I am clueless. I rebuilt the whole circuit, the trigger is working, everything is connected properly (I must be blind otherwise) and the sampling still doesn’t work.
I connected a 1MOhm pot with pin1 and 3 to -12V and +12V and pin2 to the input (pin3) of the LF398 and as you can see, the ugly line in the oscillograph is my manual input voltage from the voltage divider pot but the sampling output (pin4 of the LF398) is 0V constantly.
This is really frustrating 
I tried 7 different LF398s…
I build the thing some time ago following that same schematic, in fact I build a double one, and they have been working from the start without a hitch. See my earlier post:
I remember that the pulse must not be too long, then the LF won’t sample. There is a remark about this in the datasheet I believe.
If you get it to work then here are some implementation ideas:
You want more than 1 S&H !
The capacitors I mounted in a partial IC-socket so that I could experiment with different types and values without the need for extra soldering.
I installed 2 trigger inputs but combined the trigger ins of the 2 circuits via the switch in the input jack. The second S&H will take the trigger from the 1st unless it gets a trigger signal itself (and v.v.).
Your scope photo shows a negative going pulse, is that after the second 4093 stage (the first after the cap/resistor)? If I’m interpreting the photos correctly the probe is on the third stage. So something looks backwards to me, even though I don’t see any errors on the breadboard. Maybe I’m just misunderstanding something. Added: On a closer look it appears in that picture the third stage isn’t even connected to the second, and is connected to ground. So even though the probe appears to be on the third stage output, presumably the scope trace was from when you were looking at the second stage. So that’s all right.
I notice you omitted the voltage divider on the LF398 trigger input. You’ve tried it with and without? The datasheet seems to indicate you need a pulse height between 3 and 7 V.
If you really do have a large positive going pulse seen at pin 8, the only two other things I can think of are:
Change the pulse width. You should be able to go to 10 µs or higher with no problem.
Are all your LF398 from a single order from a single vendor? Any chance you could have gotten a bad bunch?
@tamasgal
I’ve built this same circuit in both breadboard and stripboard, following the original Schmitz schematic and I can say that it works for sure.
Do you know if your breadboard has continuous feed rails? Some of them are splited in the center e others have up to 3 or 4 divisions and you would need to join them together via jumpers.
like this:
This one, with more divisions
I say because sometimes I forget this detail.
I also use the multimeter to measure continuity all the way through each component terminal and jumper inserted into the board, to make sure there are no contact problems, some holes may have gaps.
If nothing works there may be a problem with your LF398 batch.
I used this one:
I wish you good luck!
Yes thanks, sorry, I probably moved the probe when I was holding the mobile phone to take the picture!
OK, that’s totally true. I forgot to add the voltage divider, holy cow! I was so focussed on the part before and happy that I finally saw the signal… I’ll add it immediately when I am back home and test it, I hope that should be it. If the LF398 is really that picky with the pulse height, that should be it!
Thanks for confirming, that’s motivating 
Yes that’s fine in my case, but thanks for mentioning it. It will definitely help one or the other in future!
Oh yeah, these are very good ideas, thanks! I’ll definitely build multiple of them, once I figured it out 
I am now pretty confident that the missing voltage divider is the culprit.
I added the voltage divider and play around with different values. Lower value resistors have less inductance so I get a sharper edge with e.g. two 1k resistors instead of 5.6k (I also tried the original 1.8k and 5.6k but there the voltage is 3V which is the lower limit of the specs)
I now have a somewhat good looking 3us pulse with a height of 6V but still no sampling.
I also hooked up my function generator but it’s also a bit at its limit with 3us. The pulse does not look much better than the original circuit.
I have two batches of LF398 and now tried like 12 or so from both. So no idea. This is really getting frustrating, I must overlook something stupid or I got scammed with the chips…
Maybe the edge is not sharp enough 
One funny thing I noticed it that pin6 is at -12V, that’s the pin where the 1nF styroflex is going to ground. Is that normal? Also with the oscilloscope, I get a quite stable -12V signal in that pin, I see very tiny dips here and there.
The capacitor stores the sampled voltage, which then is buffered and sent to the output pin:
so both should read the same into a high impedance probe. So the good news is the capacitor’s getting charged. But why to -12 V and then not changing?
I don’t understand either. I guess the chips are all broken, I have no better explanation otherwise. I also rebuilt the circuit on two other breadboards just to be sure it’s not a broken rail or so but I see the exact same voltages and signals.
I also realised that they are probably from the same batch.The first ten I ordered a few months ago but could not find them so I ordered another pack from the same seller. The bar code is the same on all tubes:
I’ll search for another reseller and try again.
I’ve been through this for the same reason so many times, it’s annoying.
To make sure the chips are fake, you can take one and test it with some of the typical application circuits on the datasheet. They are pretty simple.
LF398 - National Semiconductor See Page 8 to 11.
I do this to test any purchased chips or transistors as soon as possible before the return deadline… If they don’t work or are out of specification, I return it to the seller and get my refund. That way I don’t waste my time and don’t get frustrated.
Anyway, these things need to happen to keep our eyes open, very few people still believe that it is possible to earn money with honesty and dignity.
All the best.
Maybe this can help you with the chip test:
Extracted from: Electronics 53 - Sample and Hold
I finally managed to get back to this problem but I think now I can really conclude that the chips are all broken. I rebuilt the test circuit posted by @peternoya and I get similar behaviour: -12V on pin6 and 0V at the output, no matter what I do.
I used 5.6V for the gate signal height (also tried different voltages between 3V and 12V) and 6V as input voltage, as described in the test.
Can anyone recommend a good source for LF398s?
I think I got mine from RS or Mouser - tme.eu look like they have them in stock too.
In addition, TME has the LF398N8 from Analog Devices. U$3.65, but it’s Christmas, isn’t it?
As long as you don’t get another lump of coal!
So it becomes Halloween… 
I got 10 of them via the ‘FSXSEMI FengShengXIn Store’ on Aliexpress in March 2020. I had a look a minute ago and the store appears still to be selling them.
Just know that at $0.15 each, they presumably are not first quality Analog Devices or Texas Instruments chips.
Still, if they work…
Yep, the once I bought ( see picture in an earlier post ) do the job.
I bought these: LF398N Sample + Hold - Verstärker 0,004% DIP8 | eBay for 2 EUR each. Let’s see. That was the fastest I could find, which delivers within 2 days. I cannot wait longer 
