Op-amp overdrive too loud

[chrisjuby]

I'm a fairly new builder. I read the following article and built a version of the example circuit at the end using a TL071.

http://www.generalguitargadgets.com/how-to-build-it/technical-help/articles/design-distortion/

It sounds great, but the output is so high that I have to keep the level pot barely above 0 for roughly same volume as on bypass, and if I switch to the hard clipping diodes it's even louder still. I'm keen to box it up, but I'd rather have the level range a bit more sensible before I do. Any advice on what I should do?

(I've noticed that a lot of commercial circuits have output buffers but this one doesn't. Is that the issue?)

Here's the example circuit:

[GibsonGM]

Do you know what the "taper" of your output pot is?  If you use a 100K AUDIO TAPER pot, the 'change' in apparent level should happen more smoothly.

This has to do with how the human ear responds to sounds, in a logarithmic fashion...so we tend to use audio (log) taper pots for volume controls...

http://www.geofex.com/article_folders/potsecrets/potscret.htm

[Groovenut]

You could try putting a 10k-100k series resistor between C5 and R7. That will bring the total output level down by making a voltage divider with the output level pot. I would put it just before the level pot to avoid changing the way the tone control behaves. This will also alter the frequency corner for the RC filter created by C5 and R7, allowing more bass frequencies through. So you may need to adjust the value of C5 as well.

[Frank_NH]

This brings up an interesting question (for me).  What output voltage level (rms) should one strive to achieve in a typical effects pedal?  Line level is about 1 V.  Guitar signals vary from 0.1 V to over 1 V (max).  A guitar amplifier is designed to handle guitar level input voltages, so should your max output from your effects pedal have a gain of something like 5 - 10 dB more so you can achieve well over unity gain?   I note that some "boost" pedals tout gains of 20 - 30 dB and more (e.g. the SHO).

[induction]

I have to keep the level pot barely above 0 for roughly same volume as on bypass, and if I switch to the hard clipping diodes it's even louder still.

Based on this sentence, I'm going to guess that the problem is too severe for a different taper to solve it. I could be wrong though, and it's easy enough to try it for yourself to make sure.

My suggestion would be to swap the output pot for a smaller value, and add an inline resistor (between lug 3 and the C5/S2/R6 junction) to pad down the volume. If you go for a 50k Audio and a 47k resistor, this will limit your output the 0-50% range you get with the pot you have now, without changing the electrical characteristics of the circuit. Personally, I would try a 10k pot, and then choose an inline resistor to taste. But you can always experiment with pot and resistor values and see what works best for you.

Edit: ie. What Groovenut said.

[chrisjuby]

Thanks very much, everyone. All really helpful.

Slightly embarrassingly, it turns out I had the A100K level pot wired backwards - so presumably it was functioning as an inverse log. That obviously made a lot of difference straight away and the inline resistor also helps.

One further question...

There's still quite a big difference in volume between the soft and hard clipping circuits. Is there a way to reduce the output of only the hard clipping diodes?

[antonis]

Only if you replace them with Ge or Schottky diodes (or some other bypassing item with forward voltage drop less than 600mV..)

Although I think that your problem is somewhere else...

Replace OUT pot with a 47k/47k voltage divider and observe if the loudness is relatively OK..

[Mark Hammer]

I originally had a different response prepared until your last post, so....

Groovenut's suggestion of the additional series resistor is spot on.  Since you have indicated the circuit is not boxed up yet, consider use of a rotary selector switch that gives diode-pair A, diode-pair B, both diode pairs, and no diodes.  A 3-pole, 4-position rotary switch is cheap and available.  One set of contacts dictates whether diode-pair A is connected or not.  The second set of contacts dictates whether diode-pair B is connected.  And the third set of contacts straps different resistances in series with the input to R7 to set equivalent loudness for the 4 positions, so that the different connections only result in tonal changes, rather than volume changes too.

The 1N914 pair should result in about 25% more loudness than the 1N4001/1N34 pair.  So, what you will want to do is stick a 22k-27k resistance in series with input to R7 when using only the 1N914 pair, and perhaps an even larger value when not using any diodes.

[chrisjuby]

Great suggestion, Mark Hammer. Thanks very much!

[ashcat_lt]

At what point does all this series resistance start to affect tone and other things down the line?  Remember that there's a big long capacitor to ground (you call it a cable) hanging off the end of this thing, and the top of the V pot and whatever stopper R you put in there set the cutoff of that LPF.  You're also getting to where it's less than 10:1 ratio with a fairly typical 500K in-Z for whatever's following it, so that the taper and actual output becomes more and more dependent on outside forces.  And forget about plugging it straight into a Line input.

That's where you start to want the buffer that was mentioned in the OP.

But then somebody said something about using much smaller Rs, which is a half-decent workaround.

[R.G.]

At what point does all this series resistance start to affect tone and other things down the line?  Remember that there's a big long capacitor to ground (you call it a cable) hanging off the end of this thing,

Yep. There are capacitors everywhere, and they both enhance and reduce treble response. What's a musician to do?

Well, understanding is a big part. The reason 10:1 oscilloscope probes exist is not because they let you deal with bigger voltages. It's because they let you tune out the effect of the cable capacitance between the scope probe and the scope itself.

You can't make the cable capacitance zero, and even small capacitances matter at many-MHz scope frequencies. What you can do is to use a 10:1 resistor+capacitive divider, where each resistor is paralleled by a cap, and the cap in the small-resistor side is the cable capacitance. You can then use a small, and adjustable, capacitance on the big-resistor side and tune in the square-wave response for nice, square corners. That result lets you have level response from DC up to some *very* high frequency. Similarly, here you could parallel any series resistance designed to lower the level with a capacitance designed to correct for the capacitive loading of the cable run. Mostly... it's never perfect.

That's where you start to want the buffer that was mentioned in the OP.

But then somebody said something about using much smaller Rs, which is a half-decent workaround.

Both are OK, too.

[ashcat_lt]

Ahhh..."treble bleed"...That can work too!

It's also worth noting that the nature of soft clipping is such that you can never actually predict the maximum output.  Well, OK, the maximum output is where the opamp clips, but that is going to be a whole lot louder than the hard clipped option.  I can see where "typical" guitar inputs might come out quieter when soft-clipping is engaged here, but if the input gets much bigger, so will the output.  With a really hot humbucker (or worse, a couple in series with each other), an active guitar, a boost before it, or like a keyboard or something, your results could be drastically different.

[blackieNYC]

I dont think I've built one (previously established) circuit in which I can turn the output volume past 11 o'clock, if that. Most circuits seem to be very interested in "pushing the amp".  Off a cliff.  I now add a series resistor, maybe 5 times the value of the output pot, just to get within a reasonable range of the pot.