TI NE5532 Audio Opamp Changes

[tom66]

However, they do at least seem to have covered the changes under a new part number (old part OPA627AM, new part OPA627AU/OPA627BU)

I didn't spot that. In my defence it is only a subtly different part number.
Letters on the end usually indicate temperature ranges or packaging options, not a completely different class of opamp.

Personally I am less worried about that because I know our CEMs and purchasing department will always flag up a warning if a part number changes like that.  Sometimes I get queries over parts where the only difference is tape and reel vs tray packaging for instance.  There are fairly sophisticated software packages that the CEMs use that will pull specs in from databases and compare them.

What TI have done with the 5532 is malpractice IMO because it won't flag up on one of these checks; many parts will be installed on boards without this being noticed.  In some cases problems will only become noticeable once they end up at the customer's premises.  This is a problem because it's a lot more expensive to correct then.  The saying I've heard (might be from Toyota) is that a problem at the assembly line costs $1 to fix, at final test $10, at the dealership $100, and once it has been delivered to the customer and a recall is needed, it'll be $1000+.  TI has essentially pushed these problems to the customer/dealership level. 

It also impacts reputation.  Imagine if you are a recording studio* and just bought an audio deck and suddenly it isn't working so well.  It has to be returned to the factory and repaired - all the amplifier chips have to be replaced.  The unit is returned a month later.  You lose the use of that studio and you will be reluctant to consider using that company in the future.  Ok, you could argue the manufacturer should have tested things properly, but sometimes these tests are infeasible to do, you have to assume that the design as-manufactured is good if it passed your tests at the design stage.

*Hypothetical example, perhaps NE5532 wouldn't be found in such a class of device.

[floobydust]

Acquire, butcher, cheapen and increase prices. Beef up the disclaimer! Profit is paramount!
I fully expect TI to butcher Burr Brown's state of the art products.

Hey TI Change Management Dept. : "Meet or exceed" is the rule, not less for more. Or lying about it: "Anticipated impact on Form, Fit, Function, Quality or Reliability (positive / negative). NONE" 
You need to stop that shit instead of placating the scorecard.

OPA627 goin' for almost USD $50... many changes with the die redesign. I think change from Semefab Scotland 100MM DI-452 process to Freising Germany FFAB 200mm.

Oh look another 600 ohm "issue" get the broom sweep under the carpet lol.
Offset pins we can't do that so justify the DELETE by saying "you don't need that anymore". Why was it there in the first place lol.

"In the new FAB, the trim features (all package types) have been removed due to the improve process, though these are still considered as General Purpose Op Amp. Some package still show the trim pin description due to the delay in datasheet updating process. The trim feature is no longer used in the new FAB design/process."
https://e2e.ti.com/support/audio-group/audio/f/audio-forum/1517005/opa132-changed-opa132-pin-1-and-pin-8-from-offset-trim-to-nc

"The comment that the input offset voltage trim pins were removed because the device is laser-trimmed is not accurate for this device.  For other revised devices this will be true as the offset is reduced to a point where the trim pins are really unnecessary.  For this device, however, the intent is for audio applications so we do not trim the input offset voltage.  The device is trimmed for other parameters, but the specific details on what is and is not trimmed is proprietary.  For many die revisions the external Vos trim pins and associated internal circuitry are removed because most customers no longer use this feature.  That is, manually adjusting an external potentiometer is not cost effective when compared with just purchasing a low offset device.  Sorry for the inconvenience, but this device and many modern amplifiers no longer include this option."
https://e2e.ti.com/support/audio-group/audio/f/audio-forum/1589619/opa134-difference-in-laser-trim-between-rev-a-and-rev-b

[Kleinstein]

The OPA627 (and some related and similar expensive) uses a rather special silicon on isolator process. This was an expensive part from the start and is now still available for some special customers in low quantity. It is perfectly normal that it is now rather expensive. It always was a premium part.
Given the low volume they may just still have old wafers with enough parts for decades.

There are now alternatives like OPA827/OPA828 - still SOI, but a more modern, cheaper process. Still the new ones miss the trim option. At least they got a new name and thus are allowed to be different.

There are still a few cases where the offst trim pins are used for more than just classic offset trim, especially in comboned circuits with a 2nd op-amp to corret the error.

[Alex Nikitin]

Noticeable improvements there, but there are several areas it is a worse device:

- Common mode range used to be Vss-2V to Vcc+2V, now it's just 0.5V either side of the supply
- Differential mode range used to be Vs+4V, now it's just Vs
- The current output used to be +/-45mA, now it's +/-30mA
- Input bias currrent is now bipolar (if I'm interpreting the change correctly)

However, they do at least seem to have covered the changes under a new part number (old part OPA627AM, new part OPA627AU/OPA627BU)

These are not "improvements", it is again  a completely different device under the old device name (OPA627AU / OPA637AU is an old device marking as well).

Cheers

Alex

[mediatechnology]

They are using RC4580 dies, not MC33078.

