Does Op-amp Rolling Work?

[antcollinet]

I just reread your article with a focus on the OPA1612, and I noticed something. You wrote: “The low voltage noise density for the OPA1612 is notable.” That is quite confusing. In the context of “rolling,” it doesn’t really matter—but what exactly is “notable,” and under what conditions could it, if at all, cross the threshold from being merely measurable to being audible?

I also feel like I missed a scenario: “very sensitive headphones or IEMs.” Is that still only a measurable difference, or does it become relevant when the volume is turned up but no signal is present (similar to track switching)?
And the main idea may be that, in cases of ‘messed up designs,’ people can perceive a difference when choosing between something very bad and something just bad - this could explain why do they hear something as a diff.

Youve been given all the information you can use. If you lack the knowhow to understand it that is on you, not us.

What you are doing now is called sealioning. Not big, Not clever, please stop.

 

[solderdude]

I just reread your article with a focus on the OPA1612, and I noticed something. You wrote: “The low voltage noise density for the OPA1612 is notable.” That is quite confusing. In the context of “rolling,” it doesn’t really matter—but what exactly is “notable,” and under what conditions could it, if at all, cross the threshold from being merely measurable to being audible?

I also feel like I missed a scenario: “very sensitive headphones or IEMs.” Is that still only a measurable difference, or does it become relevant when the volume is turned up but no signal is present (similar to track switching)?
Mainly I mean this case: "In a very bad circuit, the op-amp doesn’t define quality—it only sets how badly things fail." and could then it be an audible influence of rolling Op-amps when: power supply noise leaking through, oscillation or ringing, input noise heavily amplified, overload/clipping, EMI pickup sensitivity, etc... ?

If anyone could prove that there are inexplicable audible differences between op-amps in a circuit would this not be known by people in the know ?
Wouldn't scientists not be all over it figuring that out ?
Would there not be peer reviewed papers on it with listening tests and links to properties.
If audio is not about waveform shapes what else is it about ?
Is there a missing signal other than the amplitude over time ?
Have you ever heard of null tests ?

Sure when noise levels become audible this is an issue and VERY measurable.
There has been many research done on audibility of distortion with test tones and music.
With music 0.1 to 1% is usually inaudible.
With test tones 0.01% is already inaudible (well depends on the type of distortion).

Why would 0.00003 percent distortion sound any better than 0.006 % ?
It doesn't.

When do op-amps matter for audio ?
In high gain circuits (phone/mic pre-amps, instrumentation amplifiers, sensor amplifiers etc.) or when used to drive loads they are not intended to drive (low impedance headphones for instance) or when certain op-amps are used in circuits where the op-amp choice matters and specific specs matter, for instance input current/slew rate for sample-hold etc.
For instance using high BW or op-amps that require minimal gain of 5 are used in such circuits the PCB layout and local decoupling of the power rails and feedback as well as input capacitance starts to matter. Using those in circuits that are low gain or do not have the proper layout/decoupling you are asking for problems like oscillations etc. This can be in the MHz region or higher.

Fact is in line level stages (pre-amps, DAC filters, power amplifiers) usually the demands aren't high and all of the cheap op-amps 5532/34 will work just as well as the 'audio grade' ones.

EVERY one that swaps op-amps hears differences. The issue is that even when you tell people you swapped in an expensive audio-grade op-amp they will hear differences even if there aren't any.
Try that on an unsuspecting audiophile friend... just open something up and pretend to swap (don't let them see you not changing anything) and let them listen.
I can practically guarantee you they WILL hear a difference.
Never tell them you tested them... let them continue to believe they heard a difference and you did not trick them... if you want to keep them as friends.

Happy experimenting ... I have done a few of those on others... and accidentally fooled myself a couple of time too. Very enlightening... have fun.

 

[MAB]

I just reread your article with a focus on the OPA1612, and I noticed something. You wrote: “The low voltage noise density for the OPA1612 is notable.” That is quite confusing.

