Filter Test inSIDents and a Hardware Hiatus

The past couple days have been mainly about dissecting the SID 6582 datasheet, with an emphasis on the filter section.

“Both capacitors should be the same value” – pg. 7

No reason for this is given. I read some claims online that matching the capacitors provides the most resonance, but wanted to see for myself. My result is the opposite, that using unmatched capacitors can provide much more resonance. The next question is… why?

“These pins are used to connect the two integrating capacitors required by the programmable Filter” – pg. 7

This implies the filter uses integrators, which likely leads to a topology that follows the schematic at Antti Lankila’s 6581 distortion page.

So I worked out the transfer functions from that circuit and tossed them into Octave (the MATLAB alternative for all of us no-longer-students) for simulation. It turns out the resonance is augmented proportionally by the ratio of the cap values. The cutoff frequency depends on the geometric mean of the capacitor values, sqrt(C1*C2).

…so what does more resonance from a SID sound like?

Filter sweep using an internal SID oscillator:
Matched caps set to maximum resonance
Unmatched caps set to maximum resonance

Filter sweep using external audio input with a drum loop:
Matched caps set to maximum resonance
Unmatched caps set to maximum resonance

For those of us using the SID in synthesizers, more resonance is very welcome if it can be manipulated properly. My 6582 is not behaving according to the aforementioned schematic in terms of the bandpass response to unmatched capacitor values. Where the simulation shows more bandpass resonance, the real 6582 is choking into a no-pass filter. Off the top of my head I do not have any guesses as to why. There are a couple other, more expected inconsistencies:

“The approximate maximum Cutoff Frequency of the Filter is given by:
FCmax = 8.2E – 5/C” – pg. 7

This is not true at all according to tests with a real SID, and is most likely based on the assumption:

“Together these [FC Lo/FC Hi] registers form an 11-bit number which linearly controls the Cutoff Frequency of the programmable Filter.” – pg. 5

It is more likely a tanh-type function, saturating in the extreme ranges. I need to do a lot more testing to confirm.

This week I am moving across the country, so as of today I am packing up all my circuit projects to ship ahead of me. It is unlikely that I will be able to resume on them until I am settled in at my new job, so all my projects will slow down for a while. In the meantime I might be able to get some measurement tools coded.

2 Responses to “Filter Test inSIDents and a Hardware Hiatus”

  1. Matthew says:

    The puns strike again!

  2. kris says:

    its already been a while but
    im just wondering if you have any news on this subject, e.g. found any good combination of capacitors to increase the resonance?
    i just received my mb6582 and would love to have a bit more bite on the filters…
    thanks and best regards