A Quad Opamp Oscillator Shootout

[DrGonzoDK]

Data sheets are one thing. Real world performance is another. Graphics cards and CPUs have benchmark comparison sites, but we humble analog circuit people don't have an equivalent. Therefore, it can be a bit difficult for an enthusiast to pick a part that matches aproblem.

So I've decided to try and help a bit. And to get things started, a shootout between 5 (6, really) different quad op amps - all of whom are very affordable and useful in DIY projects as well as "professional stuff" - I've seen quite a few of them in various products, anyway!  The shootout covers a classic oscillator application - a triangle/square output, two-opamp oscillator circuit.

You can read the article here - http://ee.david.promo/OpampShootout/ (no advertising or stuff; just too much to paste in here!)

The tested opamps are,
LM2902 (LM324, very similar to LM358) from two different manufacturers. And yes, the manufacturer seems to matter.
TL074 (similar to the TL06x, TL08x) - this particular part is from 1985. Older than me. Blimey!
LF347 (similar to the TL074, but higher bandwidth)
TL974 - a high-bandwidth (12-28MHz depending on supply voltage) low-noise bipolar amplifier.
TLV2372 - a rail-to-rail in/out CMOS op amp.

I was definitely surprised by the results - you might be, too.

TL;DR - the LF347, which is one of the cheapest opamps on test performs all-round best. The TL974 is fastest as expected, but with funny artifacts.
The TLV2372 disappoints, and the LM2902 is as horrible as you'd expect - but it differs between manufacturers.

I'd love to do more of this kind of stuff - so if you're interested in anything like this, say so!

[Zero999]

That shoot out only tests the slew rate of the op-amp, nothing else. It's good for comparing the speed but hopeless for things such as offset voltage, bias current, noise and distortion. It's virtually impossible to design a single circuit which can test everything simultaneously.

I was not surprised at all by the results. They can be easily estimated by looking at the figures given on the data sheets.

The square wave generator use the op-amp as a comparator, which is not how it's designed to be used. The slew rate of the op-amp will govern the rise/fall times of the square wave, which will limit the maximum usable frequency, in this circuit.

Different op-amps have different niches and applications. The TL974 may have come out on top here, but it would be no good in a low frequency circuit, where both inputs are expected to go to the negative rail. The LM2902 would be a much better candidate, which also has lower bias currents than the TL974.

[DrGonzoDK]

That shoot out only tests the slew rate of the op-amp, nothing else. It's good for comparing the speed but hopeless for things such as offset voltage, bias current, noise and distortion. It's virtually impossible to design a single circuit which can test everything simultaneously.

I was not surprised at all by the results. They can be easily estimated by looking at the figures given on the data sheets.

The square wave generator use the op-amp as a comparator, which is not how it's designed to be used. The slew rate of the op-amp will govern the rise/fall times of the square wave, which will limit the maximum usable frequency, in this circuit.

Different op-amps have different niches and applications. The TL974 may have come out on top here, but it would be no good in a low frequency circuit, where both inputs are expected to go to the negative rail. The LM2902 would be a much better candidate, which also has lower bias currents than the TL974.

...except that the frequency of op amps used as oscillators aren't just limited by their slew rate. This applies to sinusoidal as well as nonsinusoidal applications. The TL074 has a 50% advantage in slew rate over the LF347 (at these supply voltages), yet it performs somewhat worse in terms of actual stable oscillation speed.

The slew rate is one indicator, but it's not the truth gospel. And it can be difficult for a beginner to integrate the concepts of gain bandwidth product, slew rate, phase margin and gain margin for a specific application.

And I completely agree that the op-amps have different niches for their application use, even those that are labeled as "general purpose" (maybe even specifically those). That's why it's explicitly stated in the conclusion that,

The LF347 is about 0.2 euros (for 1 piece) from most suppliers, and it's an excellent op amp for this application. Doesn't mean it's great for other things - at low voltages, I'd use the TLV2372 any day; and if I needed precision and fast response, the TL974 would be my choice.

(In fact, the TL974 didn't come out on top because of other issues, for this application)

Rather, the hope was to contribute a little bit to helping people pick an appropriate part for this specific application. And yes, OPA1A is being used as a comparator in the circuit.

I do believe we can agree that there is no such thing as "the best opamp" though. Clearly. This is a biased test of a specific application (as is stated) and does not apply to other applications (the title of the article is Opamp Oscillator Shootout).

[Zero999]

That shoot out only tests the slew rate of the op-amp, nothing else. It's good for comparing the speed but hopeless for things such as offset voltage, bias current, noise and distortion. It's virtually impossible to design a single circuit which can test everything simultaneously.

I was not surprised at all by the results. They can be easily estimated by looking at the figures given on the data sheets.

The square wave generator use the op-amp as a comparator, which is not how it's designed to be used. The slew rate of the op-amp will govern the rise/fall times of the square wave, which will limit the maximum usable frequency, in this circuit.

Different op-amps have different niches and applications. The TL974 may have come out on top here, but it would be no good in a low frequency circuit, where both inputs are expected to go to the negative rail. The LM2902 would be a much better candidate, which also has lower bias currents than the TL974.

...except that the frequency of op amps used as oscillators aren't just limited by their slew rate. This applies to sinusoidal as well as nonsinusoidal applications. The TL074 has a 50% advantage in slew rate over the LF347 (at these supply voltages), yet it performs somewhat worse in terms of actual stable oscillation speed.