I had you on ignore, but had to peek and comment based on some responses from other members.

Douglas Self wrote that, not me. I am honored by the misattribution, but decline to take credit. The data he presents is extremely clear. Perhaps get one of his books and study, hopefully more carefully than you read my posts.

I'll leave you to it.

 

[PlanarSearcher]

Douglas Self wrote that, not me. I am honored by the misattribution, but decline to take credit. The data he presents is extremely clear.

“Sorry about that — I missed the part ‘referenced by you.’ Also, regarding the term ‘confusing,’ I probably wasn’t clear enough. In that context, ‘confusing’ refers specifically to the question above about the case where the circuit is poorly designed, which has already been addressed and clarified by Solderdude. There’s really nothing further to add except a big thank you to everyone for the help and references.

 

[Postkrunk]

https://audioxpress.com/news/akm-pu...the-dac-with-new-ak491x-premium-audio-op-amps

AKM op amps AK4911 and AK4912 are here for rolling.

I really hope they're unrollable.

I mean, there should be a reason to make a single-channel opamp in SSOP-16. I wonder if it has pins for external compensation AND balancing. Recently I thought about internally decompensated precision opamps, ahd there isn't any currently in production, to my knowing.

 

[MAB]

The datasheets are available, but not released for general availability. They can be requested if you have a business account.

AK4911ET | Audio Components | Products | Asahi Kasei Microdevices (AKM)

The AK4911 is a high‑performance single‑channel audio operational amplifier designed to meet the demands of next‑generation high‑resolution audio systems.

www.akm.com

AK4912ET | Audio Components | Products | Asahi Kasei Microdevices (AKM)

The AK4912 is a high‑performance dual‑channel audio operational amplifier designed to meet the demands of next‑generation high‑resolution audio systems.

www.akm.com

The SSOP-16 is ganged into 8 pins. It has no external comp. I regret I can't post the datasheets, but you can request them through the above links.

 

[Bruce Morgen]

Meanwhile, TI has apparently changed the circuitry of their version of the classic NE5532.

 

[solderdude]

Basically it is only really an issue when using this op=amp in circuits running on voltage rails > +/- 18V and below +/-22V, which most audio circuits don't.
The usual rail voltage is either +/- 5V or max. +/- 15V.

For some circuits that are specifically designed for this op-amp and or > +/- 18V rails you could run into issues using the newer TI version.
Fortunately there are plenty of other 5532 manufacturers that still deliver the 'original' part and one can simply choose not to use TI if unsure.

For line level low gain stages (which is where most of these are being used for) 1V/us is enough.

 

[Bruce Morgen]

one can simply choose not to use TI if unsure.

True, but this will probably result in inflated prices for "original circuit" TI NE5532s -- especially the "vintage" ones from their old U.S.. foundry.

 

[KSTR]

TI finally admitted that LM833, RC4580, MC33078 and NE5532 are now one and the same part: https://e2e.ti.com/support/audio-gr...lm833-rc4580-and-mc33078-all-now-the-same-die

For the RC4580, LM833 and MC33078 no big deal, the latter two's datasheets always suggested that they are the same thing.
But the new "5532" now is completely different in many regards and it is now a lower spec'd part than before, on several key specs.

The re-design is a new part on a new silicon process, at least that's what I would infer from TI's PCN. One would expect they'd now all have the same unified and updated datasheet but TI decided to leave customers in the dark. The 5532 datasheet is just a scam as is the part, basically.

 

[MaxwellsEq]

TI finally admitted that LM833, RC4580, MC33078 and NE5532 are now one and the same part: https://e2e.ti.com/support/audio-gr...lm833-rc4580-and-mc33078-all-now-the-same-die

For the RC4580, LM833 and MC33078 no big deal, the latter two's datasheets always suggested that they are the same thing.
But the new "5532" now is completely different in many regards and it is now a lower spec'd part than before, on several key specs.

