skrasms Posted October 17, 2008 Report Share Posted October 17, 2008 I've been doing filter experiments and found that more resonance is possible from the SID by using *different* values for C1 and C2. A lot more, actually. Does anyone know how valid the schematic is that's listed here:http://www.bel.fi/~alankila/c64-sw/I'm using a 6582. I worked out the circuit equations from that schematic and they're simulating as I would expect in MATLAB. I just want to clarify everything I can before I get too deep into possibly wasted calculations. Apologies if this is something that's been covered a thousand times before, but I did some searches and didn't find anything. Quote Link to comment Share on other sites More sharing options...
Goblinz Posted October 17, 2008 Report Share Posted October 17, 2008 Can't help I'm afraid but I'd appreciate it if you could post some mp3s showing the difference.Will this not affect the cut off frequency as well? Maybe some sort of filter switch would be a good idea? Quote Link to comment Share on other sites More sharing options...
skrasms Posted October 17, 2008 Author Report Share Posted October 17, 2008 It looks like the cutoff depends on the geometric mean of C1 and C2, or sqrt(C1*C2). So if C1 is halved then C2 must be doubled to get the same cutoff frequency. I don't have a wide range of poly film cap values on hand, only about 5. I made LP sweep files to show the biggest difference I could get. The FC register values were altered to get the same sweep range in both cases.Case 1: C1=C2=22nF, RES=15http://www.skrasoft.com/blog/blogfiles/lores.wavCase 2: C1=1nF, C2=47nF, RES=15http://www.skrasoft.com/blog/blogfiles/hires.wavThere are some issues that I am still thinking about, though. Increasing the resonance this way should make the band-pass response peak stronger, but it actually dies off on my SID... the band-pass becomes a no-pass. Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 18, 2008 Report Share Posted October 18, 2008 That schematic is close enough I guess....I've gotta get around to replying to that other thread.... This'll work (clearly - the wavs sound funky), at the expense of the cutoff frequency slope.So instead of like this: /\---' \ \ \you'll get like this: ,'---' | \ '., '(or does it go the other way? Depends which cap is increased and decreased? I don't know, it's too early ;D)Good for some music, bad for sound design. Messes with your harmonics.The bandpass thing, I'm guessing, is the opamps... think of it, rather than one big 2P filter, as a chain of 2 separate 1P filters. Quote Link to comment Share on other sites More sharing options...
skrasms Posted October 18, 2008 Author Report Share Posted October 18, 2008 That schematic is close enough I guess....I've gotta get around to replying to that other thread.... This'll work (clearly - the wavs sound funky), at the expense of the cutoff frequency slope.Good for some music, bad for sound design. Messes with your harmonics.The slope does not change according to the simulation. That's why I am asking about how valid that schematic is. I haven't run "official" measurements on the SID and hoped someone else had already done enough testing to confirm. According to that schematic, changing the resonance using only feedback from the first integrator block (what the RES register does) follows this pattern:http://www.skrasoft.com/blog/blogfiles/chgfeed.png...and changing the resonance using only changes in the external capacitor values follows this pattern:http://www.skrasoft.com/blog/blogfiles/capchange.pngIt just looks like the same standard resonance both ways, with no effect on the filter order (slope).The bandpass thing, I'm guessing, is the opamps... think of it, rather than one big 2P filter, as a chain of 2 separate 1P filters.Does this mean that you disagree with that schematic form? Each opamp section in that circuit is either an integrator or amplifier. Treating it as 2 separate 1P filters cascaded would make it an unstable low-pass filter, unless I am misunderstanding you. It is the feedback loops that make it interesting as a filter and allow different modes (HP, BP, LP). Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 19, 2008 Report Share Posted October 19, 2008 no effect on the filter order (slope).I dunno, it looks pretty different in those pics? Quote Link to comment Share on other sites More sharing options...
