as I start using it.

Edit: I remember that at one point I started doubting that the input of the synth was in working order, but I got it to work without a glitch with a signal from another CV source. But as I said, I’ll keep an eye on it.


Yes, Discourse, I know this is an old topic! But see, I wanted to ask @Dud about it because I’m considering building this S&H. I was going to ask him if he could recall why he changed the value of the integrating capacitor. And exactly how his track and hold switch was connected. Okay, Discourse? Can I do that?

Also, if I’m not completely confused, the 4093 is used only with both inputs tied together, so that makes it just a Schmitt inverter, right? So could a 40106 be used instead? Or is there something extra special about doing it with the 4093?


Yes it’s old subject, one of my 1st module, now i don’t always know what i’m doing, so …

I had seen information on a forum about the track & hold, the switch is mounted to cut the cap of 470pf.

for the 2.2nf instead of the 1nf ? surely on the same forum, i don’t know.

but I have to come back to it, it never really worked well, it seems that all the pulses (trigger) are not counted.

the value of the cap? Did I use a polystyren cap? polypropylen ?

sorry for such poor information, but @Jos built it too


There is one way to find out: don’t be afraid to try it!


Found this:

DGTom: if you put a SPST ON-OFF switch across the 470 pF cap of Rene’s YASH, you can switch between S&H and T&H modes.

Nothing there about 2.2 nF though.

And then this:

But I’m getting extreme droop in the output voltage no matter what caps I’ve used (many different 1nf caps as well as various other sizes). The best I’ve got was from a 2.2nf cap but that had pretty solid hold, but only gave me about 1-2 volts output even using a 10vpp lfo as the sample source.

but that’s not exactly a ringing endorsement of 2.2 nF.

1 Like

yes, i think the better is to follow the René Schmitz original schem, or try …

btw someone would have a source to buy a 1nf polystiren capacitor plz ?

On René’s site it says

Polystyrene caps are now difficult to source, so a polypropylene (PP), polycarbonate (PC) or polyphenylene sulfide (PPS) cap can be used instead.

and coincidentally I just got some 1 nF PPs from DigiKey a couple days ago.

1 Like

i’ve just find this one


1 Like

Alright, since Rich already resurrected this thread, let me add my frustration :see_no_evil:

I built the sample and hold circuit on stripboard from Eddy Bergman following the (YASH) design from Rene Schmitz but I could not get any sampling out of the LF398. Eddy added a DC offset pot and in a very specific position I get some square oscillations between max and min voltage but that’s it.

I checked all the connections but I could not find any error. The LED lights up nicely, the rate pot is working well but when I follow the signal path of the trigger, it vanishes after the second Schmitt trigger, right after the 470pF cap, see at the top of the schematic:

Screenshot 2021-12-15 at 01.05.55

I decided to rebuild the original circuit on the breadboard to understand what’s happening and I started with the trigger signal taming. Whatever I try, I cannot get anything other than 0V after the 470pF capacitor:

I measured the 470pF and it’s actually 400pF (all of my 470pF caps are more like 400pF)… This low pass filter has a cut-off frequency of 33kHz and with 400pF I calculate 40kHz so I guess it should be fine (I still tried adding a 75pF cap in parallel but no luck). I simply cannot see the short 3us pulses.

Then I thought my oscilliscope or probes are shitty so I fired up my frequency generator to see if I can resolve a 100kHz square wave and it works nicely.

I am clueless. If the trigger taming to 3us does not work, the whole circuit will of course not work. Nevertheless, I rebuilt it on the breadboard and as expected, it does not work:

(here, pin1 of the LF398 was not connected to +12V, but I fixed it)

Anyways, I am stuck, so if anyone has ideas, please let me know… I tried different LF398 and also CD4093BE but I get always the same, 0V after the 470pF and nothing at the output of LF398.

