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

[MrStab]

about a 1:10 ratio did gimme enough boost, Paul - i was just hoping to get a little over unity while i was at it. not sure how i wound up as high as 150-200k in my travels but for the most part i'd been using 10k:100k, which seems to be ideal... except for the clipping. i'd been approximating how much recovery i need by looking at the graph in TSC. the values i used aren't far removed from stock at all. it'd be a shame to have to redesign the whole tone section as it works quite well, but do you have a link to the linear James stack just in case?

no obvious errors, 3 builds (1 using an entirely different layout), recovery gain adequate... weird. all i have left to try is seeing how a rail-to-rail opamp would respond, in either recovery or output buffer stages.
i don't suppose anything about the output buffer could be suspect? it's about as basic as you can get buffer-wise - 1M resistor to vbias (though 10k worked too), 1k on output, 100k pulldown resistor. or maybe some power-related thing from having 2 gain stages (one of which quite extreme) on the same dual opamp?

cheers

[PRR]

Scaled for solid-state type work.



You generally don't need R5 if your "common" is at good DC level to bias the next stage. There is a DC path through the bass pot. Yes, if you are paranoid you use R5 anyway just so things don't go nuts when the pot wiper gets dirty.

[thelonious]

okay, i quickly stuck this together:

Ah, ok, that schem was helpful. If you want a gain of 15-20 out of the recovery stage, I think we're barking up the right tree---you need either rail to rail or higher supply voltages, or both. I'm interested to hear if the 2272 does the trick. If not... check out the schematic for the Maxon OD820, which uses a higher voltage bipolar supply for the op amps and also places the volume control before the final recovery stage.

PRR hooked you up with scaling values. It's done mainly for lower resistor noise in high gain circuits or to prevent the stack from being loaded too heavily by whatever follows it. But if you're happy with the noise performance and frequency response as it is now, maybe you shouldn't mess with it.

[slacker]

I don't think your problem is opamp clipping. Paul says the tone stack has 10:1 loss and you're then boosting by x 10 so that only puts the output of the gain recovery stage about the same level as the output of the clipping stage. The level of the clipping stage is limited to about 1.5 - 2 Volts peak to peak by the diodes. A TL072 clips at about 6 Volts peak to peak, so it shouldn't be clipping, if you were getting a 6 Volt signal out of it then it would be much louder than your bypass signal.
It could be oscillation caused by too much gain, maybe try some small caps in the feedback loops of the opamps. Or it could be your 4.5 volts isn't stable enough, what are you using to make that.

[samhay]

How much gain do you have before the clipping diodes? If there is enough, you could use two each in series (or LEDs) to give you more signal going into the tone stack.

[MrStab]

Scaled for solid-state type work.

thanks a ton - i don't know why the hell this didn't occur to me when i spent at least an hour tweaking the values. guess i kinda thought "well, the stock circuits must have this much loss for a reason". if the en-route TLC2272 doesn't improve things, i'll use your edit as a reference to bring my own values up to scratch. less resistor noise sounds like a plus. incidentally, my treble pot is a 470K linear, which seems to work okay.

PRR hooked you up with scaling values. It's done mainly for lower resistor noise in high gain circuits or to prevent the stack from being loaded too heavily by whatever follows it.

hmm! loading... didn't think of that.

It could be oscillation caused by too much gain, maybe try some small caps in the feedback loops of the opamps. Or it could be your 4.5 volts isn't stable enough, what are you using to make that.

just a 10k/10k voltage divider, 47uF cap to ground. not the ideal way, but surely enough without regulating it in any way. all 3 of the other sections are fine with it.
i have a 47pF cap in this feedback loop and a 100pF in the preceeding one. against all advice, experimentation led me to a low-pass filter at 12kHz at the end of everything, which seems to tame highs without losing anything. this addition is independent of all the clip-like problems, though.

How much gain do you have before the clipping diodes? If there is enough, you could use two each in series (or LEDs) to give you more signal going into the tone stack.

i have LEDs in the preceeding section, didn't think about for this one though. definitely going on my list of things to try.

okay, so my strategy is try the new opamp, then revamp the tone stack, and failing that, try upping the forward voltage of those diodes.  

thanks a lot guys, it's a weird one and there's no doubt a lot of subjectivity involved, but i think i'm one step (or a few!) closer to finding out what's not quite right.

[PRR]

> the stock circuits must have this much loss for a reason

Serendipity.

You may want 1:8 boost if there is a great discrepancy between the producer's speakers and ears and your (early 1950s) speaker and ears.

A passive network can't really boost, only cut. The work-around is to cut everything, and un-cut the parts you want to "boost". Then make-up in an amplifying stage.

Since there are always losses, a 1:8 "boost" probably needs a 10:1 overall cut.

The old standard Audio Taper pots were 10:1 at mid-rotation.

So everything is 10:1 ratios. The bass pot wiper is 1/10th of the input. This is bypassed by two caps in a 10:1 ratio, and stoppered by two resistors in 10;1 ratio. The treble side has 10:1 caps into a 10:1 pot. (The resistor from wiper to wiper isn't in any special ratio.)

That's fine for relatively large "boosts" when system levels allow 10:1:10 down/up level dips.

Sometimes we'd like less loss, and will accept less "boost" to get it. Modern 20% "Audio" pots are not a lot better. Plain Linear pots give a good 2:1 loss, and any well-planned(?) audio path can manage a 2:1:2 gain dip.

