Good opamp for that1646

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Don't hold your breath for a schematic. As I offered this is probably TMI.
Quite. Agreed "110,% all day long," (ref UK R4 Deadringers)
It is the nature of web discussions for threads to veer and ramble.
Quite. I like it. Reminds me of actual conversations 🙂
Back several decades ago when op amps were lower performance we routinely tried to squeeze out the last drop of performance with trickery... don't open the can of worms about alternate compensation schemes.

I repeat feel free to ignore this...

JR
It's all good.
I'm working in the High Voltage area atm so 36V is light relief 😃
 
BTW, how is the OPA604 compared to the 5534 for this application?

What is your source impedance going to be ?
Look at the datasheet noise specs.
It's a FET input type so better for high Z sources rather than low impedance outputs.
From the datasheet:

NOISE PERFORMANCE Op amp noise is described by two parameters—noise voltage and noise current. The voltage noise determines the noise performance with low source impedance. Low noise bipolar-input op amps such as the OPA27 and OPA37 provide very low voltage noise. But if source impedance is greater than a few thousand ohms, the current noise of bipolar-input op amps react with the source impedance and will dominate. At a few thousand ohms source impedance and above, the OPA604 will generally provide lower noise.
 
My Tonelux op-amp had it's first pole at 10K (open loop gain) and total gain was only 75 dB, and sounded more open largely because the amp wasn't at full gain and then fed back 100% when using the 10K or 20K FB resistor. With my amp, the compensation cap (not really needed) was very small. It was biased as a "in between" class A and B, where there were basically 3 diodes in the output stage instead of 2. I did that by adjusting the bias towards class A until the distortion leveled off, and stopped. I had several "golden ear" designers and engineers tell me that it was the best sounding op-amp they ever heard...
 
My Tonelux op-amp had it's first pole at 10K (open loop gain) and total gain was only 75 dB, and sounded more open largely because the amp wasn't at full gain and then fed back 100% when using the 10K or 20K FB resistor. With my amp, the compensation cap (not really needed) was very small.
I don't question what people say they hear on the internets. But I am willing to ponder op amps and NF.

I mused about this in my old (1980) article about console performance. Some popular op amps with more than 100 dB of open loop gain start the compensation pole for the open loop response as low as 50Hz, to attenuate that open loop gain enough to deliver unity gain stable performance and wide bandwidth. We want that open loop gain to be rolled off below unity gain, by the frequency when internal delays and phase shift accumulate up to 180' of phase shift turning negative feedback positive. An unintended consequence of this dominant pole integrator's open loop transfer function is that input referred error voltages (that get amplified by noise gain) will be 90' phase shifted from the output signal. In the extreme this isn't very pretty but as long as we provide adequate loop gain margin, and use modest noise gains, this is small and well tolerated.

For today's TMI in one of my esoteric phono preamps P10 (I tend to over think all of those designs). I counted on the dominant pole compensation in the typical TL07x to make the open loop transfer function pole similar to the RIAA closed loop feedback pole (3180 uSec). Holding the error voltage in phase with the closed loop transfer function eliminates one unintended consequence (in theory). Of course these phono preamps were already head and shoulders better than vinyl so squeezing out small incremental improvements was almost laughable (of course I didn't stop there).

It was biased as a "in between" class A and B, where there were basically 3 diodes in the output stage instead of 2. I did that by adjusting the bias towards class A until the distortion leveled off, and stopped. I had several "golden ear" designers and engineers tell me that it was the best sounding op-amp they ever heard...
There are diminishing returns from too much class A bias current, but it is relatively easy for op amp loads (harder when driving loud speakers).

JR
 
BTW, how is the OPA604 compared to the 5534 for this application?
I don't think it would make any significant difference, since any improvement in terms of BW, slew-rate and distortion of the 604 will be limited by the 1646.
In terms of noise, the 604 is twice noisier than the 5532, almost 10dB noisier than 5534, but the global noise performance is dominated by the 1646, which is about 20dB noisier than the unity-gain opamp.
 
Reading the 5534 manual a bit more I noticed that they claim a slew rate of 6V/us at unity gain (with Cc=22pF) and 13V/us with a gain higher than 3 (i.e. without the compensation cap). Would it make sense to have a gain higher than 3 (or example 3.5) and having a pad at the input instead of a voltage follower or other specs will be compromised with gain higher than unity?

Also, what I noticed is that although the data sheet states that the 5534 is for gains above 3, I have read many comments online where they take for granted that they unstable for gain less than 5. However, most comments are rather old. Have they changed the specs (I doubt) or I'm missing something?
 