I have the feeling that this circuit is not so well suited for breadboard or stripboard due to the parasitic capacitances of the rails, but I might be wrong. At least I see quite a few failed attempts all around the web :confused: Maybe it’s time to make another PCB

EDIT: it’s too late already but tomorrow I’ll revisit the 398-part and trigger it with 3us pulses from my function generator to see if that works. I really suspect that the 3us signal gets buried to ground on the strip/breadboard.

EDIT2: Just realised that although I am not able to display the 3us pulse signal (or any part of it right after the 470pF cap), but the trigger-LED on my oscilloscope is blinking with the oscillator frequency. This means that the signal is still somewhere, so that the oscilloscope can sense it but probably not enough (or has not the right shape) to trigger the LF398

Does the scope trigger when you probe after the subsequent 4093 stages? What about at the LF398 pin?

Presumably the pulse is either too narrow, too slow (rise is supposed to be > 1 V/µs), or too small. A larger cap would make the pulse wider. Changing the voltage divider resistors after the last 4093 stage would make it higher. Not sure what to do about too slow, other than make the pulse higher to compensate.

Nothing magic about 3 µs (Schmitz says that was for fast acquisition), you can bump it upward if you need to. In fact as @Dud’s schematic shows, you can short the capacitor and use it in a track and hold mode.

I can’t see anything in the datasheet about a minimum pulse width, maybe I’m just overlooking it. It does say “acquisition time is as low as 6 µs to 0.01%”; does that mean 3 µs is unreasonably low? But I think it just means it won’t give you 0.01% at that timing. Threshold is typically 1.4 V (maximum 2.4 V).

My understanding is parasitic capacitance from breadboard contacts is of order 10 pF, which I think is too small to be a likely culprit. But I could easily be completely wrong about that.


What levels do you see going into the 2nd Schmitt trigger - right after the transistor/led combo?

The ancient copy of the datasheet that I have seems to imply that your trigger will need to be around 7V when running it from 12V supply:

If you disconnect the output of that Schmitt trigger from the 470pF cap, can you see it toggling?
Can you try forcing to to toggle manually, just to make sure it’s doing what you think it should?


1 Like

Thanks for both of you, I’ll check your suggestions and also the levels again in the evening after work!

Here is an oscillograph I recorded in the night, which shows the trigger gates (before the 470pF cap), the triangle I generate and connected to the input and the straight line at the bottom is the output of the 398, sitting at -12V.

So far I can tell @jaradical, the levels of the trigger signal are always maxed at 12V. I see it after the first stage and the second stage, just before the 470pF and after that I see 0V (but the trigger LED of my oscilloscope still blinks with that rate!). The LED is not connected on the breadboard, I left the entire BC548 part out on the breadboard. Maybe that’s my mistake? I however don’t see any reason why it would be needed. The 4148 only protects from negative voltages and the transistor is there to drive the LED, so the input to the second 4093 stage is at around 12V anyways (coming from the first stage).

Maybe it’s too late/early :see_no_evil:

After thinking more about it, I think that I might have a problem with the 50Ohm input impedance, which maybe kills the signal from the 4093. I’ll try to add a fast opamp to see if I can catch the signal after the 470pF cap since my oscilloscope does not have a 1MOhm input impedance switch.

So this might explain why the trigger LED is lighting up with the signal’s periodicity but the rest of the oscilloscope electronics fails to bring the signal to the display.

This also means that the clock part might be OK, so I have to search for the error on the LF398 side :wink:


Really, there’s such a thing as a scope with only 50Ω inputs? I’m surprised. I went looking and found a scope that looks similar to (not the same as) what can be seen of yours

but its inputs are marked 1MΩ.


Yes you are right, that’s my scope and for some reason I thought it has only 50Ω inputs. Oh dear :sweat_smile: OK, so still hanging in the air. Let’s see what I find out in a few hours when I am at home :wink:


Alright, I can definitely see the 3us dip with my mobile oscilloscope.

I tried different probes on my should and also recalibrated them but no luck. I guess it’s really time to get a solid bench oscilloscope.

Now at least I have the right tool at hand to find the root of the problem :sweat_smile: