TI NE5532 Audio Opamp Changes

[tom66]

The problem with the 5532 at TI could become an opportunity for other smaller vendors, imho.. There is around 100 foundries around the world, afaik, most of them with older nodes. So - would it be a large problem for a smaller talented vendor to produce the 5532 within all that original specs? With even 3-5x the price for such a unique opamp people will buy it, sure (as all the discussions show)..

They might struggle with the scale.  It's likely a chip like that is sold in the 10's to 100's of millions per year.

[Kleinstein]

The problem with the 5532 at TI could become an opportunity for other smaller vendors, imho.. There is around 100 foundries around the world, afaik, most of them with older nodes. So - would it be a large problem for a smaller talented vendor to produce the 5532 within all that original specs? With even 3-5x the price for such a unique opamp people will buy it, sure (as all the discussions show)..

They might struggle with the scale.  It's likely a chip like that is sold in the 10's to 100's of millions per year.

The NE5532 is still made by OnSemi and a few others. Ti was just one of many manufacturers to make the 5532. So don't expect the price for the real 5532 to go up. The Ti part may get a bit cheaper.

[iMo]

I still think there is a niche market for a talented vendor (perhaps fabless one) producing those various oldtimers in lower quantities. With modern design tools on an older production hw in many foundries around.
Imho with $200-300k you may produce a complete design and set of masks for the older nodes, imho. Producing a quantity of even say 100k chips for say $5-6 each may satisfy their ROI fast then..

[Yansi]

With that kind of price, you can not build products that normally use it. Imagine how many of those sit even in basic A/D - D/A interface, let alone even small audio mixing desk.

//EDIT: and also, for that price, one may get a very different solution alltogether, with much better resulting spec. For that A/D - D/A example above, one may use for example much more expensive OPA1632, fully dif. opamps.  I just hope, those won't get enshitified too.

[David Hess]

I still think there is a niche market for a talented vendor (perhaps fabless one) producing those various oldtimers in lower quantities. With modern design tools on an older production hw in many foundries around.
Imho with $200-300k you may produce a complete design and set of masks for the older nodes, imho. Producing a quantity of even say 100k chips for say $5-6 each may satisfy their ROI fast then..

The larger problem is lack of consumables for older processes, and once the market for an old process gets down to one supplier, a competitor can buy that supplier and cut off the remaining fabs.  The notable legacy fab failures all seem to follow this pattern.

Also, testing is not an insignificant cost.

Some analog companies are fabless, but their process choices are very limited.

[tooki]

I still think there is a niche market for a talented vendor (perhaps fabless one) producing those various oldtimers in lower quantities.

Ummmm… did you miss the fact that beside TI, many companies make the NE5532, including OnSemi? There’s no niche to fill — there’s an abundant supply already.

As for fabless: yeah right. Analog fab processes are very different from the digital ones that fabless typically deal with.

And if a giant like TI failed to make a replacement on a modern process, then some dinky little company isn't going to do better.

With modern design tools on an older production hw in many foundries around.

It’s not just about being “old hardware”, it’s about specific processes.

Imho with $200-300k you may produce a complete design and set of masks for the older nodes, imho. Producing a quantity of even say 100k chips for say $5-6 each may satisfy their ROI fast then..

Ridiculous, not a chance you could do it for that investment. Especially when you can still buy originals from OnSemi for pennies.

[julian1]

It is a problem to create bad fakes of your own products.
Dave's video shows that many asian fabs can meet the DS specs.
TI could have just contracted one of them to reliably make legitimate fakes.

[tom66]

It is a problem to create bad fakes of your own products.
Dave's video shows that many asian fabs can meet the DS specs.
TI could have just contracted one of them to reliably make legitimate fakes.

Well, you assume that those parts do meet those specs.  That isn't guaranteed, especially for smaller vendors.

The other factor is, sometimes companies have to source from the USA/not-China, due to national security reasons.

TI's enshittification is bad, but I can bet plenty of companies will still buy from them.

[julian1]

Would it even make sense cost-wise to respin the design taking advantage of the features of a different process node - to then have go through the steps to re-test/re-qualify all the specs.
Versus taking an existed fully supported product, and just getting an intern/chat-gpt to just merge the datasheets together.

[Alex Nikitin]

TI's enshittification is bad, but I can bet plenty of companies will still buy from them.

