Choosing of best sounding OP AMPs for the lowest possible THD+N -really the best Way?

[Mark Tillotson]

You say that but LM4562's are not screened for excess 1/f noise(*) and anyway have _much_ more current noise than an NE5534A - still a great choice for a MM phono preamp...


[ incidentally the NE5534A is better than the NE5532 for preamp duty with its guaranteed noise performance, but isn't
unity gain stable as is, and is single not dual - between them they are probably still about the best value for money for
audio signal chain in the 10k impedance range, about 20cents per opamp ]



Some of the newer JFET opamps are quite impressive, OPA1652 for instance, assuming they don't suffer from the same
production spread of 1/f noise.


(*) LME49720: popcorn noise - Audio forum - Audio - TI E2E support forums

 

[kozard]

From the datasheets at 1kHz:

OPA2228 0.4 pA/√Hz (!)
NE5532 0.7 pA/√Hz
LM4562 1.6 pA/√Hz
OPA1612 1.7 pA/√Hz
​

I guess my thinking was that this was lower down on the priority list at least in my application for the PCM1794 or ES9038Q2M. In other words not the factor that I am most concerned with for device selection here. Is that incorrect?

Then again the Ti datasheet for the PCM1794 makes use of NE5534 and I must admit that I was impressed with the PCM1794 using just Ti NE5532 before I obtained the OPA1612. The resistors being used are 750, 560 & 270 Ohms or 820 & 360 Ohms. (Depending on which datasheet circuit, figures 24 & 25 in the Ti datasheet.)

 

[Tfive]

if you really want to find a definitive answer regarding noise you have to look at the surrounding circuitry of the opamp and take all imedances into account. Somewhat similar to what I tried here:

I/V conversion circuit - noise considerations


Looking at current and voltage density noise numbers without context will can be misleading.

 

[kozard]

Yes. I agree. Thank you for doing the math for my use cases! (ES9038Q2M & PCM1794.)

if you really want to find a definitive answer regarding noise you have to look at the surrounding circuitry of the opamp and take all imedances into account. Somewhat similar to what I tried here:

I/V conversion circuit - noise considerations

Looking at current and voltage density noise numbers without context will can be misleading.

 

[Tfive]

I just updated the calculation result slightly. Beware.


The formula did not change though.

 

[HumbleDeer]

Holy F, that double OPA228 is expensive.

I'll take a look at the LM4562. I was always kind of under the impression the ne5532p had some or a lot of flaws regarding distortion. I'm of course trying to make this differential output an improvement over just single ended!

 

[dotneck335]

The only "flaws" I am aware of in the NE5532 were the T.I. versions, which were reported to have more HF distortion than other brands. The Signetics/Philips brands were the ones used in all the classic mixing consoles from the late 70s onward.

 

[kozard]

The only "flaws" I am aware of in the NE5532 were the T.I. versions, which were reported to have more HF distortion than other brands. The Signetics/Philips brands were the ones used in all the classic mixing consoles from the late 70s onward.

Interesting. Are there any good links with data for this? Of current production parts which is best? I assumed that Ti used Ti NE5534 for the reference PCM1794 designs.

On Semi? JRC? Ti?

Is anyone else still making them?

 

[HumbleDeer]

I'm not sure if this counts as an op amp as such, but I suppose newer IC's like the SSM2142 will do better for my purpose if I can get my hands on some designs. These are apparently dedicated amplifiers/opamps/drivers for single to diff

 

[kozard]

Holy F, that double OPA228 is expensive.

I'll take a look at the LM4562. I was always kind of under the impression the ne5532p had some or a lot of flaws regarding distortion. I'm of course trying to make this differential output an improvement over just single ended!

Yes. That is why I mentioned the LM4562 first.


I think I have four OPA2228 which I bought in 2002. I don't remember them being so expensive back then. (But the online order history does not go back that far so I can not be sure.)

 

[kozard]

I'm not sure if this counts as an op amp as such, but I suppose newer IC's like the SSM2142 will do better for my purpose if I can get my hands on some designs. These are apparently dedicated amplifiers/opamps/drivers for single to diff

Not sure but perhaps look at OPA1632? Just don't get them online (eBay/AliExpress/etc). I bought four online and of course they were fake...

 

[dotneck335]

You can get them online with no problem----just order from Mouser or DigiKey. No fakes.
As far as the new NE5532/34s, the ON Semi is the best. Unfortunately, they only offer it in SOIC-8----no PDIP-8s anymore.

 

[HumbleDeer]

SOIC-8 is not really a problem. You can solder those to double sided perfboard easy as.

Not sure but perhaps look at OPA1632? Just don't get them online (eBay/AliExpress/etc). I bought four online and of course they were fake...

Dear-oh! These are sufficiently cheap I'll just buy one of the boards off aliexpress and replace the chips themselves if needed (edit: I'm pretty proficient in soldering & repair.) Interesting, thanks! I have honestly never looked into this part of true design. I always stuck with repair and bodgery on basic amplifier things and programming computer things

 

[xtraforcz]

BTW is the LM4562 burst issue mentioned earlier real? Any reason behind this sudden discovery after it exists for more than a decade?

 

[jean-paul]

You mean this opinion?

LM4562 / LME49720 low frequency noise - Audio forum - Audio - TI E2E support forums

Fake news I tell you.

 

[xtraforcz]

Yeah, I am not sure was that an actual event even after googling around. I was quite worried after I had stockpiled a bunch of LM4562's at a huge discount in 2019.

 

[jean-paul]

You can have mine. I will never use these again after having noticed the what I call unreliable behavior of LM4562. The huge discount may have had a reason

 

[xtraforcz]

Thanks jean! But I have enough to last a few builds already.

 

[jean-paul]

Please use IC sockets with these. For new designs aim for SMD as the choice is better and OPA1642 and OPA1656 can be used.

 

[xtraforcz]

Couldn't it withstand common soldering temperature?

I am considering OPA1656, but the availability is quite low, currently no stock on major distributors in my area.

My current consideration: OPA1656, OPA1622 (INA1620 for a smaller project), BUF634A, OPA1612

Yes im a TI fanboy