while building my second ssm2044 shruthi i asked olivier a couple of questions by email. i thought i’d post my questions and his answers here, so he won’t have to answer them over and over again.
for your edification and entertainment i’m making this into a dialogue - for best dramatic effect please read out loud with different voices for olivier’s and my part. here we go:
q (me): i have a question about ic6. on the pcb, it says tl 084, but in the bom it says tl 074. which is correct?
a (pichenettes): Both are totally fine, no difference.
q: i read on the interwebs that tl 074 is better for ac, and tl 085 better for dc signals.
a: In the case of the SSM2044, the quad op-amp is processing CVs, so it
must have better DC performances; and this is why I spec’ed a TL084 in
the first place.
However, this “widely known” fact that the TL074 is better for AC,
TL084 is better for DC is actually true only if you buy the
"industrial grade" and “A grade” versions of the chip. In the other
application / performance bins the TL074 is always better than the
TL084. This became apparent to me only a few weeks ago.
But then, we’re talking about this on a synth with 8-bits converter,
and with an OTA-based VCF… Don’t worry, whatever noise or distortion
there will be in the circuit will surely not come from a bad choice of
op-amp. Weakest link, etc…
q: i also have a question about those audio grade np capacitators c6, c10, c34, c39: they are spec’ed to be 3.3u. can i use 4.7u caps instead?
a: In this context, the bigger the cap the better. I spec’ed 3.3uF in the
BOM because some 4.7uF caps or larger are a bit obese and wouldn’t fit
on the board ; but [if you find slimmer 4.7u ones, they will work just fine].
q: on my first ssm board, i did use 3.3u ones - will the two shruthis sound any different from each other if i use 3.3u on one and 4.7u on the other?
a: The 3.3uF version will have a low-frequency roll-off point at 0.5 Hz
(frequencies below this point will be seriously attenuated) ; the
4.7uF version will have a roll-off point at 0.3 Hz. Given that your
speakers will probably have a roll-off point at 20 Hz or so, and that
humans don’t perceive low frequencies as sounds, it doesn’t matter and
I dare anybody to perceive a 2dB loss in a frequency they can’t hear,
1/ some people might argue that all these low-frequency loss might add
up through an entire signal chain so it’s better to design with 1 or 2
orders of magnitude of margin below / above the human threshold of
2/ It would be really confusing if the BOM listed a part value as “it
doesn’t matter dude, just make it large enough”.
Almost half of the parts in the filter board circuits could withstand
a +/- 30% change, and if given a synth circuit, you wonder why such
resistor is equal to such value, the most likely answer is that it was
what the designer had on his desk at the time.
q: thank you very much for this most instructive conversation!
oh, incidentally, i found that on the ssm pcb v0.2, there are two sets of holes for different sizes of 6.8n film caps for c21, c22, and c36 - but still not for c11! this is not really a problem since the legs of a smaller 6.8n cap like the ones sold by reichelt can easily be bent to fit the holes for a wider cap. but still, you might want to fix this for aesthetic reasons on a future v0.3 revision.
Shit. I Ordered 3.3µFs and have a huge pile of 4,7µF on Stock…
glad i could be of assistance.