TI NE5532 Audio Opamp Changes

[SiliconWizard]

what TI thinks it stands to gain by doing it.

I think they think that 99% of engineers won't notice a difference

Which would be totally covered by saying “we are discontinuing the NE5532 and NE5532A. NE5532C is on a new process and is a drop-in replacement for typical applications. Please review your design if you rely on the following parameters which have changed: {list of changed specs}”. The 99% who don’t care happily buy the new chip and don’t review the design, and the remainder know what to check for. And most importantly, every TI NE5532 and 5532A would actually be an NE5532 or 5532A, rather than there being incompatible impostors in circulation.

Agreed. "Normal" vendors send PCNs, months in advance to their customers, when something like this happens. I wonder if anyone concerned has received any PCN for this change or for other TI chips for that matter.

I more or less understand what happened though at TI and that's kind of sneaky. They're right in their answer regarding the operating voltage range: they stated a max operating voltage of +/-15V in their NE5532 datasheets as early as  early 2000's, so basically 20 years ago. +/-22V was the absolute maximum and you should rarely design at the absolute maximum ratings unless you like playing with fire.

The sneaky part is that in those old datasheets, they advertised up to +/-20V on the front page while recommanding +/-15V max a couple pages later. And I'm talking about their 2004 datasheet revision.

As to nearly halving the slew rate, that's a bit drastic. The value in the 2004 datasheet was, as expected, 9V/µs.

[David Hess]

What irritates me about this entire situation is that I don’t see what TI thinks it stands to gain by doing it. In reality, they’re discontinuing old designs, which is sometimes annoying, but fundamentally fine. I see only downsides from introducing a new design under the old part number: angry customers, possibly lots of e-waste from nonfunctional circuits that were built with the new parts.

I mean, if customers have to re-qualify their designs, they may as well do so with a different part number. Whether that’s an NE5532C or OPA something or another, it’s the same process…

Polluting the stock of existing NE5532s may convince some customers to not use NE5532s at all, including the good ones from other manufacturers.  If Texas Instruments is not going to sell their bastardized ones, then nobody else should be selling good ones either.

Texas Instruments has a long history of datasheet shenanigans, including outright lies.

[exe]

About avoiding TI part, I'm not sure there are alternatives it comes to high-spec opamps.. IMO only two companies specialize on that: TI and AD. AD being even worse in terms of availability and price.

So, for generic parts an Asian opamp would be fine, but if you need precision, or high-end audio opamp, then it's TI (may be AD). Or if you need a detailed datasheet.

[Geoff-AU]

They're on my shit list at this point for future orders, I'd rather buy other brands than second guess bog standard parts. I can't imagine many people who make a living in electronics have time for shit like this.

TI have been on my spidey sense radar ever since one of their datasheets or app notes described an open drain with pullup as "totem pole". 

But no.. if you enshittify existing part numbers..  I have no time for that.

I prefer Analog at the moment because they gobbled up Linear Tech / LTSpice but I can divorce myself from them too if I need to..

Totally fine with process changes but you can't make your specs WORSE.

[tooki]

what TI thinks it stands to gain by doing it.

I think they think that 99% of engineers won't notice a difference

Which would be totally covered by saying “we are discontinuing the NE5532 and NE5532A. NE5532C is on a new process and is a drop-in replacement for typical applications. Please review your design if you rely on the following parameters which have changed: {list of changed specs}”. The 99% who don’t care happily buy the new chip and don’t review the design, and the remainder know what to check for. And most importantly, every TI NE5532 and 5532A would actually be an NE5532 or 5532A, rather than there being incompatible impostors in circulation.

Agreed. "Normal" vendors send PCNs, months in advance to their customers, when something like this happens. I wonder if anyone concerned has received any PCN for this change or for other TI chips for that matter.

They released the PCN in late 2023. I see two issues: IMHO, changing fundamental product specs far exceeds a “process change”, which is what the PCN basically says it is. And even worse, they didn’t update the datasheet until nearly two years later. THAT is inexcusable if you ask me.

[Kleinstein]

Changing a part that much and still call it NE5532 is crazy. This is especially as Ti is only a 2nd source for the NE5532.
There new chip may be an OK replacement in most cases, but it is no longer a NE5532 it is more a RC4580 derived part (if not just different testing).