And we’ll have more stories like that one:

https://www.eevblog.com/forum/blog/eevblog-1752-wtf-texas-instruments-big-ne5532-changes/msg6276732/#msg6276732

Cheers

Alex

[aeg]

The problem with the 5532 at TI could become an opportunity for other smaller vendors, imho.. There is around 100 foundries around the world, afaik, most of them with older nodes. So - would it be a large problem for a smaller talented vendor to produce the 5532 within all that original specs? With even 3-5x the price for such a unique opamp people will buy it, sure (as all the discussions show)..

NJR and ON Semi already make 5532s. The key is whether they're savvy enough to add a part number suffix to their existing parts so people can call out a "real" 5532 on a BOM and not risk substitution to TI. If not, this is an opportunity for other vendors to step in if they're interested in low-margin, high-volume jellybeans. Diodes Inc and Microchip could probably do it. Or maybe it's some obscure Chinese chipmaker's time to shine.

[David Hess]

NJR and ON Semi already make 5532s. The key is whether they're savvy enough to add a part number suffix to their existing parts so people can call out a "real" 5532 on a BOM and not risk substitution to TI. If not, this is an opportunity for other vendors to step in if they're interested in low-margin, high-volume jellybeans. Diodes Inc and Microchip could probably do it. Or maybe it's some obscure Chinese chipmaker's time to shine.

Diotec has been picking up jellybean transistors, so they might be interested.  They make a 317 clone.

[Someone]

Comes up on google for "Texas Instruments 5532 1990", the document is SLOS075A
http://www.richard-dj1pi.de/Techn_Data/NE5532.pdf
That has the +/-3V to +/-20V marketing puff on page 1 but still recommended +/-15V.

The front page literally states:

Peak-to-Peak Output Voltage Swing
32 V Typ With VCC± = ±18 V and RL = 600 Ω

In other words, it's fine to use ±18V supply and even earns a "typical" parameter spec.

The new part completely deletes this specification; it doesn't even specify what the swing would be with 2kohm load or at ±15V supply.

TI took a part that had a defined top-line specification of performance, changed the part materially and deleted the specification and all references to it because they knew it was not comparable and would be noticeably worse than the competition.

The various allusions to 18V operation that I acknowledged in my opening post!

On the other hand they tacitly support +/-18V  operation by having some specifications at that operating point.

TI have not changed all this suddenly as people keep trying to claim. The recommended operating conditions were +/-15V as far back as I can see.

References to +/-20V or +/-18V disappear between SLOS075H and SLOS075I, May 2004 to April 2009

So this has been sitting there for 17 years at least.

[16bitanalogue]

I know you're just using this as a single example, but I think it sums up a bit of the flawed logic going on here. You (as in TI) can't assume that everyone using these at voltages higher than +/-15V is doing that in a new design. 99% of my work is repair work. And I'd say the amount of circuits I see using 5532s at or below +/-15V is maybe 30%.

They are using it beyond the stated, very clearly I may add, limits which did not change. This is universal and should be known by any engineer. This limit is vendor agnostic. The answer to "Can I use this above the maximum operating specficiation as long as I am below ABS MAX?" is yes, but the ECT is not gauranteed to hold. It's your judgement, and the condquences are yours and yours alone. I have had this conversation too many times with fellow engineers who should know better.

This is very much balancing the engineering, business, and legal waranties of the datasheet. Part of me empathizes from an engineering side, the other part of me says, "You didn't RTFM, and you want to complain?" That's some hillbilly, junior engineer bullshit. I watched part of Dave's video and he made an off hand comment that it is used to be +/- 22V part. Unequivocally wrong. It has always been a +/- 15V part.
Edit: You can even use the part beyond ABS MAX by a couple of volts because its still below transistor breakdown, but that's on the engineer who can't read.

As I said, if you all are so passionate about it ring up your local TI signal chain FAE or vent on thier support forum. Just remember you all didn't read or understand the instructions.

[16bitanalogue]

TI have not changed all this suddenly as people keep trying to claim. The recommended operating conditions were +/-15V as far back as I can see.

References to +/-20V or +/-18V disappear between SLOS075H and SLOS075I, May 2004 to April 2009

So this has been sitting there for 17 years at least.

100%

If the datasheet had curve or graph that showed +/- 18V then that would be something worthy to cruxify TI about because of the brazen inconsistency. The older datasheets did allude to +/-18V on the marketing puff page; engineers go to the ECT, graphs, and design sections. If I had the resonsibility to support this part, I would certainly have argued with the marketing engineer even in regards to the puff page. The datasheet would have been revised for clarity in spite of marketing's objections.

