Help! My buffer is distorting!

[tdkrause]

I'm working on an overdrive design that sounds great but has a really annoying flaw.  Since I had a spare op amp stage I decided to use it as the output buffer.  The signal coming from the first op amp stage is so hot that I had to move the volume control to just before the output buffer so that the signal can be attenuated before the buffer input.  That definitely helped, but now if I turn the volume up past unity gain (compared to the bypassed signal) the buffer starts distorting again and sounds terrible. Even running it at 18v I can't turn the volume up too high without it distorting. After the volume control is a .22uf capacitor, then the op amp buffer (biased with a 1M resistor to the reference voltage of a 68k/68k divider), then out through a 10uf cap. Would it make a difference to have another resistor after the volume control? Could it be an impedance problem?

[samhay]

Something doesn't quite add up, unless: (i)  you have an additional voltage amplification between the two op-amp stages; (ii) your 'buffer' has gain; or (iii) my favourite, you have a perpetual energy machine.
I think a schematic is probably going to be the easiest way to get to the bottom of this. However, for arguements sake, say you have a rail-to-rail op-amp. The first stage can put out a 9V peak-to-peak signal if powered with a +9V supply. Your second op-amp, if on the same supply, can also put out 9V P-P, so your first signal can never be too 'hot'. My guess is that your biasing is not working right - are you using the (+) or (-) input of the buffer op-amp?

[Thecomedian]

I built a buffer design of my own that purposely has an "overdrive" distort effect, and got the +/- voltage peaks just about identical from a Jfet/NPN combo. What I noticed while playing with the original buffer design is that a buffer WILL distort if too much voltage going to the "buffer stage", you will get distortion. In extreme cases, with a single silicon transistor you can get an "overdriven" positive voltage peak, and cut off for the negative voltage input. The two possibilities I see for this circumstance is that the signal is not "small signal" anymore, so the amplification has moved into Curve territory, OR, there is insufficient feedback biasing to keep the transistor within the "Q point" to provide 0.9999 gain and impedance transformation of the signal (buffer being to transform a higher impedance output into a very low impedance output).

I suspect that since a buffer should work in any signal chain (which means you could have gains moving a signal from 10mv to 5V+), the problem is the latter: more voltage in raises the Q point. That would mean lack of feedback stabilizing.

Option 3: as samhay pointed out, maybe you don't have the op-amp for the buffer taking the output into the right pin, or some other variation.

Check and see if the circuit is following the schematic exactly, and if it is post the schematic.

[slacker]

Is the reference voltage bypassed with a big cap, if not perhaps it's some sort of oscillation.

[Paul Marossy]

perhaps it's some sort of oscillation.

If everything else checks out, could be a possibility. Might be oscillating ultrasonically and causing ugly distortion in the audio band. If this were the case, a small cap to ground somewhere around the buffer might fix it (not sure to try input or output to buffer, probably on the output I guess). Maybe 47pF a would do it.

Also, is there a small cap in the feedback loop? If not, try that too. Try a 47pF to start with.

[tdkrause]

Thanks for the posts. You,ve given me a lot to think about. I'm away on business for a couple of days but as soon as I get home I'll try to post a schematic. I do have a bypass capacitor in place so I don't think it's any kind of oscillation and I double checked the pin out to make sure I didn't connect the input to the wrong pin. One thing I didn't think of is the possibility of a bad preamp tube. It's a long shot, but worth looking into. It seems to me I had already ruled that out but it would be nice if it were that simple. Also, is it possible for an op amp to be so far out of spec or for one stage of an op amp to be damaged by ESD that it would exhibit this type of behavior?

[Paul Marossy]

Also, is it possible for an op amp to be so far out of spec or for one stage of an op amp to be damaged by ESD that it would exhibit this type of behavior?

It could be a hosed up opamp. I've had defective brand new unused opamps before, so it's worth swapping it out for a different one to see if the problem persists. It just happened to me the other day, although in my case it had an extremely low output, no oscillation problems.

[Johan]

You said the buffer is biased through a 1M resistor, but didn't say what kind of opamp. That would probably work if this is a tl072, but not if its a 5532. not enough bias current at nine volts. Try reducing your bias r to 100k

[tdkrause]

Interesting, Johan. I hadn't thought of it as a current problem. I've only been thinking of it as a voltage problem. The op amp is a TLO72 but I suppose it's still a possibility. I'll try a 100k resistor when I get home. Being self-taught and having no education in electrical theory, I tend to just use 1M resistors everywhere without giving it much thought. I'm usually so glad when something works that I don't care what the output impedance is (not that I could calculate it anyway) or whether my power supply has enough current available.

[Johan]

The bias current doesn't have to be the problem, just one more thing to consider. It's usually not something we worry about much in this forum, but if you study different schematics you'll see that tl07x is often biased through 1M, 4558 rarely over 470k and 5532 rarely over 100k. If there is not enough current, the bias voltage at the opamp will stray and the output will be offset. I've used this effect in the past to get a dynamically shifting asymmetrical clipping, but would not recommend it for anting other than one offs.....it might not be your problem, but one more thing to look at

[samhay]

The bias current doesn't have to be the problem, just one more thing to consider. It's usually not something we worry about much in this forum, but if you study different schematics you'll see that tl07x is often biased through 1M, 4558 rarely over 470k and 5532 rarely over 100k. If there is not enough current, the bias voltage at the opamp will stray and the output will be offset. I've used this effect in the past to get a dynamically shifting asymmetrical clipping, but would not recommend it for anting other than one offs.....it might not be your problem, but one more thing to look at

You're right, but as RG might say, the (other) details (also) matter - i.e whether you are using a bias resistor to Vb and whether you are using the op-amp in an inverting or non-inverting way.

I've used the 'dynamically shifting asymmetrical clipping' trick too - it can work really nicely.

[R.G.]

Also note that some opamps, including particularly the TL07x family, will oscillate when loaded with a capacitance on their outputs, particularly when they are operating at a gain of unity. It's one of those quirks. If you see small-value resistors, 10R to 1K, in series with the output of an opamp, it's usually for this reason or to limit current if the output is shorted.

Your situation is one where an oscilloscope is ...really... useful.

[tdkrause]

More useful info! Next time ill just let you design the whole thing for me. And I always wondered why I often saw small resistors at the output of a buffer. While we're on the subject, what's the purpose of the 4.7uf capacitor between the emitter and VR on the Pete Cornish buffer? Is it a pull-up to bring it up closer to unity gain? RG, I have a couple old Tektronix scopes but I have no idea how to use them. Can you recommend a good tutorial?