Open loop gain measurements of NE5534 / NE5532 / LM318 from various manufacturers

One thing, if you compare samples from different manufacturers, then you should first do tests from one manufacturer.
of course, the are no significant difference

Interesting. All 5534 samples are within the tolerances.
no, the are no unity gain bandwith(unity gain frequency) in the datasheet and associated tolerances, the same with AC gain tolerances (only in 1034 but with correction capacitor 22pF and load capacitor 100pF but without tolerances).

5534(and many others opamps) from different manufacturers are in fact different opamps, philips has ugb=31.8MHz and phase margin 47deg, onsemi has ugb=14.6MHz and phase margin 70deg, and can be used with closed loop gain as low as +1.
 
Last edited:
AX tech editor
Joined 2002
Paid Member
This is very nice information, thanks for that.
It is nice to see the gain for the compensated 5532 being about 6dB less than the de-compensated 5534 types.
Am I right that the top Y-axis is 60dB, with 10dB/div? Not used to see graphs with no axis labeling.

A question: is the info also available for lower (audio) frequencies? It would be nice to see the ol bandwidth.

Jan
 
Last edited:
  • Like
Reactions: 1 user
Account Closed
Joined 2010
Interesting. All 5534 samples are within the tolerances. One thing, if you compare samples from different manufacturers, then you should first do tests from one manufacturer. Because the chances are hight, that the tolerances are the same as comparing different manufacturers.
This would only outline once more that the manufacturers do have high standards.
Randomly choosing only one sample from each manufacturer, if all the manufacturers show similar measurements give an automatic feedback on all the manufacturers standards .
It's highly improbable to be just luck that one manufacturer's sample was the only good one in all their samples if it was randomly chosen.
If you do 10 samples from one manufacturer you need to check ten samples from each manufacturer and the result will be simply similar unless the probability rules don't work anymore with great numbers.


You have only one exception to the rule that might be applied here:
The older designs of ne5534/32 like Philips or Signetics were already checked for noise specs and batched accordingly before being sold, but at the same time the technology made possible for larger numbers with similar specs to be manufactured in the same amount of time so 1000 hand picked op-amps from the past might have the same tolerances as 100 000 randomly chosen op-amps of today .
 
Last edited:
Thanks for that information - very interesting if even being within spec.

If I remember correctly, Philips and Signetics shared the same manufacturing process etc for their 5532/4s and were often favoured above the likes of TI.

I use an MCI recording/mixing console full of 'House numbered' 5532s and 5534s which are, in fact, cherry-picked Signetics devices chosen by the manufacturer for being very tightly within spec., particularly their noise spec.

Cheers

Mike
 

Interesting. I learned from a reputable source that CMRR was one of the criteria that Analog Systems used to select their opamps back in the 80's. Since then I've been using CMRR test to select opamps, and find it the best single qualitative indicator for non-inverting and diff-amp performance. I see much greater unit-to-unit variation across the audio freq range than your test is showing, with some exceptions. Of the 'oldies', the Signetics and Raytheon 5532 were the most consistent. Some batches being really excellent (CMRR -100dB and flat beyond 20kHz). The TI 5532 sucked. One of the best, but very inconsistent batch-to-batch, was Nat Semi LF351/353 and LF411/412. Some batches were -120 to -130dB!

Manufacturing consistency has improved greatly since the 80's. The LM4562 and OPA1642 CMRR equal or beat the best of the old 5532's every time. I've never seen a dog of either one.
 
AX tech editor
Joined 2002
Paid Member
I learned from a reputable source that CMRR was one of the criteria that Analog Systems used to select their opamps back in the 80's.

Apart from whether this is popular lore or not, it doesn't seem too smart.
In an inverting stage there is no CM signal so in these cases the opamp would be selected on a performance point that is irrelevant.

Jan
 
Apart from whether this is popular lore or not, it doesn't seem too smart.

Not just 'lore'... Deane Jensen was in a position to have direct knowledge.

In an inverting stage there is no CM signal so in these cases the opamp would be selected on a performance point that is irrelevant.

Perhaps you didn't read, I said "non-inverting and diff-amp". Of course it makes no difference for inverting. But if you needed SOTA performance at that time for a non-inverting application (not just audio), then the Analog Systems selected IC's were the bee's knees. My Anritsu MS-420K network analyzer is full of 'em. Amber distortion analyzers used them in key places.
 
Since then I've been using CMRR test to select opamps, and find it the best single qualitative indicator for non-inverting and diff-amp performance.
CMRR number has nothing to do with non-inverting distortion performance.
For such purposes you need to measure "CMRR nonlinearity" or in another words you must measure and plot Offset vs Input CM voltage.
 
There seems to be an epidemic of reading disability today.
I didn't say anything about "non-inverting distortion performance".
Nor did I say it was the sole indicator of performance.
All I said was, I find it the best single qualitative indicator. I have never found an instance of an opamp that measures CMRR vs freq poorly and sounds good in non-inverting or diff use.
Sheesh.
 
AX tech editor
Joined 2002
Paid Member
Not just 'lore'... Deane Jensen was in a position to have direct knowledge.



Perhaps you didn't read, I said "non-inverting and diff-amp". Of course it makes no difference for inverting. But if you needed SOTA performance at that time for a non-inverting application (not just audio), then the Analog Systems selected IC's were the bee's knees. My Anritsu MS-420K network analyzer is full of 'em. Amber distortion analyzers used them in key places.

OK. Point taken.

Jan