Op amp distortion

Just to sort of remind about the types of distortions we are likely talking about:

Crossover distortion generally sounds very bad. It sounds worse at lower amplitudes. It is a short span over which the op-amp doesn't act well. It also seem the worst at lowish frequencies because the harmonics all can be heard.

Slew rate distortion sounds worse at there are more high frequency components to require the op-amps output to move quickly. The TL082 IIRC can be trusted up to a slew rate f about 5V/us A 3KHz signal with a 4V peak voltage gives 4*2*pi*3K/1E6= 0.075 V/uS so slew rate isn't likely to be the problem

Clipping happens at high volumes so we can rule it out too I think, unless you have made a serious mistake. If the output of the op-amp is very heavily loaded, the built in current limiting can cause clipping.

A heavy capacitive load on an op-amp can cause all manner of bad things to happen. One of those bad things is oscillation at RF frequencies.

If you have a scope, you could observe the signal as it goes through the chain, and identify he [roblem stage by eye..without a scope it will be very difficult

Marc

That OP-AMP falls more into the family of logic.. The output pairs on the OP-AMP do not have internal biasing to maintain minimum current on the pairs. so in short, the current becomes discontinued in the cross over region. This will cause distortion for AC wave forms if you're using a differential rail supply. (+/- rail with common). placing a (-) feed to the op-amp output via a resistor will force the op-amp + side of the emitter follower pairs to maintain steady current how ever, using the (-) output signal still generates some distortion. So basically, you most likely removed 50% of it at least.

In the case of a single rail supply, the OP-AMP is biased so that the output cross over point is at 50% of Vcc for example. This places the OP-amp out of that region how ever, a load resister is still required to common. With out this a couple of side effects take place..

Try loading the output to common via a resistor..

--
"I\'m never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

"Mike Faithfull"

** FORGET using LM358s for audio (other than maybe a guitar stomp box).

Stick with the tried and proven TL072 / 82 series, the NE5532, the RC4558,

4560 series and even the LM833.

All these can be trusted to be free of x-over problems and to have low THD.

Note, the TLs and the LM833 can display a tendency to oscillate at frequencies in the MHz range if wiring layout or supply decoupling is poor OR if the output pin is not isolated from load capacitance by 100 ohms or so in series.

Such oscillation will often result in significant audio distortion.

...... Phil

Thanks for all replies. Regrettably, my days of regularly getting to grips with this technology have long gone, so no 'scope is available, only the trusty old AVO 8 and a cheap digital test meter that doesn't actually say "made in China" on it anywhere but we all know don't we .......

The LM358 was chosen originally because I got the impression from the marketing stuff that it was "optimized for single supply operation" and I wanted to use a +12v power source available from the power amp and avoid the need for additional cost/components to make a +/- 15v supply. This meant that I was having to make modifications to "textbook" circuits to run off a single rail and that caused me some problems. Therefore, after a lot of experimenting to no satisfactory conclusion, I bit the bullet and did the right thing !!

The balanced line input stage circuit is almost a copy of the one shown in Figure 2 found at

. With the benefit of hindsight I would probably have been better off "upgrading" to Project 87 and building it on a pcb purchased from ESP, but with the exception of the ICs I had (almost) everything I needed in my junk box to do it the cheapskate way. The "almost a copy" in my version has the following differences from the published schematic:

C1 not fitted ('cos the only things small like that I had to hand were committed elsewhere (see below) ) R11 is 27K and R10 is 22K and a 10K pot in series (so I could "tune for max common mode rejection" and because I had used the last of my 10K resistors anyway!) R12 is not fitted and the output on pin 7 connects through a 10uF capacitor to the 'top' of a 10K pot in the mixer stage which is grounded at the other end. The input to the next stage is then taken off the slider via another

10K resistor. Pin 7 was temporarily connected through 3.9K to -15v to try to eliminate the crossover distortion like in the mixer stage, but while it had the desired effect in the latter, it didn't seem to make any difference here.

(Little light comes on in head ...) So perhaps crossover distortion is not the problem here ... The cicuit is built on stripboard and although I did connect some small value ceramic caps between the supply pins and ground at each IC location, I haven't otherwise paid much attention to layout or the possibiity that the circuit mught be hooting at some undetectable frequency as Phil suggested.

Time to see if I can borrow a 'scope from somewhere I think.

I really am most grateful for all the help I'm getting with this, by the way. Given the reason for the project, failure will inevitably bring egg on face and comments about "thinking he's cleverer than he really is" from other band members.

But surely life is partly about making mistakes and learning from them isn't it? If you never make a mistake you're not trying very hard! :-)

There are old scopes available on ebay and elsewhere for a fraction of the new cost. Although you need reasonable fidelity, you don't need really high bandwidths for these sort of measurements - of course, such older stuff may actually have a price premium as antiques ;)

Cheers

PeteS

Mike Faithfull wrote: (snip)

Changing from single supply to dual supplies does not change the cross over distortion situation, if the output load is capacitor coupled. In either case, you decrease the distortion by adding a load resistor from the negative rail to the output pin, to maintain a small DC bias current through the output stage. The required resistor value depends on the AC load impedance and output swing. The combination needs to produce a unidirectional load on the output. But even with this band aid, the LM358 is not a very high performance opamp at the upper end of the audio spectrum. Okay for low gain, low swing applications, or low frequency (audio limited to 5 kHz, say) because of its low slew rate (only .3 volts per microsecond) and low gain bandwidth product (of only about 0.7 MHz). Normally used for audio only if very low cost is important.

Are you located in the US? I have several old ones to sell for a little that I got from eBay and checked out.

Yes, my research indicated this ... 3.9K felt about right (and I had several to hand!)

(snip)

I'm in the UK - south east, just north of London. At one point I did contemplate buying a 'scope, but it would probably not be used again after this project and would only sit and gather dust in the loft with the MicroVax and the HP pen plotter, the antenna rotator, the DECMate and all the other stuff I can't quite bring myself to get rid of, so good sense prevailed .......... well, thus far anyway ;-)

It isn't. I like the TL072 (and friends) for dirt-cheap, "don't care" amplification purposes.

robert