Has that been officially confirmed anywhere, either by TI or by someone opening one of the current devices and inspecting the insides? I have some completed products on hand some of which have "real" NE5532A in them, some have new NE5532A. Based on the offset voltage of the new parts, which is much better than the either the NE5532 or RC4580 spec, they seem more like MC33078, behaviour wise, but I have not yet had a chance to do much extensive testing to compare them in other ways, I just happened to notice that when production changed to the new parts, offset improved, and my guess was that since both the 5532 and 4580 have had datasheet updates whilst the 33078 has not, that would more likely mean that the dies in those parts have changed whilst what's in the 33078 has stayed the same. Of course, its also possible they just haven't gotten around to updating the 33078 datasheet yet.

I do recall TI confirming that they are identical die.
EDIT: https://e2e.ti.com/support/audio-group/audio/f/audio-forum/1652528/ne5532-ne5532-lm833-rc4580-and-mc33078-all-now-the-same-die

Here is a relative FFT comparison of the error voltage of an NJM5532 to a TI NE5532 and the new TI BS5532:
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=20694#p20694

This is a comparison of the BS5532 to an RC4580. They have identical signatures:
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=20700#p20700

Test conditions in the thread but basically +20 dBu at 15 kHz into 604R||10K (105 mW). Supplies +/-15V. Inverting gain of +20 dB.
Error voltage of the op amp A inverting input has another 20 dB of non-inverting gain using the second half of the DUT. The DUT is loaded only by the non-inverting input of op amp B.
The NJM5532 is the clear winner, the NE5532 a close second and the 4580/BS5532 tied for dead last.

I will eventually do a non-inverting null test - the inverting configuration is advantageous to all of the devices. A non-inverting follower test will reveal CM differences.

[David Hess]

...

There are now alternatives like OPA827/OPA828 - still SOI, but a more modern, cheaper process. Still the new ones miss the trim option. At least they got a new name and thus are allowed to be different.

There are still a few cases where the offst trim pins are used for more than just classic offset trim, especially in comboned circuits with a 2nd op-amp to corret the error.

I wish they would note in selection guides which operational amplifiers have offset null as a feature.  As it is, I have to read every datasheet, and now some of the datasheets lie anyway.

The same applies to external compensation, external overcompensation, and clamping.

[mediatechnology]

The OPA627 (and some related and similar expensive) uses a rather special silicon on isolator process. This was an expensive part from the start and is now still available for some special customers in low quantity. It is perfectly normal that it is now rather expensive. It always was a premium part.
Given the low volume they may just still have old wafers with enough parts for decades.

There are now alternatives like OPA827/OPA828 - still SOI, but a more modern, cheaper process. Still the new ones miss the trim option. At least they got a new name and thus are allowed to be different.

There are still a few cases where the offst trim pins are used for more than just classic offset trim, especially in comboned circuits with a 2nd op-amp to corret the error.

Dielectric isolation is a very messy and expensive process that requires grinding the back side if the wafer to expose the bottom of the isolating tubs. I've seen it done and a very wet slurry from substrate is produced which is the complete opposite of that found in a clean room environment. This manufacturer did the intermediate lapping process at a completely different site from fab for isolation and used a lot of water getting them clean before shipping them back for final processing. This manufacturer is still doing boutique DI products on a 4" line. Large 300 mm wafers don't lend themselves to DI due to warping. The OPA627 may still be being made on the only remaining 200 mm line.

I have a couple of examples of 5534 offset pins being used to attach external bipolar and FET input stages. They can be used for purposes other than physical trim pots something the current generation of IC designers may have never seen.

[David Hess]

Dielectric isolation is a very messy and expensive process that requires grinding the back side if the wafer to expose the bottom of the isolating tubs. I've seen it done and a very wet slurry from substrate is produced which is the complete opposite of that found in a clean room environment. This manufacturer did the intermediate lapping process at a completely different site from fab for isolation and used a lot of water getting them clean before shipping them back for final processing. This manufacturer is still doing boutique DI products on a 4" line. Large 300 mm wafers don't lend themselves to DI due to warping. The OPA627 may still be being made on the only remaining 200 mm line.

Isn't there a more modern way where the wafer is somehow sheared between the silicon and implanted silicon oxide layers?

I have a couple of examples of 5534 offset pins being used to attach external bipolar and FET input stages. They can be used for purposes other than physical trim pots something the current generation of IC designers may have never seen.

The oldest example I know of where that was done was with the LM318.  A different example is shown below.

[G0HZU]

Is there a definite way to buy the new (revised) Ti 5532 from Mouser?
There are NE5532P, NE5532PSR and NE5532PE4 Ti parts available from Mouser. The PE4 part looks to be EOL so I assume this is an older version.

I have a setup here that can measure the noise figure of amplifiers (across various source impedances) so I could buy the new Ti version of the NE5532 compare against a known good (classic) NE5532. I have some really old NE5532 parts here from Signetics for example. My setup would be as in the lower image below where the test fixture feeds to a low noise analyser that has a 50 ohm input.

I can also test the voltage noise of most opamps using the other two (simpler) test setups in the other image below. I prefer to use the one with the 120R feedback resistors as it reduces the risk of damage to the analyser even though I have protection diodes included. The 120R resistors add some extra noise, but not much.