Last edited:
Reading the 5534 manual a bit more I noticed that they claim a slew rate of 6V/us at unity gain (with Cc=22pF) and 13V/us with a gain higher than 3 (i.e. without the compensation cap). Would it make sense to have a gain higher than 3 (or example 3.5) and having a pad at the input instead of a voltage follower or other specs will be compromised with gain higher than unity?
If you're application requires a slew rate of 13V/us instead of only 6V/us such as because you're transmitting ultrasonic pulses from a submarine for echo location, then yes.
 
Reading the 5534 manual a bit more I noticed that they claim a slew rate of 6V/us at unity gain (with Cc=22pF) and 13V/us with a gain higher than 3 (i.e. without the compensation cap). Would it make sense to have a gain higher than 3 (or example 3.5) and having a pad at the input instead of a voltage follower or other specs will be compromised with gain higher than unity?
It doesn't make sense to add un-needed closed loop gain just to milk out higher slew rate, but there are old engineering tricks to accomplish that. An added RC from the - input to ground can fool the 5534 into thinking it is running at a closed loop gain of higher than 3x for stability at very HF, while the RC can be tuned to be out of the picture for audio signals.
Also, what I noticed is that although the data sheet states that the 5534 is for gains above 3, I have read many comments online where they take for granted that they unstable for gain less than 5. However, most comments are rather old. Have they changed the specs (I doubt) or I'm missing something?
No the spec has not changed AFAIK.

JR
 
Depending on the specific manufacturer and datasheet or application note then the gain for uncompensated stability is stated as 3 or 5 as far as I can see. Go with 5 for convenience.
 
Depending on the specific manufacturer and datasheet or application note then the gain for uncompensated stability is stated as 3 or 5 as far as I can see. Go with 5 for convenience.

Don't quote me on that but I think only TI makes these chips nowadays.
 
Reading the 5534 manual a bit more I noticed that they claim a slew rate of 6V/us at unity gain (with Cc=22pF) and 13V/us with a gain higher than 3 (i.e. without the compensation cap). Would it make sense to have a gain higher than 3 (or example 3.5) and having a pad at the input instead of a voltage follower or other specs will be compromised with gain higher than unity?
You must put it in perspective with the fact that, in order to reproduce a 20kHz sinewave at full-blow (+20dBu), one needs only about 1.5V/us.
Datasheet shows that power BW with a 47pF compensation cap exceeds 40kHz, 60kHz with 22pF.
 
You must put it in perspective with the fact that, in order to reproduce a 20kHz sinewave at full-blow (+20dBu), one needs only about 1.5V/us.
Datasheet shows that power BW with a 47pF compensation cap exceeds 40kHz, 60kHz with 22pF.
That is the math to avoid slew rate limiting, ASSuming the circuit behaves perfectly up to the edge of overload. In practice it is prudent to design with some modest amount of slew rate headroom. My preference is to rise time limit audio paths but that is TMI for this discussion.

Resist the temptation to design in crazy hi slew rate parts as they sometimes deliver those metrics by gaming how the circuit behaves while slewing. Some high speed op amps designed for use in sample and hold circuits bypass around the input LTP during slew events to charge the compensation cap faster. The net result may be a faster acquisition time, but is not very useful for linear circuits.

A lot of good music managed to survive 1/2V/usec audio paths. -10dBV bedroom recording gear managed high quality audio somewhat better by using a lower nominal 0VU.

JR
 
The LM4562 represents a real advance on the 5532. It is, however, still a good deal more expensive and is not perfect –it appears to be more easily damaged by excess common-mode voltages, and there is some evidence it is more susceptible to RF demodulation (small signal audio design)

https://www.audiosciencereview.com/...pecially-of-the-lm4562-lme497x0-family.10687/
and it some other places I can't remember right now.

I think I'm just gonna go with the 5534 and call it a day :)

Just replying to this as quite a bit of stuff in the link and associated with it.
Yes - it's interesting. In summary I wouldn't be overly concerned with the radiated em issue.
I can see that the issue seems to measure worse than other devices including alternative bipolar input opamps.
But we are looking at a huge 86dB gain for the opamp in the test configuration there.
What I find more concerning is the reported inconsistency in low frequency noise characteristics. Quite a bit to untangle there. Such as whether it is associated with a particular period of manufacture / process / fabrication geometry. And what is the current position.
 
Don't quote me on that but I think only TI makes these chips nowadays.

JRC still makes them. We use NJM5532 on our modules.

I have some NEVE line amp cards that have the 5534 running at 6dB gain, and the unbalanced input is padded (voltage divider). 5534 is driving an output tx. So, that my be the reason the designer chose to configure it that way. But as I mentioned before it is a work around.
 
JRC still makes them. We use NJM5532 on our modules.

I have some NEVE line amp cards that have the 5534 running at 6dB gain, and the unbalanced input is padded (voltage divider). 5534 is driving an output tx. So, that my be the reason the designer chose to configure it that way. But as I mentioned before it is a work around.
The 5532 are still available but the 5534 are discontinued AFAIK.
 
Back
Top