Faint Castor ramp bleed into Pollux mix out
nebularnoise opened this issue · 6 comments
For context: this is a DIY build based on hardware v2 9/26/2020, with fixed transistor ramp core.
I've noticed that with all volume pots at zero, there is a faint bleed of castor's ramp in pollux's mix out.
I've yet to calibrate the unit, but this seems like pure analog spookery at play, here.
Please note that in the following report, part numbering follows latest schematic (3beb279 as of today).
With preliminary probing, here's what I've tested:
- No bleed at pin 8 of U12
- Bleed after r47 (== net pollux_mix_out) -- this is the reference volume
- Bleed is fainter at pin 3 of U14a
- Bleed is equal at pin 1 of U14a
Setting Castor's ramp volume up, and playing with the crossfade knob can cancel out the bleed. This seems to indicate the bleed is 180° out of phase with the ramp output. Maybe netlist castor_ramp_pre is bleeding in ?
I'll tinker a bit further when I get the time.
On production hardware there is a very small (6mV peak-to-peak, about 500 parts per million out of the 12 V peak-to-peak output) noise that can be observed on castor_mix_out, pollux_mix_out, and mix_out. Part of the noise is influenced by Castor's oscillator frequency, but it is not anywhere near audible. From what I can tell, this is normal power supply ripple and parasitic capacitive coupling (castor_ramp_out and pollux_mix_out run parallel to each other, and also are adjacent on the pin header).
Bleed is fainter at pin 3 of U14a
pin 3 of U14a is the output of a voltage divider that halves pollux_mix_out, so that tracks.
Bleed is equal at pin 1 of U14a
pin 1 of U14a is the output of the amplifier and should be louder, so this tracks as well.
This seems to indicate the bleed is 180°
The mixed waveforms are inverted.
I'll tinker a bit further when I get the time.
Good luck! I'm happy to help where I can, but please keep in mind that we can only provide limited support for DIY builds.
we can only provide limited support for DIY builds.
Of course! This issue was only to document what I found in my journey 😄
6mV peak-to-peak
Oh, on my unit, its muuuuch more than that.
This morning, I realised I had not properly cleaned the jackboard. I probed a puddle of flux and the bleed was there 🙃
I'll be back after cleaning everything thoroughly. Most likely the adjacent pin headers are capacitively coupled through the flux goo...
I'm going to close this for now since it doesn't affect production boards, but if there's anything else I can do, let me know.
It was 100% the flux.
The outputs are now perfectly silent when pots are down 🤏 chef's kiss