For the NE5534A, 3.5 nV/√Hz and 0.4 pA/√Hz
OPA1612: 1.1 nV/√Hz and 1.7 pA/√Hz
The problem with so-called ultralow noise op-amps is that they usually have ultralow equivalent input voltage noise, but ultrahigh equivalent input current noise. For RIAA amplifiers for moving-coil cartridges, the current noise doesn't matter much due to the small source impedance, for RIAA amplifiers for moving-magnet cartridges, it's usually dominant.
A simple rule of thumb is that if it is the total RIAA- and A-weighted noise you are interested in, you have to multiply the current noise by the cartridge impedance at 3852 Hz, about 12 kohm for a typical 500 mH MM cartridge, to compare it to the voltage noise. With 12 kohm:
NE5534A, 3.5 nV/√Hz and 4.8 nV/√Hz
OPA1612: 1.1 nV/√Hz and 20.4 nV/√Hz
So, noise-wise, an OPA1612 would be a good choice for moving coil, but for moving magnet, it is outperformed by a large margin by the NE5534A. @sa-mo didn't specify whether the cartridge is moving magnet or moving coil.
Unfortunately, the NE5534A is a single op-amp, so it doesn't meet sa-mo's requirements. The NE5532A is a dual op-amp, but with somewhat worse noise specifications
Last edited:
A 'great design' wouldn't need it!
If a manufacturer wants to out-spec a rival, they might stoop to such trickery and then imply this means their opamps sound the best, aided by the emotive names for them.
I'd like to see double blind listening tests comparing the old stalwarts like the NE5532/34 and OPA134/2134. with the super duper op amps.
If a manufacturer wants to out-spec a rival, they might stoop to such trickery and then imply this means their opamps sound the best, aided by the emotive names for them.
I'd like to see double blind listening tests comparing the old stalwarts like the NE5532/34 and OPA134/2134. with the super duper op amps.
Yes I know, is interesting
Problem is TI wont work against itself so the data is correct.
Problem is this is ideal test conditions.
And rule of thumb equations are correct yes .
Problem is nobody bothers to measure it in real circuit conditions.
And compare same 2 op amps also in same conditions.
So to make yah feel better the expected input current noise of 20.4 nV expected from 1612 at 3852 Hz is actually more like 31.76nV
Unfortunately 5534 input current noise is not assumed 4.8 nV it is also 32.1nV
So 1612 actually a slightly better but basically the same. Guess when you scale the graphs 1612 could be expressed as way better.
Im not gonna debate the meaningless difference of 31.7 to 32.1
But overall noise from input to output, then distortion and numerous other factors. People just have a hard time
accepting modern engineering has surpassed old opamps.
With filters in RIAA circuit, distortion analysis its pointless to compare distortion = same with filters.
Without filters driving various loads then distortion is roughly equal to datasheet
But there is numerous tests to change that up to watch 5534 fall to .01% and 1612 will hold up .001%
so no it is not .000019 either as datasheet shows. Again in test conditions data is correct
real conditions everything changes. Basically the pops and crackle of vinyl is way way way over what
a opamp does far as noise. People can drop in a TL072 and swear they hear the difference.
It doesnt, imagination probably, real world= no
Last edited:
Exactly because then there is no difference.
And not down talking vinyl either cause I love it very much.
Crackle n pops are louder than the cheapo opamps.
Might as well do the comparison with magical 2000 dollar tweeters
compared to a well done 30 dollar aluminum tweeter.
listening to old jazz and big band recordings which I also love.
The detail of even a well done cheapo tweeter or ribbon.
You can hear how lovely microphone distortion is on the original recordings.
Something a 300 dollar amplifier or 3000 dollar magic amplifier wont ever change.
either way I still cry a little listening to some of Dukes old stuff
So just build the best you can. And know it will be miles above other restraints
in the hobby
So what exactly did you measure under what conditions and how? Source impedance, weighting?
By the way, my rule of thumb is based on the calculations in "Noise and moving-magnet cartridges", Electronics World October 2003, pages 38...43, https://worldradiohistory.com/UK/Wireless-World/00s/Electronics-World-2003-10-S-OCR.pdf Mind you, Electronics World drew one of the sections of the gain switch in the wrong state in figure 5 and I mixed up the terms spectral density and power spectral density.
Noise wise, the OPA2210 outperforms the NE5534A: less noise voltage, about the same noise current. The OPA1612 is just optimized for lower source impedances, such as sa-mo's moving coil cartridge: more input stage tail current, hence less noise voltage and more noise current.
Opamps for phono stages is not a "one size fits all" type of problem.
Moving coil (MC) cartridges have very low impedance, so you'll want an opamp with low voltage noise but you can tolerate some current noise. The OPA1612 springs to mind, though that's not available in DIP. The LM4562/LME49720 is.
Moving magnet (MM) cartridges have relatively high impedance due to the inductance of the pickup coil, so you'll want an opamp with low current noise but you can tolerate higher voltage noise as long as it doesn't end up dominating over the current noise. It's been a while since I last looked at the simulations, but I seem to recall that the NE5534 was very hard to beat. OPA627/637 were options but they're ungodly expensive (and not dual). The OPA1642 (dual, but not DIP) was a strong candidate in my sims.
