DIY distortion: Have to clip opamp (i think) to reach unity :(

[MrStab]

Hi,

i don't have an up-to-date schematic drawn out, and haven't managed to test the exact source of the overload, so please bear with me:

here's my setup: opamp input buffer, clipping LEDs in an opamp feedback loop (or after opamp, tried both), james tone stack (with grounds connected to vbias) and standard non-inverting opamp recovery stage (biased by tone stack, no cap between), volume pot, opamp output buffer.

sounds awesome. i don't want to change a thing because after a coupla months of tweaking, i love how it sounds. trouble is, in order to get it well and truly above unity gain (buffered bypass), i have to use a 100k resistor in the recovery stage's feedback loop and 10k to ground via. 10uF cap (so gain of 11?) - which clips either that opamp or the output buffer after it (or both). safest ratio i can go is 4.7:1, which is a little below unity on some settings.

any tips? as i say, i'd really rather not change anything tonally. i was thinking maybe i could try ditching the output buffer, but then there'd be a risk of all kinds of interactions with whatever comes after the volume pot. maybe just having a variable resistor in the recovery feedback loop would be a fix? but then it could still cause similar problems with other pedals' inputs. i've tried stuff like diodes between inputs, as per the BD2 (which i know nothing about), but no improvement. i'd rather not lower the bypass output to compensate.

any help appreciated!

[MrStab]

wondering whether or not i should connect the 10uF to ground in the recovery stage to vbias instead... though this post suggests it's okay as-is http://www.diystompboxes.com/smfforum/index.php?topic=49053.msg783530#msg783530

[slacker]

Could you put the volume pot between the tonestack and the recovery stage? That should allow you to turn the volume down so louder settings don't clip the recovery stage .

[MrStab]

i was thinking of moving it - or at the very least, the alternative of making some fixed attenuation, but i'd rather have it so it never, ever saturates the opamp, as opposed to having to fix it by manual attenuation. unless that's not quite what you had in mind.

cheers!

[samhay]

I assume you are using a 9v supply? If so, why don't you increase the supply voltage - a charge pump will give you +/- 9v for a hand full of parts, and this will usually save you a few caps in the audio section.

[anchovie]

Are you running a naked guitar into this, for your clean sound to be louder? Even if you aren't using rail-to-rail opamps you should still have at least 6V peak-to-peak of voltage swing before opamp clipping occurs, I would have thought.

[slacker]

I think what he's saying is some settings of the tonestack cause so much signal loss that he needs a gain of 10 to get back above unity.  Other settings have less loss so need less gain and therefore clip the recovery stage. Thinking about it a bit more this sounds a bit odd to me, maybe there's an error in the tonestack, schematic would be useful.

[thelonious]

I assume you are using a 9v supply? If so, why don't you increase the supply voltage - a charge pump will give you +/- 9v for a hand full of parts, and this will usually save you a few caps in the audio section.

+1. Unless slacker is right about something strange going on in the tone stack, this should solve your problem. And if you like the tone as is, you could keep the first op amp supplied by 9v and just run the second one off the higher voltage.

Anchovie mentioned a rail to rail op amp. Again, to keep your tone the same, you could switch just the 2nd opamp to a TLC2272 or TLC272.

[MrStab]

thanks a lot for the thoughts so far, guys.

a charge pump sounds like a perfect idea - and i have a coupla ICL7660S gathering dust - but i'd really rather avoid it to keep the footprint down. i'm on "version" 4 so the rest of the design is more or less consolidated. there's hope in that line of thinking yet, though - a different opamp. so you guys think a rail-to-rail CMOS opamp could help out here?

i'm using TL072s, i forgot to mention that. i'm broke as hell but could maybe see what i could score on eBay.
currently i (think i) have LM833s and OPA 2132s in my stash - not rail-to-rail, afaik, but i'll check the datasheets for fun.

if i could just squeeze a tiny bit more performance out of an opamp, then i could probably get away with a gain of 7-8. i'm happy to provide measurements if needs be.

oh - there are clipping diodes going to vbias in the path before the tone stack - as in no cap between: i THINK i remember cutting these out altogether to no avail whilst debugging, but this was a few versions ago, so i might have another look at those, Slacker, in case they are causing a problem.

cheers

[MrStab]

tried OPA2132 (in both buffer & gain positions) just to see if there was any difference, and as you'd expect, it still clipped but with different characteristics. i also cut out the diodes - no improvement (incidentally, there's another thread about soft and hard clipping together going on at the moment, but cutting either out of my build definitely takes away some texture from the sound. i use the "wire cutter A/B test". lol)

guess i'll order some TLC2272's then - shame i'll be affecting the sound of the first clipping stage, but i'm kinda out of options at the moment. maybe i could swap the priority of the opamps (ie. input buffer & gain 1 on TL072, boost & output buffer on TLC) - was planning to do something like that for layout reasons soon anyway. won't be ideal to test on the 3 iterations i have here, though (i try to keep them all up-to-date for "control" purposes).

[MrStab]

i only had a 100R resistor load on the recovery gain opamp, upped it to 1k (also tried 10k), and found that while i needed a higher gain to compensate, it does seem to help a little bit. it's that annoying Atari explosion kinda sound as the note decays that i'm trying to weed out. does that offer any clues?

