Hey guys, so after playing with a handful of designs I’ve noticed some differences here and there when it comes to how people design their decoupling and whatnot. I figure a picture is better than words:
Some designs call for 10Rs, some call for ferrite bead, some say either. Others nothing at all.
Schematics don’t really make it clear where abouts some of these components should be—i have learned a few things (placement of the IC decouples thanks @fredrik) but still have some questions 
Thanks
for the 100nf of IC (1&2) it would be better to put them apart, because there they go from +12 to GND and from -12 to GND (which besides are often added to the electrolytic cap of 10uf)
but for IC they go from pin + of IC to GND and from pin - from IC also to GND
I don’t know if I’m clear 
Different schools:
Bypass capacitors should be close to the power pins for the respective chips. Note that the bypass caps in your schematics are too tiny for most circuits; 0.1 uF (100 nF) is the usual size, some argue that with modern components 1 uF multilayer ceramics would make even more sense.
thank you very interesting links ! (mostly : mistakes in diy synth)
You also sometimes see diodes from the rails to ground instead of in series.
Whichever way you draw it they’re from the rail to ground, and drawing them in the schematic near the power supply is commonly done (and less distracting/cluttering/confusing than drawing them near the IC) but they do belong physically close to the IC.
thank you, i didn’t know
it seemed less clear to me like that, since there were no ICs
but in fact it seems logical
The first and most important thing is to use the correct power connector and connect the pins properly.
Because shorting pins 8&9 like you did above, will do no good.
Well, it is less clear, especially if you’re not aware of the conventions, but it’s often more practical that way (*). Ideally you’d have room to place everything where it should be, and draw out as many wires as possible so you don’t have to hunt for cryptic netlabels, but people are lazy
*) in textbooks, bypass caps are often not shown at all, so having them somewhere is a step forward in comparison
Depends on what footprint you use, surely.
I always forget if it’s odd even or whatever always have to fix it down the line. I’m in real early stages in working on this
Yeah, but not using the pretty well standardized connector and having +12V instead of -12V on pin 1 is the best way to test how good your polarity reversal protection circuit really is.
That being said I’d like to draw up a nice “typical” circuit with no errors and post it here
And I’ll have to read that polyfuse thread a couple more times to hammer it home. It’s been a long week lol
Do we have the correct Symbol and Footprint for KiCad listed in the Kosmo Specs Thread?
But yes, as long as the pin designation of the footprint matches the symbol, it does not matter from a construction point of view. I have been tripped up by this on a header, not power though.
Sam often referes to the decoupling caps at the IC’s and leaving them out , same with the ferrite.
So is there any noticeable differenc with/without?
I have not designed anything from 100% scratch so far so gone from other peoples schematics, although where there are Decoupling CAPs I may have put them on the PCB in random places as I keep forgetting about their purpose.
Interesting good information though.
I have seen a different variant for decoupling a tl074, where they put a single 100nf between +12V and -12V instead of the usual two 100nf between the each rail and GND. Is this also a viable option?
This was on some kassutronics project and he seems to know his stuff…
as a engineering prof, what you stated is correct and great for folks thanks
Sounds a bit unusual – usually the output load is relative to ground, and that path is what you want to keep clean. On the other hand, as you say kassutronics stuff is usually carefully designed. Do you remember what circuit it was?
The ladder vcf. The schematic is in this pdf:
Hmm. There’s nothing unusual about the opamp use there what I can see, so that seems weird.
If not else the TL07x datasheet is pretty clear on what they expect:
EDIT: Checked his latest designs, and he’s still using that pattern. I’m not convinced; I think I’ll stick to the more regular way of doing things – it makes more sense if you look at the overall circuit, and if you can only afford one, you should probably stick it on one of the supply lines (which one depends on the opamp internals, but often the negative rail) instead of between two supply lines. But it’s not like a 100 nF ceramic is an expensive component (typically ≤ 1 cent each, also in small volumes)
