U6 definitely isn’t the issue which is nice, voltage reference section is working nicely!
2 Likes
Can’t agree more!
I haven’t paid attention when I first looked at your schematic, but I’m not sure if you can drive the next module input directly at the output of the clipping stage. I had issues with VU007/VU007B taking the output of the RGB clipping section directly to the input of the CD4053 used for blanking as it is done in Cadet II: with hard edged signals and/or signals exceeding 0-1V, it would cause some voltage spike under 0V and above 700mV, which wasn’t an issue at the Composite/S-Video output, cause I guess the AD724 could tolerate some negative/positive excursion. However it caused sync issues on the Component output where the blanked RGB signals are summed together directly using op-amps, as the negative voltage spike present in Y was interpreted as sync by the receiving device, making it glitch. The solution was to use an additional buffer between the clipping section and the blanking section (U15.1/2/3)
When you say “Gain sections and clipping seem to be working but I’m getting no output” do you mean you’re able to see those sections working correctly with a scope, but not when you connect the output to the input of another module? It may sounds like the second clipping stage cannot drive the load, though I haven’t tried driving an output at the output of the clipping stage so I could be wrong.
3 Likes
Hi Bastien, sorry I realise I’ve not uploaded the most up to date schematic, I’ve since added an output buffer from the clipping section. I’ll upload the updated schematic when I get in this evening.
In terms of no output I’m just getting 0V on the output of the buffer.
2 Likes
No worries, I had missed Lars comment about adding a buffer at the output at R3/D6 junction, at least you’re sure that it is not what’s causing the issue.
4 Likes
So this is my current clipping and output section. I’ve disconnected R2 from the previous section and connected a signal generator producing a 0-1V sine wave. My test probes are set to 1x. When probing the output I just get +1V DC output. Can anyone see any issue with this section that I’ve missed? Or is it an issue with how I’m testing it? Am I buffering the output incorrectly? Should I be buffering the signal generator before it hits R2?
I’ve isolated the output buffer and it worked fine so can hopefully rule that out?
1 Like
Looks good to me, simulates correctly too.
2 Likes
tips:
check your pcb layout file for errors (if you haven’t done so already)
replace IC’s , maybe you got a dud
look where the signal is present and where is is lost. It is possible that there is a solder issue somewhere
2 Likes
Great advice. Went through every connection to check the PCB, checked the chips then finally discovered R1 was 100R rather than 1k! Swapped out and things are working on the clipping and output sections.
Pictured is some nice asymmetrical clipping!
In my tests I also found I’d got the inputs on one of the op amps the wrong way round in the attenuverter section so next step is to see if I can do a board mod! We’re getting there, albeit slowly 
12 Likes
+1
Best practice is a continuous ground plane layer adjacent every signal layer if you have any say in it. The signal doesn’t flow on the traces, it’s a field - and how obnoxious it is depends on your layout. If there’s a ground plane directly below or above your signal trace the field stays tight right under your trace, the return path is in the most efficient route possible. If you don’t leave that gnd space available the field blooms & causes EMI, crosstalk all that nonsense as the return makes a wider loop & rams into everything trying to make it’s way back home.
It’s pretty cool, our universe has autorouting on for return paths by default - just pour a plane & it’ll sort itself out in the most efficient route possible!
I just noticed that if you can swing it you might try to keep it at or under 100mm tall, seems like you can duck the $35 “Engineering Fee” on JLCPCB that way. Think the cutoff is like 102mm or something. Paying that fee isn’t so bad if you’re doing a run but for prototyping OOF!
Awesome work, fun following along 
2 Likes
Soft Key Version 2 coming very soon! Will do a final check of everything and then get things ordered this weekend. Once the second version is tested and hopefully working I’ll be ready to do a batch of orders if people are interested.
11 Likes
Three weeks later, just ordered the V2 boards!
7 Likes
Another update, managed to put the wrong gerbers in when I ordered v2, so ended up with another batch of v1… either way I’ve reordered v2 boards today so fingers crossed I should be in a position to offer boards for sale in a month or so!
I’ve been playing about with values on the V1, I’ve swapped out the linear 10k pot on the gain control for a logarithmic 10k and much prefer the level of control it gives over shape.
5 Likes