Power configuration for multi-sid

[aminoplacid]

I just got my single SID to work earlier this week after way too long since I began this project.  Naturally, I was thrilled :)

On to my question...

I built the optimized PSU with the intent that I would build the 4x SID setup.  I think it is a bit sloppy looking with the daisy-chained 5volt power going to each of the SID and CORE modules (see the optimized psu pdf).  I was considering redoing my PSU board but instead of daisy-chaining the 5v line to the J2 on each board, I wanted to use a star configuration (I think this is the correct term) to output a single line to each from the psu with the idea that they would be connected along a ribbon cable until they reach their intended PCB.  It seems like this would work except with the SID module, the last board is supposed to end with a 2200uF Cap.  If I used this approach, would I have to connect every SID module to the 2200uF Cap since they wouldn't be chained?

Just trying to minimize the cable clutter in my future case.

Thanks for any advise.

[Wilba]

I think with a star configuration, you wouldn't need the 2200uF cap, this is intended to reduce digital noise when it's a single 5V supply to all SID PCBs. I actually do not use this 2200uF cap in the MB-6582 PCB which has one track for 5V through all the SID modules (like the optimized PSU circuit) and have not noticed digital noise from the rest of the circuit.

Instead, you might benefit from separating the 5V supply to the SID from the 5V supply to the shift registers, and also separating the SID's ground from the ground of the shift registers, i.e. connecting it to an "audio ground" shared with the audio buffer circuits and the ground of the audio inputs/outputs.

[lucem]

Having a star topography power distribution is a definately a good idea.

It reduces not only high frequency noise on the supply lines, and resulting problems from that, it also reduces the series resistance in the power supply to the respective module.

This (comparatively high impedance) is the source for the low-frequency hum that's especially feared in audio applications, usually at 50 or 60 Hz, so exactly the net frequency of the supply net you get from your power provider.

Long power trails, and the resulting higher impedance of the supply lines, especially the ground lines, act like an antenne that catch up the magnetic part of the EM-radiation coming from your power lines, which creates that hum on the audio.

So, I'd go for a star power distribution, and have the 2200µF capacitors reduced to between 470 and 1000 µF, but have those on each core; additional 100nF parallel to them sufficiently block digital noise (electrolytic capacitors go "blind" at high frequencies, so their blind impedance goes up instead of down, in the ner of up to 1 kOhms) and are to be mounted on every module (not only the cores).

If one (in the rare situation) encounters digital noise and interference event with such a setup, a ceramic capacitor of 22pF can help; if money's not an issue, one may think about inductors in the power lines, in the range of maybe 100µH or higher; ferrite beads on long cables might also dampen noise peaks and voltage peaks breaking through the voltage regulator.

If hum and noise build up inside the SID module, one can get rid of if (at least dampen it) by reducing the input impedance of the analog driver at the SID output (thus increasing the load); self-induced noise on resistors etc. builds up squared over the impedance, so even a small impedance reduction can have great results.

Of course, ferrite beads etc. can be applied here, too, as well as a bigger coupling capacitor (always make sure that the plus side of it is "inside" the circuit).