It would have been acceptable if they EOL there NE5532 (as that process may run out or gets too expensive for a low cost part) and link to there replacement part(s). The new replacement than should have really got a different name. If there is really demand (e.g. ON also dropping it) they could still have restarted or keep the part at a higher price (e.g. for repairs and odd special cases that really want the same part).

They did a similar mess with the TL072H and somehow did not learn the lesson.
The new datasheet that covers 2 chip versions gets rather confusing and error prone. All the typical graphs would only apply to one of the versions.
They still have to somehow keep the old data so that the info on the old chips stays around.

[tom66]

TI's PCN for NE5532 only mentions that the fab site changed and the die has changed as a result - it doesn't mention any characteristic changes.  In fact, it explicitly states:

Anticipated impact on Form, Fit, Function, Quality or Reliability (positive / negative):
None

https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/5726/PCN20231114002.1.pdf

It'd be very easy to see someone in production engineering not realising that something significant has changed with this part and it being missed.   

[iMo]

Imagine the chip vendors in current geopolitical and economic situation. There is perhaps XX hundreds of opamp types fabricated in last 60years. The vendors usually bake X million dies in one shot and slowly selling them out. Any die changes are very expensive. To bake the same chip dies with time is also much more expensive. Also the market for some analogue and digital parts is shrinking because there are none products requiring XXX thousand chip orders anymore. If I were the ADI's or TI's boss I would have a pretty short hand in EOLing dozens of types.
For us here it is not a good message, but producing the same dies for centuries is not feasible, imho (unless Elon builds up the chip foundries in the space while using the material off the asteroids, powered by Sun, operating by AI fully autonomously, receiving orders via Starlink, tapes and reels with chips delivered via StarshipCargo).. 

[Alex Nikitin]

There is no problem in an obsolescence per se, if properly understood and controlled, it is a fact of life and happens all the time. It is TI brazen lies about it which cost other people money that are completely unacceptable. And as I’ve said before, the result is a complete loss of trust in TI as a supplier. I, for one, would now avoid any TI parts manufactured over the last five years (and will advise my customers to do the same).

Cheers

Alex

[iMo]

Thus my free of charge advice to ADI and TI: do not confuse customers - write in clear text:

"This XYZ product has ended up and we do not sell nor produce or plan to produce the original dies or parts with the parameters as stated in the DS XYZ from 1992.
Basta."

or something like that.. 

PS: "Managing customer expectations means communicating honestly and transparently about what your business can realistically deliver. By setting clear boundaries and proactively sharing potential limitations or delays, you can prevent misunderstandings, build trust, and turn angry complaints into customer loyalty."
From google..

[MT]

They did a similar mess with the TL072H and somehow did not learn the lesson.
The new datasheet that covers 2 chip versions gets rather confusing and error prone. All the typical graphs would only apply to one of the versions.
They still have to somehow keep the old data so that the info on the old chips stays around.

Not just TL072/74 but also 082/84 and 064 but not for 062 then there is just a note about a new better device 072H... but the main point of the 062/064 the very low power consumption
for every one single 072H opamp five 062 can be driven.

[EEVblog]

I more or less understand what happened though at TI and that's kind of sneaky. They're right in their answer regarding the operating voltage range: they stated a max operating voltage of +/-15V in their NE5532 datasheets as early as  early 2000's, so basically 20 years ago. +/-22V was the absolute maximum and you should rarely design at the absolute maximum ratings unless you like playing with fire.
The sneaky part is that in those old datasheets, they advertised up to +/-20V on the front page while recommanding +/-15V max a couple pages later. And I'm talking about their 2004 datasheet revision.

Here is a 1990 datasheet saying +/-20V in the banner headline spec.
But +/-15V max recommended in the spec table.
https://img.gme.cz/files/eshop_data/eshop_data/3/310-024/dsh.310-024.1.pdf


Rev J from 2015 is here:
https://www.mouser.com/datasheet/2/405/ne5532a-556776.pdf
The Absolute Maximum Ratings has indeed been changed from +/-22V to+/-18V
Onsemi, Fairchild, NXP, and a bunch of adian brands are all +/-22V
I found one asian brand that is +/-18V max though:
https://www.lcsc.com/datasheet/C48677906.pdf

[pope]

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

[EEVblog]

I'm especially impressed by removal of the offset adjustment pins on the OPA134 (and a simultaneous increase in the typ/max input offset values in the datasheet ! )  .

WTF.