For the full picture, I suggest including a model of the cartridge in your simulations (or calculations).
Tom
Moving coil (MC) cartridges have very low impedance, so you'll want an opamp with low voltage noise but you can tolerate some current noise. The OPA1612 springs to mind, though that's not available in DIP. The LM4562/LME49720 is.
Moving magnet (MM) cartridges have relatively high impedance due to the inductance of the pickup coil, so you'll want an opamp with low current noise but you can tolerate higher voltage noise as long as it doesn't end up dominating over the current noise. It's been a while since I last looked at the simulations, but I seem to recall that the NE5534 was very hard to beat. OPA627/637 were options but they're ungodly expensive (and not dual). The OPA1642 (dual, but not DIP) was a strong candidate in my sims.
For the full picture, I suggest including a model of the cartridge in your simulations (or calculations).
Tom
They use very high open-loop gain and high bandwidth, as well as trying to get very linear open-loop response too. The results speak for themselves, sub-ppm distortion.
Not sure what "distortion cancelling techniques" means in this context other than having a very linear open-loop response? All circuit design is trickery surely? That's the point, but the tricks have to work.
"Distortion cancelling techniques" is a phrase I would associate with things like using a differential pair or some other balanced structure instead of a common emitter stage to get rid of even-order distortion, or using a multitanh stage instead of just a differential pair to also get rid of some odd-order distortion, or applying emitter resistors in a complementary emitter follower stage to partly cancel its distortion. Many of these are quite old techniques.
Not the case, here's my measurements for OPA1612 v. NE5534A with 500mH (800 ohms ESR) as the source impedance simulating a real MM cartridge:
OPA1612:
Note the big hump across the whole HF part of the audio spectrum - current noise hits 500mH, big rise in noise, basically hiss.
NE5534A:
Perhaps 12dB better at the top end, definitely quieter from about 500Hz upwards. It is difficult using discrete inductors to get this measurement because they act as magnetic loop antennas, this spectrum was recorded mobile, laptop powered, about 1/4 mile from nearest power line / building and still it picks up some interference (more visible than the 1612 case as its not so masked by the chip itself).
Lets redo those noise calculations:
500mH+800R at 3852Hz is 800R + 12.1k reactive = 12.1k impedance.
With 1.7pA/√Hz that's 20.6nV/√Hz, so we expect the 1612 to have about 21nV
With 0.4pA/√Hz that's 4.85nV/Hz, so the 5534A should have total voltage noise of 6nV/√Hz, so should be about 11dB quieter at 3852Hz. That agrees with the plots pretty well.
At 10kHz the figures are 31.5k reactive, 1612: 53nV, 5534A: 13nV, so 5534A should be 12dB quieter, again in agreement with the plots.
Interestingly the 500mH (made from 5 100mH inductors on a small breadboard), seems to be resonant with stray capacitance of the cables at around 16kHz, not unlike a real cartridge.
BTW a good JFET opamp like OPA1652 performs better than the NE5534A at top end, but only just - there's still some hump, probably due to LC resonance and the 47k load resistor's current noise.
Last edited:
Last week I cobbled together a pretty good differential input flat response phono preamp to use with the balanced line level inputs of my USB interface. I'm applying the RIAA EQ digitally. I found the low current noise LT1037 works well and it is in the 8 Pin DIP package. The 60 Hz line noise comes in right at the noise floor of the preamp, so it really works well. The added spice directives calculate the input and output noise levels and they are found in the output log file. CRTL L on LTSpice opens it. I pasted the example values in blue and calculated the approximate SNR based on the expected output of 3 mV from the cartridge. I use it with the Stereo Lab software from pspatialaudio.com to "rip" records to digital. The RIAA EQ can also be applied using equalizeAPO or other means of DSP filters to play them in real time. I built the schematic below without the 1 and 20 ohm resistors. It actually has a simple RC output, R17 and C4 , network just for the spice model, that produces the RIAA lowpass roll off to get the noise figure a bit closer to what would be there with the EQ added. I attached the spice models for this and a more complex version that actually includes the differencing. The noise measurement was performed with the configuration shown in my basement with the turntable plugged in, using the bench power supply and the laptop running from battery. I forgot to put the 0.5 uF power supply bypass caps in the schematic. With phono cartridge distortion in the .1% to 10% range, preamp distortion isn' t the major challenge.
Last edited:
Hello I am with Tom and Mark here.
The TI OPA 165x op-amps use a 3 stage format with feed-forward to cancel distortion, much like THX design see the Functional Block Diagram in the Data Sheet.
Yes real people do measure real circuits. See the plot attached of 3 OPA1656 in a Bugle clone by Audiowind A-300 ($40 on line)
A AT91 MM cartridge is in the mail to measure the real circuit with a real MM cartridge.
Thanks DT
Yes.
I can't speak for all opamps, but I do know that the LM4562/LME49720 uses good design, good layout, and also a modern analog-optimized semiconductor process to get the good performance. They also cranked up the bias current in strategic places to increase linearity.
The OPA1612 (and other Burr-Brown products) uses feed-forward if I remember right. Whether that's electrickery or good design is up to you.
Tom