I guess there’s enough room to squeeze bypass caps in…
4 Likes
Not being a stripboard person, I did a PCB.
I added an output jack for the clock, an attenuator for the input signal, and a switch (from a post by @dud, thanks!) that allows shorting the capacitor in the clock processing. That changes the circuit from a sample and hold, which grabs the voltage on the rising edge of the clock gate and holds that, to a track and hold, whose output tracks the input as long as the gate is on and holds the final value once the gate turns off. (Really it does the same thing in sample and hold mode, it’s just that the gate is internally shortened to 3 µs.) There are a few other little tweaks but no other significant changes.
You might notice an unusual feature on the circuit board, a rectangle of copper around two of the LF398 pads and one of the capacitor pads. This is a guard ring, as recommended in the LF398 datasheet.
What else to say? It’s straightforward. I built it, plugged it in, it worked. I love it when that happens.
Schematics, KiCad design files, Gerbers, and documentation in the GitHub repo: GitHub - holmesrichards/yash: Sample and Hold synth module in Kosmo format.
9 Likes
Nice, your pcb definitely looks much neater than the stripboard version I built. May I ask why you panel-mount your potentiometers, as opposed to pcb mounting? I find the panel wiring the worst bit of using stripboard.
I dither on that. Most of my last several modules have had panel mounted pots but I’ve done a lot with board mounted. Those are definitely easier to build but harder to design. Jacks and pots have to be on separate boards, and if the rest of the components are on the board with the pots then you have to work around them. You have to use pin headers/sockets to connect the boards, and then you have to route all kinds of stuff from everywhere on the main PCB to the headers. With panel mounted pots you can just run their wires right to where they need to be on the board which keeps layout simpler. Also a single board is cheaper.
If there are a lot of pots and switches I definitely prefer board mounting, but for two or three I tend to figure the wiring’s not that bad. I built a Eurorack sequencer once (and only once!) with eight pots, eight LEDs, four jacks, a 7-position rotary switch, and a toggle switch, all panel mounted; three pots is nothin’!
Or so I tell myself, but after nine Molex connectors in three days it’s already getting old.
5 Likes
That all makes good sense, I agree that arranging components to fit around the hardware does complicate the design process somewhat, and has left some of my previous projects with less than logical pcb layouts. While I have not designed any finished modules in a while, I do have a pcb on order with no hardware on the board, just all the connections routed to pin headers. I am still undecided if I am going to produce additional pcb’s or panel mount. It’s good to know that panel mounting when producing pcb is not such a bad idea. Thanks for your insight. 
2 Likes
Another reason…
I saw that photo earlier, I can not think of a quick fix if these are already on a pcb.
1 Like
The quick fix if the jacks are on a separate board from the main circuit is you get rid of that board and wire the jacks to the main board.
If the jacks and the main circuit are on the same board you’re hosed. Fortunately not the case this time!
2 Likes
If it had been the case, chassis-mounted jack sockets may help, but then the pcb has far less or no support.
1 Like
5 Likes
big physics guy here and math and electronics, so we had physics in the same year possibly high school?
I finally got around to modifying my YASH: 10 nF instead of 1 nF for the integration cap and 4.7 nF instead of 470 pF for the gate width cap. The former change makes the droop rate about 10x slower and the latter makes the integration gate 30 µs instead of 3 µs to compensate for the longer charging time. I can still see a little droop on time scales of about 10 seconds but it’s a big improvement over what it was. I updated the design files and docs at https://gitlab.com/rsholmes/yash.
4 Likes