[thelonious]

I hear that a TLC2272 is basically an updated, rail-to-rail version of a TL072. It might not change the tone of the clipping stage that much to drop one in there. The main drawback is the price. It would save you $1.50 per pedal if you kept the buffer and clipping stage on a TL072. If the price doesn't matter to you, heck, drop two in and see if you like them in both positions.

Especially since you have clipping diodes going to Vref before the recovery stage, I'm really surprised you're clipping the op amp. Between those diodes and the tone stack loss, that just shouldn't be happening (says me, anyway). If you up that 100 ohm load resistor to 1k (assuming that comes before the clipping diodes to Vref), I'd put a 1k resistor in series with the clipping diodes just to see what happens.

[MrStab]

^that's reassuring re. the tone, Tristan - after some tweaking, i think i'm ready to bite the bullet and order a couple just to try out.

yeah, it just doesn't seem to make sense. at least 2 different builds i have in front of me do this. the 100R-turned-1K is actually on the output of the recovery stage, so in terms of coupling, that whole section is diodes > tone stack > recovery stage. it did the same thing when i had it all anally separated with caps, too. there is a 1k resistor coming out of the first clipping stage, but there's a cap between that and those diodes.

i know it's a bit of a pain without a schematic - things keep changing so much that i can't keep up with making them.

as it stands, it's not totally atonal, but meh... the TL072 doesn't seem to clip gracefully. i tried a LM358 in there to see if its low slew rate would help, and it did kinda, but it sounded far too muddy compared to the TL072.

[slacker]

Are you sure all the stages are biased properly, inputs and outputs sitting around half the supply voltage? I'm also still a bit confused as to what the problem is, are you saying you can't get the output the same level as the input without the opamps clipping?  Or do you mean that at some settings you need a gain of 10 to get to unity which then means that other settings which don't need as much gain cause clipping?

[thelonious]

i tried a LM358 in there to see if its low slew rate would help, and it did kinda, but it sounded far too muddy compared to the TL072.

TLC2262 has a lower slew rate than the 2272; you might want to grab one or two of those to try out if you liked the lower slew rate of the LM358 but didn't like the mud.

[MrStab]

Are you sure all the stages are biased properly, inputs and outputs sitting around half the supply voltage? I'm also still a bit confused as to what the problem is, are you saying you can't get the output the same level as the input without the opamps clipping?  Or do you mean that at some settings you need a gain of 10 to get to unity which then means that other settings which don't need as much gain cause clipping?

biasing seems fine. everything works great up to about 5-6 recovery gain, but in order to reach unity, i need so much gain after the tone stack that it clips either the non-inverting gain stage or the following opamp buffer. even when i roll back the Gain pot from the previous stage (basically a glorified TS clipping section), it still clips. i thought of maybe moving the volume pot to make some kinda "pre-gain" type control, to attenuate what goes into the offending section, but i feel that'd be a poor compromise. there is a f***-ton of distortion going on, but that's kinda the point.

TLC2262 has a lower slew rate than the 2272; you might want to grab one or two of those to try out if you liked the lower slew rate of the LM358 but didn't like the mud.

damn, just ordered some 2272s. my goal with going rail-to-rail would be to avoid clipping altogether, but i'll see how the 2272s work out and maybe the 62s could be a good compromise.

nagging feeling: just putting a resistor from the gain stage's - input to vbias instead of via. a cap to ground really wouldn't make a difference... right?

thanks again, guys. i feel like i'm so close to having this working and i like the sound so much - inelegant though it may be - that i don't wanna have to do a drastic redesign!

[thelonious]

damn, just ordered some 2272s.

You'll like 'em, I think. Even when they do clip, they clip a bit more gracefully than normal op amps.

just putting a resistor from the gain stage's - input to vbias instead of via. a cap to ground really wouldn't make a difference... right?

If the cap to ground is big enough to pass all guitar frequencies, I don't think you'd notice any difference. But there's one way to find out...

Probably a dumb question, but I assume you're using a version of the James stack that has been impedance scaled to work well with op amps instead of the tube amps it started in? Also, if you get a chance to pencil-scratch out a messy schem of the diodes > tone stack > recovery stage section, that would help. I'm just wondering about things like the tone stack loading the op amps and also exactly how the recovery stage is getting biased.

...but maybe you will find that the rail to rail op amp takes care of the problem and won't have to worry about anything else!

[MrStab]

i used the James tone stack schematic verbatim from the Tone Stack Calculator app. so basically this, only with different values and ground becomes vref:


that 1M resistor is what goes on to bias the opamp's + input. impedance-scaled, you say? i did try lower values as Zrc source in TSC, and everything seems fine, but nothing beyond that.

i'll try to stick some schematics of the different sections together asap.
just tried a JRC4885 i ripped from an old amp - the opamp clips a bit better, but it's not the sound i'm going for.

cheers

[MrStab]

okay, i quickly stuck this together:



there's a cap before all that, forgot to add it in. i plan to have a hard/soft switch at some point, even if having both simultaneously is futile, so i'd rather keep those diodes if possible.

[PRR]

The James with correct-taper pots has a loss of 10:1 (or 11:1). If you have a happy no-tone system, and add a James, you need to find/add a gain of 1:10 (or 1:11) after it.

If you need more gain than that to recover a James, then you have a wiring or value error.

There is also a "linear James" which only has 2:1 loss (and only 5dB boost).