I find that high-resistance pots give more trouble. 10K or 100K is my happy value. 10K pots lead to around 1K midband impedance which is rather low. Hence 100K. The mid-high frequency impedance is suitable for the hiss of common transistor/chip amps.

[MrStab]

i think i get what you mean about accepting less "boost" - it correlates with my observation that there seems to be less-intense a mid scoop when maxed in the linear version. maybe i could have it just a little deeper, but hopefully overall functionality won't be affected much either way.

so your example values were 10x less than the default values in Tone Stack Calculator [edit: i need to pay more attention] - i can't seem to simulate that with my own values, would it even be possible to simulate the updated version in TSC? i'm SPICE-illiterate, as it stands. any tips on "converting", or should i just start from scratch? i could type out the values and ask for a handout, but i'd learn more by figuring it out on my own.

i'm liking this solution a lot more now i've had time to absorb it - it's making me think the tone stack i have at the moment has an excessive load for the levels involved.

thanks again

[MrStab]

i'm getting there - forgot to change Zsrc to 2k. think flat at -10dB might be a decent compromise, or still too much on the low side of things? the right amount of make-up gain did seem to be just on the verge of doom, so maybe i'll be okay with that.

[PRR]

> simulate the updated version in TSC?

My image above _is_ TSC; just sorted a bit for posting.

> change Zsrc

The ideal Zsrc is zero. As a practical matter, 1/10th the value of R1+R3 gives a 1dB slump from bass to treble which may be acceptable (and can mostly be fudged-out by running the treb knob a bit over half-rotation).

[MrStab]

i figured it out - at the time of posting i'd misunderstood your drawing a little bit. i can't conclude much yet as i've only listened on headphones (3am), but the revamped tone stack seems to work fine and a gain of 2 seems ideal (going for the nearest common part, 10k:10k in the gain stage). might raise it a little bit, but i'll be conservative as i go. i'll test at full-blast on a speaker tomorrow.

interestingly - and again, i can't say this conclusively as i've just used headphones for now - there's virtually no susceptibility to squealing in true bypass mode anymore. i'd had it fixed with shielded wiring & other precautions so it was audibly just on the edge of squealing, but now nothing.

hopefully this is the fix, i'll report back. in any case, thanks a lot, Paul and everyone else, because this may have smoothed out the whole thing and i've gained (pun not intended) an even deeper understanding of the circuit in the process.

[MrStab]

okay, so i have one version with a less-loading tone stack - no more opamp farts at unity+, but disappointing change in tonal characteristic - and another with the old tone stack but using TLC2272s - no clear opamp fart, it's definitely better, but i still can't tell if it's quite right.

i've listened so much that the ol' psychoacoustics are screwing with me. it's that end-of-chord breakup i can't figure out, say after you've let a power chord run out. it's not obviously bad, which is a start, i guess. i'm testing on a crap solid-state Marshall MG, which has seen better days, so tomorrow i'm gonna try with a decent tube head & 4x12 cab.

i can't really tell whether the remaining breakup is opamp breakup or just bad amp breakup. i want to be sure it's not the former. any tips on how i could establish this objectively? or any recordings i could compare with? i'd record a sample, but i'm overdue re-installing Cubase and for some reason my recording setup is a bit hissy - maybe it'll be clear enough for someone to identify whether things are right or wrong, though. any volunteers? if any recording isn't horrific, ofc.

also, would it help to omit the 100R resistor right after the recovery stage? increasing that had a considerable effect on the amount of gain needed, i just wonder if it's contributing to the problem.

cheers, and sorry for the continuous harrassment!

Edit: took some time to reset my cochlea, gonna try & tweak the linear James stack a tiny bit more & see how that pans out before jumping to conclusions.

[MrStab]

this thread is dead, but i thought i'd let everyone know that after a few days' hassle, everything sounds awesome. 2 builds sound near-identical, and the 450V carnage of my tube head trumps the ~30V on offer with my SS amp for testing. all doubt erased. i like this distortion.

i'd share it, but there's no concrete schematic drawn up yet & i fear it's quite unremarkable, design-wise. but it does the job like a beast. 'scuse me while i pry my gonads off the wall. thanks again to everyone for all the help! Paul's tone stack suggestion was the saving grace, but everyone else's suggestions forced me to analyse the circuit in greater depth, in ways that helped me tweak it further.

cheers!

[pappasmurfsharem]

this thread is dead, but i thought i'd let everyone know that after a few days' hassle, everything sounds awesome. 2 builds sound near-identical, and the 450V carnage of my tube head trumps the ~30V on offer with my SS amp for testing. all doubt erased. i like this distortion.

i'd share it, but there's no concrete schematic drawn up yet & i fear it's quite unremarkable, design-wise. but it does the job like a beast. 'scuse me while i pry my gonads off the wall. thanks again to everyone for all the help! Paul's tone stack suggestion was the saving grace, but everyone else's suggestions forced me to analyse the circuit in greater depth, in ways that helped me tweak it further.

cheers!

Clips and Scheme.

Even if its just YATS its always fun to hear how different components interact.

[MrStab]

this is all i can manage for now, a computer "scope" reading (which i no doubt mis-used and is probably clipping the audio interface lol):



i used a tone generator from a laptop at 1kHz. gain fully-maxed, tone controls flat.
i'm stupid-busy lately, but i really need to fix my recording setup for my band's sake so watch this space!