H2O Posted October 19, 2008 Report Share Posted October 19, 2008 I dunno, it looks pretty different in those pics?Try the same scales (see attached).Sounds like some nice sounds you've found there Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 19, 2008 Report Share Posted October 19, 2008 Try the same scales (see attached).Yeh I thought that might be it, but there is a difference in the slope even in your image... Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 19, 2008 Report Share Posted October 19, 2008 Well I was looking at the graphs, and the big difference in cap values, and I'm figuring that tiny difference is some glitch... so I went back to the schem to reconsider it and refresh my memory... It would appear that the two caps are actually controlling a separate filter mode each. That explains the 'no-pass' bandpass, and why the lowpass seems to be behaving similarly despite the big difference in values, and why they're not acting like a 'chain'.... Might pay to compare the curves between HP and LP... Quote Link to comment Share on other sites More sharing options...
skrasms Posted October 19, 2008 Author Report Share Posted October 19, 2008 Well I was looking at the graphs, and the big difference in cap values, and I'm figuring that tiny difference is some glitch... so I went back to the schem to reconsider it and refresh my memory... It would appear that the two caps are actually controlling a separate filter mode each. That explains the 'no-pass' bandpass, and why the lowpass seems to be behaving similarly despite the big difference in values, and why they're not acting like a 'chain'.... Might pay to compare the curves between HP and LP...Which tiny difference? The maximum resonance difference between changing cap values and changing the RES register that in those graphs is not accurate. Those are just a couple simulations I ran with arbitrary values to show that both cap ratios and the RES register control resonance in the same manner. With the right cap values simulated I could make them line up perfectly, but that wasn't what I was trying to show. I don't think we're talking about the same thing when we say "slope." I'm talking about the db/oct of the filter. That does not change at all in the pictures I posted. The transfer function gets a peak at the cutoff frequency, but that is normal resonance. You mentioned messing with harmonics that was bad for sound design, but unless you consider resonance as a general concept to be bad then I'm not sure what you meant with regards to changing cap values. What explains the no-pass? I mentioned before that it isn't just a couple filters cascaded in series. There is feedback involved so changing one value anywhere has an effect everywhere. Did you work out the transfer functions from the circuit? I have them all on paper and in MATLAB, and they do not have the no-pass effect. It only exists in real testing with my 6582, not at all in the schematic (is there confusion here?). If you derive equations from that circuit that do have it, it would be helpful to compare yours with mine. Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 19, 2008 Report Share Posted October 19, 2008 Which tiny difference?The really small one :DI don't think we're talking about the same thing when we say "slope." I'm talking about the db/oct of the filter.Yep same here. That does not change at all in the pictures I posted.It's out by two pixels but as I said, I think that's a glitch.It only exists in real testing with my 6582, not at all in the schematic (is there confusion here?). That schematic is close enough I guess....I guess not. Quote Link to comment Share on other sites More sharing options...
H2O Posted October 20, 2008 Report Share Posted October 20, 2008 It's out by two pixels but as I said, I think that's a glitch.Any two-pixel variation is likely to be my fault ;) Quote Link to comment Share on other sites More sharing options...
skrasms Posted October 20, 2008 Author Report Share Posted October 20, 2008 Score! I traced some names involved with the 6581 distortion page and came across many pictures of the chip die filter section up close. What's even better, though, is this:http://oms.wmhost.com/misc/sid_filter_plan.gif Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 21, 2008 Report Share Posted October 21, 2008 Score! Nice one skrasms! I knew there'd be a "real" schem out there somewhere. That's even better than the one I knew from before.Can't wait to see what you're going to hack now :D Keep us posted dude! Quote Link to comment Share on other sites More sharing options...
skrasms Posted October 23, 2008 Author Report Share Posted October 23, 2008 I'll let you know what I come up with :)For the next few weeks I'll be busy starting a new job! Quote Link to comment Share on other sites More sharing options...
stryd_one Posted October 23, 2008 Report Share Posted October 23, 2008 Good luck! Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.