The slew rate is one indicator, but it's not the truth gospel. And it can be difficult for a beginner to integrate the concepts of gain bandwidth product, slew rate, phase margin and gain margin for a specific application.

I agree that slew rate isn't always the determining factor in oscillators but it appears to be in the circuit you've constructed. The propagation delay of the comparator may be a factor but, I think the slew rate is still dominant. The LF347 performing better, than the TL074 here, despite having an inferior slew rate specification, could easily be put down to variance. It would make more sense to compare several parts, from different manufactures.

You'll probably get better results in general, if you used a comparator, or, even better, a Schmitt trigger IC (try two sections of the CD4106, or 74HC14 in series) for OPA1.

And I completely agree that the op-amps have different niches for their application use, even those that are labeled as "general purpose" (maybe even specifically those). That's why it's explicitly stated in the conclusion that,

The LF347 is about 0.2 euros (for 1 piece) from most suppliers, and it's an excellent op amp for this application. Doesn't mean it's great for other things - at low voltages, I'd use the TLV2372 any day; and if I needed precision and fast response, the TL974 would be my choice.

(In fact, the TL974 didn't come out on top because of other issues, for this application)

Rather, the hope was to contribute a little bit to helping people pick an appropriate part for this specific application. And yes, OPA1A is being used as a comparator in the circuit.

I do believe we can agree that there is no such thing as "the best opamp" though. Clearly. This is a biased test of a specific application (as is stated) and does not apply to other applications (the title of the article is Opamp Oscillator Shootout).

How about performing a similar test with a Wein bridge oscillator, with a low output signal amplitude? Then see which op-amp does the best.

[DrGonzoDK]

That shoot out only tests the slew rate of the op-amp, nothing else. It's good for comparing the speed but hopeless for things such as offset voltage, bias current, noise and distortion. It's virtually impossible to design a single circuit which can test everything simultaneously.

I was not surprised at all by the results. They can be easily estimated by looking at the figures given on the data sheets.

The square wave generator use the op-amp as a comparator, which is not how it's designed to be used. The slew rate of the op-amp will govern the rise/fall times of the square wave, which will limit the maximum usable frequency, in this circuit.

Different op-amps have different niches and applications. The TL974 may have come out on top here, but it would be no good in a low frequency circuit, where both inputs are expected to go to the negative rail. The LM2902 would be a much better candidate, which also has lower bias currents than the TL974.

...except that the frequency of op amps used as oscillators aren't just limited by their slew rate. This applies to sinusoidal as well as nonsinusoidal applications. The TL074 has a 50% advantage in slew rate over the LF347 (at these supply voltages), yet it performs somewhat worse in terms of actual stable oscillation speed.

The slew rate is one indicator, but it's not the truth gospel. And it can be difficult for a beginner to integrate the concepts of gain bandwidth product, slew rate, phase margin and gain margin for a specific application.

I agree that slew rate isn't always the determining factor in oscillators but it appears to be in the circuit you've constructed. The propagation delay of the comparator may be a factor but, I think the slew rate is still dominant. The LF347 performing better, than the TL074 here, despite having an inferior slew rate specification, could easily be put down to variance. It would make more sense to compare several parts, from different manufactures.

You'll probably get better results in general, if you used a comparator, or, even better, a Schmitt trigger IC (try two sections of the CD4106, or 74HC14 in series) for OPA1.

And I completely agree that the op-amps have different niches for their application use, even those that are labeled as "general purpose" (maybe even specifically those). That's why it's explicitly stated in the conclusion that,

The LF347 is about 0.2 euros (for 1 piece) from most suppliers, and it's an excellent op amp for this application. Doesn't mean it's great for other things - at low voltages, I'd use the TLV2372 any day; and if I needed precision and fast response, the TL974 would be my choice.

(In fact, the TL974 didn't come out on top because of other issues, for this application)

Rather, the hope was to contribute a little bit to helping people pick an appropriate part for this specific application. And yes, OPA1A is being used as a comparator in the circuit.

I do believe we can agree that there is no such thing as "the best opamp" though. Clearly. This is a biased test of a specific application (as is stated) and does not apply to other applications (the title of the article is Opamp Oscillator Shootout).

How about performing a similar test with a Wein bridge oscillator, with a low output signal amplitude? Then see which op-amp does the best.

One thing I didn't mention - and should - in the article is that I tried several parts. Actually, 5 of each. Now, I couldn't find my USB flash key, otherwise I'd happily have added some fancy graphs of the averages. The fact is, though, that they were pretty consistent - except the LM2902 which had a pretty decent difference between the two manufacturers.

I'd love to add it with a Wien bridge or phase shift oscillator. test Maybe even the Bubba. The point is not to point out "the world's best opamp", rather to provide some concrete guidance to picking a part that isn't massively expensive for a particular job. :-) So another oscillator test would be a good thing!

Also, similar results were acheived with reduced gain (and voltage swing, ?v/?t, relative to voltage supply) in the circuit. Very important point, that one.

[DrGonzoDK]

Other interesting application tests would be Sallen/Key and Multiple Feedback-filters... Suggestions are welcome!

[rstofer]

My interest would be in the linearity of integration from -10V to +10V (+-15V supplies).  My only application for op amps is in the realm of analog computing and I have delusions of accuracy.  Low bias current and offset voltage seem like they would be importan given that I am using a 1 ufd capacitor and a 1 meg resistor for integration.