[DanielSkevington]

They are using RC4580 dies, not MC33078.

Has that been officially confirmed anywhere, either by TI or by someone opening one of the current devices and inspecting the insides? I have some completed products on hand some of which have "real" NE5532A in them, some have new NE5532A. Based on the offset voltage of the new parts, which is much better than the either the NE5532 or RC4580 spec, they seem more like MC33078, behaviour wise, but I have not yet had a chance to do much extensive testing to compare them in other ways, I just happened to notice that when production changed to the new parts, offset improved, and my guess was that since both the 5532 and 4580 have had datasheet updates whilst the 33078 has not, that would more likely mean that the dies in those parts have changed whilst what's in the 33078 has stayed the same. Of course, its also possible they just haven't gotten around to updating the 33078 datasheet yet.

I do recall TI confirming that they are identical die.
EDIT: https://e2e.ti.com/support/audio-group/audio/f/audio-forum/1652528/ne5532-ne5532-lm833-rc4580-and-mc33078-all-now-the-same-die

I am aware of that. It is not an RC4580 die though - the decapped chip pictured up thread confirms that whats inside is a part labelled "6113" which as far as I can make out doesn't seem to be a device in its own right - I guess some kind of bipolar op amp "master part" that they can bin up and package as multiple other parts as required. My point is I don't think it's really accurate to say that the current 5532 is a 4580, although the design is closer to a 4580 or 33078 than a 5532, they are all really something else.

[Yansi]

Do you know what is inside the currently produced 4580? Comparing with old die pictures of 4580 does not make sense now.

[DanielSkevington]

Do you know what is inside the currently produced 4580? Comparing with old die pictures of 4580 does not make sense now.

According to TI, current 4580s have the same "6113" die inside.

[squadchannel]

Is there a definite way to buy the new (revised) Ti 5532 from Mouser?

It's best to create a TI account and buy directly.

But I think the current stock on the market has already been replaced the BS5532.
Why don't you ask when that stock arrived?

Do you know what is inside the currently produced 4580?

@Noopy

[mediatechnology]

Is there a definite way to buy the new (revised) Ti 5532 from Mouser?
There are NE5532P, NE5532PSR and NE5532PE4 Ti parts available from Mouser. The PE4 part looks to be EOL so I assume this is an older version.

I have a setup here that can measure the noise figure of amplifiers (across various source impedances) so I could buy the new Ti version of the NE5532 compare against a known good (classic) NE5532. I have some really old NE5532 parts here from Signetics for example. My setup would be as in the lower image below where the test fixture feeds to a low noise analyser that has a 50 ohm input.

I can also test the voltage noise of most opamps using the other two (simpler) test setups in the other image below. I prefer to use the one with the 120R feedback resistors as it reduces the risk of damage to the analyser even though I have protection diodes included. The 120R resistors add some extra noise, but not much.

I've started a registry of 5532 devices from Mouser and 6 days ago the NE5532A was the "new" part.
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1592

Not all of the NE5532 from Mouser are new stock as of 6/15/26. In fact 3 of the 4 devices were not BS5532s. That obviously can change quickly.

IMHO the most revealing test is HF distortion and this thread shows 5532/BS5532 comparisons. https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=20694#p20694

Two posts down for the one linked above you'll see a 4580/BS5532 comparison with almost-identical distortion (error voltage) signatures.

I'm not sure it matters if the 4580/BS5532 are the same die or not - despite TI's admission that they are - because both parts suck equally when actually measured.
I'm done with conjecture.

[mediatechnology]

Here I compare the IM of the error voltage for the RC4580, BS5532, NE5532 (150mm), NJM5532 and the XR5532.
https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=20706#p20706

Once again the IM signature for the RC4580 and BS5532 are identical.
I think, based on actual tests, we should take TI on their word that they're the same die and move on.

The NJM5532 still outperforms the 150 mm NE5532.

The Exar 5532 is king daddy and in a league of its own.

[tooki]

Once again the IM signature for the RC4580 and BS5532 are identical.
I think, based on actual tests, we should take TI on their word that they're the same die and move on.

Nobody was disputing TI’s claim that the new RC4580 and new NE5532 are based on the same die.

The unanswered question is “are the BS5532, BS4580 (and others) the same as the old RC4580 die”, which is what had been suggested. I kinda doubt it, but we don’t have a conclusive answer.

[mediatechnology]

Once again the IM signature for the RC4580 and BS5532 are identical.
I think, based on actual tests, we should take TI on their word that they're the same die and move on.

Nobody was disputing TI’s claim that the new RC4580 and new NE5532 are based on the same die.

The unanswered question is “are the BS5532, BS4580 (and others) the same as the old RC4580 die”, which is what had been suggested. I kinda doubt it, but we don’t have a conclusive answer.

According to this PCN the 4580 (et al) die has changed as a result of the move to 300mm and the TIB process: https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/5726/PCN20231114002.1.pdf

In the middle of page 1:

The die was also changed as a result of the process change.

The 4580 is in Group 1 on page 2.