Rust Bucket Anti-alias filters and mods

[lars]

Just looking over the Rustbucket schematic, there doesn't appear to be any anti-alias filters for the MN3007. IC4 looks like an op amp gain stage, and then there is just C7 as a decoupling cap. I believe the MN3007 is clocked for 13ms of delay in this circuit, which would mean the clock frequency is 39.4kHz. That means there should be an anti-alias filter set for 19.7kHz.

The whole reason I bring this up is that when you put an audio probe on pin 7 or 8, there is a lot of background noise, especially in the high-frequencies. Also, it's difficult to not get some distortion on picked notes, even with the BBD biased correctly and the edge switch off. I think adding a proper anti-alias filter would clean a lot of this up, which could also improve the tracking.

Please let me know if I've way off base on this (my understanding of BBDs is pretty limited).

[Scruffie]

You're right there isn't much filtering, there is a filter formed by R18 & C5 which is set low but only a single pole filter.

It doesn't necessarily sound like anti-alias filters are your issue though, the lack of them causes noise and artifacts but not really distortion as such... but it depends on the exact nature of the distortion you're hearing.

You get noise at pin 7 & 8 of the BBD but do you still get it at pin 6 of IC12? That's the wet audio so if the compression is taking care of the noise... but if you have high output pickups, there is a fair bit of gain, maybe making R24 - 68k and then just bringing the volume back up on the gain trim would do the job. Of course dropping the BBD input gain will raise the noise floor some so then maybe a little of low passing on the output would help... you could put a cap in the feedback path of IC12 but gain trim setting will alter the cut off frequency.

Bit of a balancing act and dependent on the exact nature of your issue.

[lars]

Thanks for the ideas.
I think the best first option would be to put an audio probe on each step of the circuit before the BBD, and see if I'm getting an unusual amount of noise from one of the op amps. I do find it odd that the Harmonics section is before the BBD. It would seem the last thing you would want to do to a BBD is add distortion and/or "enhance" the frequency response that it already has a tough time passing. If only there was a schematic for the Pigtronix Attack Sustain for comparison.

I did find that swapping out op amps is pointless. The Attack Decay circuit is very specific to the op amps called for. Putting in anything else just adds noise; NE5532's are worse.

[Scruffie]

Seems like a sensible idea, the 2 x 4558s & 741 are where your audio path is so I wouldn't worry about the other stuff, but the 741 could definitely be changed to a lower noise opamp.

5532s are quite current hungry, possible the charge pump just can't supply enough to run them in such a big circuit. The 1458s are doing envelopey jobs so they're best left as they are anyway. So perhaps give TL072s a go in place of the 4558s and a TL071 for the 741?

The harmonics section has its input signal cut by R19 & 20 and then the boosted signal is limited by the clipping diodes and there's no frequency emphasis going on (in fact there's a little bass cut from C6 and bass is where BBDs struggle with distortion) so it's really not so different to the clipping you'd see in a flangers input stage like on the collosalus. Also if you think about S/N, you want your gain up front, if the harmonics section does add noise from its gain (which, invariably it will a bit) you wouldn't want that post BBD.

Edit: Another thought, while IC2 is in the feedback loop so shouldn't be such an issue, audio does go through IC11 and while I haven't personally experienced it, I have heard people complain of noisier CA3080s... take that as you will.

[lars]

Ah, I could try swapping the two CA3080s around and see if I hear any difference. That's a good idea with the 741. I'm pretty sure I have some TL071s around to test out.

As far as the anti-alias filters, it wasn't worth the effort. I was able to test out some configurations, but the filters made no noticeable improvement in the sound, although the tracking did seem to be a little better. Still, not worth it to make a daughter board and cram it in an already busy enclosure.

*update*
I tried several different op-amps in the 4558 positions (TL072, NE5532, LM833) and also an LF351 in the 741 position. These made little to no difference in noise. But, I put an audio probe to various spots in the signal path, and oh man. Coming off pin 7 of IC1 is just noise city. It sounds worse than a vintage triangle muff at idle. And the problem is, that noise just keeps getting amplified by IC4.
At pin 1 off IC4 the noise floor is ridiculous.
On pin 7 off of IC4 the noise is outright unbearable. With the harmonics pot at 0, pin 7 of IC4 just has a continuous drone that sounds like an audio test tone. Funny thing is, my pedal seems to work and sound like all the demos I watched. I think this is just how the circuit is!
Anyway, just for kicks I put a 120pf capacitor across pins 6 & 7 of IC1, and it was a massive difference in the noise floor all the way through the circuit. I did have to adjust the sensitivity a lot higher to get it to trigger, but it was a huge improvement.
To me the gain arrangement between pins 6 & 7 of IC1 looks kind of strange (I don't claim to understand it, it just looks different than usual. Is it set for frequency dependent gain?). Anyway, that definitely seems to be the culprit in the circuit for the excessive noise. I'm going to try pulling the legs of R6 and R7 so I can run some jumpers to a breadboard and try some different options for that part of the circuit.