[David Aurora]

I know you're just using this as a single example, but I think it sums up a bit of the flawed logic going on here. You (as in TI) can't assume that everyone using these at voltages higher than +/-15V is doing that in a new design. 99% of my work is repair work. And I'd say the amount of circuits I see using 5532s at or below +/-15V is maybe 30%.

They are using it beyond the stated, very clearly I may add, limits which did not change. This is universal and should be known by any engineer. This limit is vendor agnostic. The answer to "Can I use this above the maximum operating specficiation as long as I am below ABS MAX?" is yes, but the ECT is not gauranteed to hold. It's your judgement, and the condquences are yours and yours alone. I have had this conversation too many times with fellow engineers who should know better.

This is very much balancing the engineering, business, and legal waranties of the datasheet. Part of me empathizes from an engineering side, the other part of me says, "You didn't RTFM, and you want to complain?" That's some hillbilly, junior engineer bullshit. I watched part of Dave's video and he made an off hand comment that it is used to be +/- 22V part. Unequivocally wrong. It has always been a +/- 15V part.
Edit: You can even use the part beyond ABS MAX by a couple of volts because its still below transistor breakdown, but that's on the engineer who can't read.

As I said, if you all are so passionate about it ring up your local TI signal chain FAE or vent on thier support forum. Just remember you all didn't read or understand the instructions.

Are you serious?

TYPICAL OPERATING CONDITIONS ARE NOT THE SAME AS ABSOLUTE MAXIMUM RATINGS.

A typical/recommended rating is picked as a balance point for reliability/performance/power consumption/etc and used for measurements because the data sheet would be the size of a bible if you listed performance under every possibly condition.

Using +/- 15V as typical or recommended does NOT mean it's bad practice to use it on different supplies. Like I said earlier, are you now going to say op amps can't be used at +/- 12V, or on single supply, or any of the other million configs people use because that isn't what's listed in the typical/recommended spec table?

All of this goes for what Someone is banging on with too.

First, neither of you seem to understand that parts CAN be used in different supply configurations, typical/recommended is not a firm usage rule. This is nothing new, it's been done like this going back to vacuum tube data sheet days, it's well understood by anyone with an ounce of practical electronics experience.

Second, you're arguing about a part where literally millions of them have been safely and reliably used like this for DECADES. This is not fringe, experimental use, the part has always been well understood to be capable of working like this.

[squadchannel]

It’s common sense that you can apply voltage up to the absolute maximum voltage.
TI is completely crazy for suddenly changing the absolute maximum voltage of the jerry bean 5532 on a whim.
They even said in the PCN that this change has absolutely no impact. If this causes a lot of devices to fail, wouldn’t we win a lawsuit? They’re doing something incredibly stupid.

(1) Operation outside the Absolute Maximum Ratings may cause permanent device damage. Absolute Maximum Ratings do not imply functional operation of the device at these or any other conditions beyond those listed under Recommended Operating Conditions.
If used outside the Section 5.3(Recommended Operating Conditions) but within the Absolute Maximum Ratings, the device may not be fully functional, and this may affect device reliability, functionality, performance, and shorten the device lifetime

[0xFFF0]

Texas Instruments – Acquisitions (Semiconductor / Electronics Companies only)
============================================================================

Year                | Company                           | Amount                | Focus / Products
--------------------|-----------------------------------|-----------------------|---------------------------------------------
2026 (announced)    | Silicon Labs                      | 7.5 billion USD      | Wireless Connectivity, IoT, Mixed-Signal
2011                | National Semiconductor            | 6.5 billion USD      | Analog Chips (broad portfolio)
2009                | Luminary Micro                    | undisclosed          | ARM Microcontrollers (Stellaris)
2009                | Ciclon Semiconductor             | undisclosed          | Power Management Semiconductors
2006                | Chipcon                           | 200 million USD      | Wireless (Zigbee, Low-Power RF)
2003                | Radia Inc.                        | ~320 million USD     | Wireless / WiFi Front-End
2001                | Graychip                          | undisclosed          | DSP Technologies
2000                | Burr-Brown Corporation            | 7.6 billion USD      | Precision Analog & Data Converters
1999                | Unitrode Corporation              | undisclosed          | Power Management & Analog
1999                | Telogy Networks                   | 457 million USD      | VoIP / Communications Semiconductors
1999                | Libit Signal Processing           | 365 million USD      | Wireless / Signal Processing
1999                | Butterfly VLSI                    | ~50 million USD      | Wireless Technologies
1999                | Power Trends                      | undisclosed          | Point-of-Use Power (DC/DC)
1998                | GO DSP                            | undisclosed          | DSP Software & Technology
1998                | Harris Semiconductor (product lines)| undisclosed        | Standard Logic (CD4000, HC etc.)
1997                | Amati Communications              | 395 million USD      | DSL / Broadband Communications
1996                | Tartan, Inc.                      | undisclosed          | DSP / Embedded Software

