Amazing opamp measurement shoot-out

[KSTR]

You guys should not miss this:

http://recforums.prosoundweb.com/index.php/t/24704/0/

- Klaus

 

[Juergen Knoop]

thank you!
I remember Ben Duncan? or someone else publishing opamp distortion specs in EWW in the 90's .
But not at this detail level.
regards

 

[Salas]

Excellent. Long time needed.

 

[syn08]

Amazing work and results. I hope the author will keep up with more opamps (there are quite some [worth evaluating] devices, in particular from AD and TI).

Nothing really surprising, only the LT1115/LT1128 are dissapointing and somehow conflict with my own measurements (but then I did not evaluate the common mode linearity and the input impedance linearity).

I also hope the discrete opamp manufacturers will send samples for evaluation (to help answer the third question) but somehow I doubt they will...

Anybody asking "but how does it sound?".

 

[andy_c]

That is a great article. I think congratulations to Scott are in order.

 

[Wavebourn]

Unfortunately Samuel did not measure couple of my favorite opamps, though I've asked him...

 

[Bonsai]

Great Paper with very interesting commentary.

Interesting how quickly things degrade once you start to drive heavy loads and/or high output levels on most of the devices..

I might be wrong, but it looks like the AD749 comes out tops overall and the LME4xxx from Natsemi a creditable 2nd place.

The author also commented that extra care is needed on the AD device. I think th e NS device is a bit easier to use, so thats what its got in its favour.

As mentioned alreadt, congratulations are in order to Scott Wurcer!

 

[syn08]

I might be wrong, but it looks like the AD749 comes out tops overall and the LME4xxx from Natsemi a creditable 2nd place.

There are a few very strong contenders for the first place that were not yet evaluated:

OPA211, THS4031, OPA827, OPA2822, OPA842 from TI
ADA4899-1, AD823, AD8066, AD8610, AD8599 from Analog Devices
EL2227 from Signetics/Elantec

...and probably more...

 

[h_a]

Excellent piece for reading - even for non IC-guys!

Have fun, Hannes

 

[lumanauw]

Why is it needed to use topology that give noise gain 60dB more than signal gain?

 

[andy_c]

This reduces the feedback, which increases the distortion so that it's above the residual of the test equipment.

 

[lumanauw]

Since the noise is magnified 60dB, will it interfere with the measurement, like in FFT?

 

[forr]

I was interested by the bootstrapped current mirror in the one stage topology mentionned in appendix B. Are there discrete implementations of it ?

 

[andy_c]

Since the noise is magnified 60dB, will it interfere with the measurement, like in FFT?

Well, the noise will get worse, but according to his text, he imports the raw data into the MATLAB program and does a bunch of averaging and filtering to reduce it. I gather the noise is less of a problem than the residual distortion of the AP System 1.

 

[Wavebourn]

There are a few very strong contenders for the first place that were not yet evaluated:

OPA211, THS4031, OPA827, OPA2822, OPA842 from TI
ADA4899-1, AD823, AD8066, AD8610, AD8599 from Analog Devices
EL2227 from Signetics/Elantec

...and probably more...

...5562, CA3140. The last one has an excellent for audio output stage, but noisy input.

 

[syn08]

...5562, CA3140. The last one has an excellent for audio output stage, but noisy input.

CA3140 - I have evaluated a few samples from Signetics and I was not happy with. The output stage might be very good, but it's a slow opamp, to slow to avoid slew rate induced distortions.

5562 - What is this?

 

[Wavebourn]

CA3140 - I have evaluated a few samples from Signetics and I was not happy with. The output stage might be very good, but it's a slow opamp, to slow to avoid slew rate induced distortions.

In applications where I use them 15V/uS is enough.

5562 - What is this?

Oops... Sorry for the typo, I mean 4562

 

[syn08]

In applications where I use them 15V/uS is enough.



Oops... Sorry for the typo, I mean 4562

LM4562 is (AC) identical to the already tested LME49860. The slightly higher supply voltage allowed by the LME49860 is most likely the result of sorting the chips.

 

[scott wurcer]

This explains a call I got the other day. Did anyone notice the unbiased and unbypassed 6800uF caps in the test circuits?

 

[lumanauw]

Did anyone notice the unbiased and unbypassed 6800uF caps in the test circuits?

Unbiased, do you mean the + of the 6800uF should be more positive than the - of the 6800uF? How can we do it in a simple cct?