LM324 and crossover distortion

You're right, the whole point of the trick is to make the output class A. Whether that's a good idea depends on what the load is. It helps that the 324 is usually used in single-supply applications, where the load is generally referred to the negative rail.

Cheers

Phil Hobbs

Get a better opamp!

John

Yes, the point is to make it a class A stage. I'm not sure about the LM324, but on the LM358 you can use a pull-up resistor that can not only make the stage class A, but will let the op-amp output go up to the positive rail instead of 1.5-3V short of it.

Just about any op-amp is going to have crossover distortion. The question isn't "is it there" but "how bad is it". Just run a sine wave through an amplifier using an older op-amp and look at the output -- there'll be a flat spot in the middle, that gets more pronounced as the frequency goes up. That's from the crossover distortion of the output stage interacting with the thing's internal slew rate limit.

Hello Mark,

I agree on this. As you can guarantee that only the upper or lower output transistor is conducting, there is no cross-over distortion anymore.

Best regards,

Wim PA3DJS

Even shifting crossover to a higher voltage can be an improvement depending on the kind of signals being processed. If the signals have a varying amplitude about 0V, crossover will bite you every "cycle". If the crossover distortion is shifted to some other voltage, preferably well towards the rail, it will bite you much less frequently for many signals. And in proportion to signal level, the distortion is reduced.

But I'm with John L, the 324 family sucks - get a better opamp. The technique of shifting the crossover can reduce distortion even with a better opamp, but unless you're scrutinizing every penny there are many better choices out there.

Cassiope,

Why don't you produce a list of "better" OpAmps and their cost per OpAmp? ...Jim Thompson

The '324 must have been one of the first quad op amps around, mid seventies ?, We were quite excited by it's arrival at the time. Turned out to be too noisy for audio work, but plenty good enough for a myriad of other applications. The little bit later Raytheon 4136's and similar were better, but the 324 was still a major milestone in the art, imo.

Regards,

Chris

Not forgetting the ubiquitous LM339 quad comparator, probably around the same time and still using in designs now. Not bad at all for a 35+ year old design. Dirt cheap as well...

Regards,

Chris

In my experiments the problem with the LM324 for audio is also its slew rate - with a slew rate of 0.3 V/usec the maximum amplitude of a sine wave at 20kHz would be about 1.5 volts before running into slew rate limitations, and at the same amplitude a square wave will stop looking like a square wave at a much lower frequency.

I'm sorry, I did the math wrong - that should be 15 volts! But the problem with square waves still occurs!

Sorry, I did the math wrong - that should be 15 volts P2P! But the problem with square waves still occurs!

I have buckets of both in my parts bin :-)

Also loads of TL08x, which had become my favorite jelly bean OpAmp by around 1980. ...Jim Thompson

What do you think of the 4558? It seems a lot of consumer audio equipment uses that one.

Neeerp !-) Keep trying...

S = Omega*VP

So VP = S/Omega

My very first OpAmp design, MC1530/31, circa 1963-64, had 6V/us slew-rate :-) But it required external compensation.

And it's still being sold, 48 years later...

I've even had inquiries to re-design it on a modern process :-) ...Jim Thompson

Never had occasion to use it. I stopped doing any significant number of discrete designs by ~1990 and instead specialized in custom mixed-signal chip design. ...Jim Thompson

Whoops! I forgot the factor of 2*pi, and that it should be the peak amplitude, rather than the peak to peak amplitude. I got 6 hours of sleep last night, that's my excuse.

Is the LM324's slew rate so poor mainly because it's designed to be low power?

So? That's about my norm... us old farts don't need much sleep ;-)

Partially. It's a complex function of first stage gm, the use of "pole-splitting" compensation (only two stages of gain), etc. My gimmick in the MC1530/31 was the use of a ZERO (lead) in the compensation.

...Jim Thompson

Forgot another bragging point for the MC1530/31... sliding-class-A output stage... no cross-over distortion ;-) ...Jim Thompson

Tim -

Is that true for the TL072 whose specs say .003% THD?

John