BBD noise elimination

[StephenGiles]

Found it:
https://www.dropbox.com/scl/fi/gnfrj0lj8020ic5rrvwgk/TDA1022_c-1.jpg?rlkey=flm9b3tgcybpgiaedzcqi6p1y&st=o4nkbagc&dl=0


See the little circuit on top right which is connected to the base of each transistor - isn't that a current source, and I think it's there to minimise noise.

[PRR]

The transistors are wired as current sources. This would give a dB less loss in the BBD output followers; which would be cumulative in long chains like the one shown. Yes, S/H gating is a standard tool for jaggy noise. But this all loses the simplicity of re-masked memory chips. Now we have fast dense RAM we should move to real digital audio.

[jorg777]

Found it:
https://www.dropbox.com/scl/fi/gnfrj0lj8020ic5rrvwgk/TDA1022_c-1.jpg?rlkey=flm9b3tgcybpgiaedzcqi6p1y&st=o4nkbagc&dl=0


See the little circuit on top right which is connected to the base of each transistor - isn't that a current source, and I think it's there to minimise noise.

Indeed, and they work surprisingly well in practice.  They do two things which cumulatively work nicely:
(1) The very high impedance load results in higher gain (i.e. very close to a gain of 1 for this case).
(2) Greatly reduce distortion, by forcing Vgs of the BBD output transistor to be constant.  This allows you to pump a much larger signal through without distortion.

[jorg777]

The transistors are wired as current sources. This would give a dB less loss in the BBD output followers; which would be cumulative in long chains like the one shown. Yes, S/H gating is a standard tool for jaggy noise. But this all loses the simplicity of re-masked memory chips. Now we have fast dense RAM we should move to real digital audio.

Yes, and while we're at it, let's chuck all those silly analog pedals in the trash.   
BBDs are not re-masked memory chips.

[Eb7+9]

let's chuck all those silly analog pedals in the trash ...

save a few for posterity's sake

without spilling the beans, one of your suggestions above actually helps lead us towards smoother waters - away, so to speak, from what one might call the Japanese "metallic chorus" era ...
I came up with my own variant a while back - a bit more complex but worth exploring

yep - still lots of untapped art in them stupid BBD's

[ElectricDruid]

I have to say the PIC's NCO module is pretty much ideal as BBD clock generator. You can adjust the dead-end period in terms of "X counts of the system clock" (So 8MHz or 32MHz or something = sub-usec resolution) and you can generate frequencies up to 500KHz with ease. Plus the outputs can drive a 20mA load, so there's enough "oomph" to push the clock pin capacitance of quite a few BBD stages. The only problem is dealing with modulation, since a NCO only has discrete frequency steps.

I'd like to revise this. There's a problem with NCO-based clock generation, which is frequency jitter. The NCO mostly doesn't generate a single frequency, but jumps between two different period lengths. The *average* period length is exactly what you're after, but that's not what you get. The jumps between different periods have a frequency themselves, and depending on the specific frequency increment of the NCO, that can be an audible frequency.
The net effect of this is a weird warbley watery noise in the background when using an NCO to drive a BBD! Since this noise is caused by the clock, efforts to eliminate clock noise from the BBD also help reduce this jitter noise.

Using a timer to perform frequency division is another way to produce a clock from a PIC or other uP. That avoids the frequency jitter problem, but the possible output frequencies get sparser as the frequency gets higher, so it makes modulation basically impossible. It's not a bad solution for uP-controlled tap tempo delay though.