[EEVblog]

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

The change note does not mention the 5534, and I checked and there doesn't seem to be a new version datasheet with any changes since 2014

[tom66]

Seems a lot of this comes down to TI shutting down the SFAB and moving production to RFAB, which has different process technology.  It might have been rushed, and they had to respin chips quickly, without enough time to optimise them for the new process resulting in parts with noticeably different performance.

I am not an analog IC expert, but I imagine analog ICs require a few more rolls of the dice before performance is optimised.  Analog simulation can get you quite far, but nothing compares to testing the real chip.

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

I don't think so.  It's possible this was already set up at an existing fab under TI, or subcontracted. Datasheet hasn't been revised since 2014.

[Haenk]

I wonder if the same happened to the 4562 - got them from a chinese vendor advertised as "new start" (which probably was mistranslated from "new production") about 1.5 years ago...

[Alex Nikitin]

IMHO, to talk about parameter matching or production difficulties or device features is to completely miss the point. And the point is actually very simple: "Thou shalt not bear false witness to the device thou makes" .

Cheers

Alex

[pope]

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

The change note does not mention the 5534, and I checked and there doesn't seem to be a new version datasheet with any changes since 2014

TBH, I'd be surprised if they haven't changed the 5534 from 150mm to 300mm. I developed trust issues with TI so the fact that the datasheet isn't updated, means nothing to me.

The 5534s went EOL a few years back and after a year or two they magically reappeared. Pretty suspicious.

[Simmed]

i remembered JRC
so it looks like JRC/nissinbo is still using 22v on their version

https://www.nisshinbo-microdevices.co.jp/en/products/operational-amplifier/spec/?product=njm5532c

https://www.lcsc.com/category/1441.html?scene=FULL_MATCH&globalKeyword=njm553&s_z=n_q_njm553

lets hope they dont update their die revision 

[SiliconWizard]

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

The change note does not mention the 5534, and I checked and there doesn't seem to be a new version datasheet with any changes since 2014

Haven't used these parts in years but from memory, I think the 5532 was actually a slightly different design and not just a dual version of the 5534. IIRC (but don't quote me on that), the 5532 had worse noise figures.

[Simmed]

Does anybody know whether the new 5534s are also f*cked up? I'm pretty sure the answer is "yes" but I'm curious if anyone has any at hand.

The change note does not mention the 5534, and I checked and there doesn't seem to be a new version datasheet with any changes since 2014

Haven't used these parts in years but from memory, I think the 5532 was actually a slightly different design and not just a dual version of the 5534. IIRC (but don't quote me on that), the 5532 had worse noise figures.

yes you are right 5532 5nv/hz vs 5534 3.3nv/hz

[Hydron]

@EEVblog - if you want any more content along these lines for a potential video, then search the forum for the issues some of us have had with LMH6518; while not a fab change, TI/NS royally screwed up the datasheet and omitted a critical maximum rating, then pissed about for years before revising it and clarifying the issue. One consequence is that a number of us have had our RTB2k scope frontends (which use this chip) die when operating the scope well within advertised specs - I've had it happen twice, and R&S quoted a silly price the first time (and now probably won't even deal with me now with their new anti-private-buyer policies) so I've had to replace the chip myself both times.

[floobydust]

I think this thread is about TI's corporate malfeasance that we all are finding unacceptable.
What happens is executives and employees worship the scorecard and it becomes a bad corporate culture. It always gets worse with time until the plane crashes.

Cost improvement doing a die-shrink results in weaker power transistors - reduced SOA and less heat transfer to the frame.
It shows up in the NE5532's weaker output stage that TI puts the "lipstick on a pig" by deleting any such revealing specs. 600 ohm drive is a basic requirement in pro-audio for many decades.
It shows up in the LM317M SOT-223 "new chip" thermal resistance 30% worse than the legacy part. Maybe they even saved a milligram of metal. You won't get "900mA" for long, test unknown, probably "new chip" does it for only a few usec lol.
Die-shrink surely the comp caps are smaller and challenge stability as well.

I have to swear here, the OPA134 is fuckin' over 15x the price of the NE5532! Overpriced that cow did you?
Nevermind TI's bastardization and ruining of the Burr-Brown designs, it's a travesty taking their innovations and enshittifying them, all the while jacking up the prices.

[factory]

wow
this should be a new model
like NE5h1t

Hitachi almost did that for a batchcode (5H1 T). 

David