[Scruffie]

I'm not surprised Pin 7 of IC1 is noisy, if you look at the circuit, it works by using a sort of DIY compander, so that stage with the CA3080 in the feedback loop is a compressor and compressors are inherently noisy, then the expander section (IC11) is open and closed by the attack/decay circuitry.

It seems to me that you'd be better off putting a feedback cap in IC4B... in fact if I was going to try and reduce noise my self i'd put a 22pF in parallel with R3 (don't want anything too large as this feeds the dry signal too, that should roll off the very high end hiss though) a 100-150pF in parallel with R25 to roll of some of the high end hiss going in to the BBD and then up to 1nF in the feedback loop of IC12.

[lars]

Thanks for all your help with this Scruffie; your ideas have pointed me in the right direction towards getting the noise in this thing tamed.
I found out several things when I disconnected R6 and R7 and ran jumpers to a breadboard to try some mods. I found the a small cap across pins 6 & 7 of IC1 is essential. 100 - 120pf works best.
R6 and R7 really should be around 47k instead of 470k. The two 470k's provide so much gain to that stage, it pretty much drives IC4 into distortion no matter what. With 47k resistors in that position, the sensitivity does have to be adjusted up more, but that's a good thing. The sensitivity control is really just a volume control for the first gain stage of IC1, which is your cleanest sound. It's way better to have that louder than bumping up the volume of subsequent stages.
C3 is very interesting. This seems to control how abruptly the Decay feature shuts off. Too small a value here and notes hardly trigger at all, but too large a value and you add a lot more distortion in the bottom end and the decay kind of fizzles off (like a microsynth). I found that a 150n cap works really well for C3. It cleans up your lower notes, but also provides a more abrupt and clean cut off to the decay, which sounds more like tape reverse. This is of course in conjunction with the other mods.

Overall I'm much happier with the performance of this circuit with these mods. It actually feels like something that could be used, rather than just a novelty effect.

[Scruffie]

Fantastic! Glad you got it sorted, I was worried fiddling with the compressor section would alter the attack/decay response portion too much as it feeds in to that too (and I can't admit to completely understanding everything in the circuit) but it seems my worries were unfounded.

I remember reading Howard Davis saying all these vintage EHX circuits were designed with the low output single coil pickups of the day in mind so it makes sense input gain needed dropping, although i'm surprised you had to go all the way down to 47k!

Although increasing the sensitivity improves S/N, I do wonder if any residual noise could be helped with the cap suggestions I made, there really is very little filtering (I suspect as the circuit was so big already any added caps were considered over the top and an unnecessary expense haha).

Consider C3 as an 'emphasis' capacitor so the larger it is, the more the full signal is passed, so if you have too much bass you get distortion from the OTAs and BBDs which both have limited headroom, but if it's too small and it's all treble, there's not a full enough signal to 'activate' the detection circuitry, socketing it for the users specific pickups/playing style etc. etc. might be worthwhile.

Once again, well done on your investigation.

[lincolnic]

Just wanted to say I've been following this discussion with great interest, as I felt my Rustbucket was also a bit noisier than it ought to be -- I'll definitely be trying these mods out when I get the chance. Big thanks to lars and Scruffie!

[pogart]

I think I will try these mods as well, I was having some trouble getting settings I liked or could experiment with. Thanks for taking the time to post your results.

[lars]

I will definitely second the idea of socketing these changes, in case you don't like them. Or better yet, just lift the two connections from R6 and R7 and run some jumpers to a breadboard. Then you can really go to town and try many different configurations. I actually used the original EHX 16 second delay schematic as a starting point for comparison of the compressor configuration.

My guitar has lace sensor pickups, so the bridge and neck are fairly hot, 12k +. So that is something to consider when adjusting the gain of IC1b. If you run standard strat style pickups you might see better results only halving the gain with some 220k's or so.

I found another modification that I feel is a "must do". Replace the 100k Bias trimmer with a "custom" trimmer:  Take a 20K trimmer and solder 39k resistors to legs 1 and 3 respectively. This gives you a trimmer that the circuit will "see" as 100k, but gives you much greater control over the sweet spot. I've been able to eliminate a lot of distortion in my circuit with just this mod. In conjunction with the other mods, it gives me the ability to crank the gain trimmer higher. It's quite the balancing act between the two trimmers, but when you get it just right, you get a very nice loud signal with much less distortion.

[lars]

One other idea I have, i believe the harmonics section of the circuit is always connected, even with the harmonics pot turned all the way down. I think this might be where some faint background distortion is coming from, much like the distortion from vintage big muffs that aren't true bypass.
Maybe the edge switch could be changed to instead bypass the harmonics section all together, from C5 to pin5 of IC4. The great thing about that, the distortion feature could then be on a stomp switch instead.
Hmmmm....