The re-design is a new part on a new silicon process, at least that's what I would infer from TI's PCN. One would expect they'd now all have the same unified and updated datasheet but TI decided to leave customers in the dark. The 5532 datasheet is just a scam as is the part, basically.

This is atrocious behaviour by Texas Instruments! How can they be so utterly unprofessional? There are thousands of devices in the professional field which will require supply from other manufacturers.

 

[Ropeburn]

This is atrocious behaviour by Texas Instruments! How can they be so utterly unprofessional? There are thousands of devices in the professional field which will require supply from other manufacturers.

Hey now, it's clearly a highly professional scam. Don't you see it?

 

[antcollinet]

Meanwhile, TI has apparently changed the circuitry of their version of the classic NE5532.

Jeez - this could cost 100k's for a volume manufacturer, just in re-qualification cost. Millions if it results in field failures.

 

[diddley]

Okay, I made it through the video and understand where the problem is. But this guy is talking and acting as if the world is going under, very tiresome.

 

[antcollinet]

Okay, I made it through the video and understand where the problem is. But this guy is talking and acting as if the world is going under, very tiresome.

I think he is reacting exactly as any engineer who has worked in a manufacturing organisation would react - to a component manufacturer doing something quite this dumbass.

Certainly my incredulity matched his.

 

[MaxwellsEq]

Okay, I made it through the video and understand where the problem is. But this guy is talking and acting as if the world is going under, very tiresome.

It's likely to have a huge impact in the professional world. A big studio may have thousands of 5532/4s.

 

[MAB]

It's likely to have a huge impact in the professional world. A big studio may have thousands of 5532/4s.

But does any studio equipment use supplies greater than 15V?

I don’t think any of the other parameters will make a difference, better or worse.

 

[KSTR]

There is one issue when repairing existing studio products in a few cases which exploited some features which are gone now, where a non-informed tech might replace NE5532 with the new TI fakes.
But word of mouth is already spreading fast and TI 5532's with date codes later than 2024 will be blacklisted.

The bigger issue is that some pro and consumer audio products made in the last year or so may have been fitted with the fakes without the manufacturer being aware of the change, prone to cause unexpected problems/failures in end-of-the-line testing, or worse, later in the field. The TI support forum is seeing a surge of such cases at the moment.

 

[MAB]

There is one issue when repairing existing studio products in a few cases which exploited some features which are gone now, where a non-informed tech might replace NE5532 with the new TI fakes.
But word of mouth is already spreading fast and TI 5532's with date codes later than 2024 will be blacklisted.

The bigger issue is that some pro and consumer audio products made in the last year or so may have been fitted with the fakes without the manufacturer being aware of the change, prone to cause unexpected problems/failures in end-of-the-line testing, or worse, later in the field. The TI support forum is seeing a surge of such cases at the moment.

Do mean extended supply voltages?

For sure that will be a big issue if people are indeed running supplies above the NOM. I am sure there are a few.

Not saying this change isn’t big. I’ve done several process change and node consolidations. Like lead to copper bumps, aluminum to copper interconnects, consolidating 130 and 90nm to 65nm node, etc. There is incredible financial incentive to do this, with risk. I’m just trying to wrap my head around the actual risk here.

 

[solderdude]

Let's face it all the gear that already has 5532 in it are O.K. and will continue to do so.
Should one of those op-amp ever break you AND performance could be affected > 18V rails or 5Vus slew rate is an issue then one can still buy 5532 from other brands.

Any new gear using TI 5532 will probably work within design parameters and should there be the odd design with >18V supply that production batch will likely fail and the issue will be found (broken 5532). The solution is simple... buy 5532 but not the ones from TI.

Most designs I know of that are using this part usually have voltage rails of +/- 15V max.

Waiting for the first reports of op-amp swappers hearing differences between legacy and new TI 5532 (all sighted of course).