============================================================================

[tom66]

TI have not changed all this suddenly as people keep trying to claim. The recommended operating conditions were +/-15V as far back as I can see.

References to +/-20V or +/-18V disappear between SLOS075H and SLOS075I, May 2004 to April 2009

So this has been sitting there for 17 years at least.

100%

If the datasheet had curve or graph that showed +/- 18V then that would be something worthy to cruxify TI about because of the brazen inconsistency. The older datasheets did allude to +/-18V on the marketing puff page; engineers go to the ECT, graphs, and design sections. If I had the resonsibility to support this part, I would certainly have argued with the marketing engineer even in regards to the puff page. The datasheet would have been revised for clarity in spite of marketing's objections.

The front page of the datasheet literally stated it's rated for 32Vp-p from a +/-18V supply into 600 ohms.  Not only is that gone now, but so is the rating for any peak-to-peak swing at any load from any supply.

[aeg]

If anyone wants to start making 5532s, here's the equipment you need: https://www.gesemi.com/featured-project-texas-instruments/

[squadchannel]

https://www.eevblog.com/forum/blog/eevblog-1752-wtf-texas-instruments-big-ne5532-changes/msg6277242/#msg6277242

As they are kind of ending the old process, other part should be effected too from changes. It looks like this also applies to the gold old uA741:
The new 2026, rev. H datasheet gives quite some changes:
no more offset adjust - though some pictures still show it, there are other oddities with the bias current
much lower supply current (0.13 mA compared to 1.7 mA typ)
much lower input bias , higher input resistance
near RR output swing
80 mA typ output current
+-18 V supply only   -- AFAIR they had a +-22 and +-18 V max version. At least for the LM741 they have.
--> not sure if this is still a BJT based part. Most specs are an improvement, but some could still break old designs.
     the list the TLV9301 as a similar part and from the specs - it could be a similar CMOS part.

On the positive side there is still the LM741 listed with more the original specs. The question may be for how long.

He found it. uA741 too.

https://mm.digikey.com/Volume0/opasdata/d220001/medias/docus/5485/PCN20230713001.1.pdf


I found this on the TI forum for the uA741.
https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1650472/ua741-lot-code-and-timing-of-revision-change-on-ua741cp-so-546434

    If the date code starts with 22 or lower, then it contains the legacy die with the offset trim pin function.
    If the first digit on the top marking is a 2 or lower, then it contains the legacy die with the offset trim pin function.
    If the year listed in the date code is 24 or higher, then the part could be the new die, and does not have the trim pin function.

The PCN was issued in 2023.
The die was changed in 2024.
The datasheet was revised in 2026.

Of course, the PCN says

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

TI quietly lowered the maximum voltage and changed the pin to NC without any notice.
It doesn’t get any more ridiculous than that.
The SFAB to RFAB conversion is a disaster. There are surely still ICs out there that are being affected in ways people aren't even aware of yet.

[David Hess]

Reliability suffers when operating at the absolute maximum rating, with some differences between technologies.

Bipolar parts can be graded for higher operation voltage by testing quiescent and bias currents.  Tektronix did this to use 36 volt 741s at higher voltages even though 44 volt 741s were available.

This will not work with CMOS parts because gate oxide breakdown is fatal.  For instance operating the National LMC (16 volts) or Texas Instruments TLC (18 volts) CMOS parts above their absolute maximum voltages, even momentarily, will almost certainly destroy them quickly in my experience.  Their maximum recommended operating voltages are lower.

[Alex Nikitin]

So, more or less anything analogue (not sure about digital) from TI , manufactured from 2023 is suspect and should not be used, full stop.

Cheers

Alex

[squadchannel]

I wonder what SFAB used to manufacture....

[MT]

Suggested replacements for 5532 could be OP275, OPA2134, LME49720, LME49860 but these are priced at least 4-5x more even in volumes. NJM2068 if still manufactured.