A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

[mediatechnology]

Loaded the prototype back on the bench and got an accurate measure of gain which is 62 dB.

Test conditions:

Rs = 3.3Ω
Rg = 1Ω
Rfb = 316Ω (X2)
Gain = 56 dB
Cross-coupled output gain 6 dB
Total Gain = 62 dB


ZTX851 Preamp NJM2068DD Gain 62dB Noise Floor

Results:

Output noise measures -80.96 dBu. (20 kHz BW)
Subtract 0.7 dB for converter loading and we get -80.26 dBu.
Referred to the input the Ein is -142.26 dBu.
That's an Rnv of 11Ω.
4.3Ω of that 11Ω is Rs and Rg.
Add another 4Ω estimated for 2X transistor rbb and that leaves 2.7Ω Rnv "unaccounted for."
NF is approximately 5.23 dB.
Noise density works out to be a about 0.43 nV√Hz.

The output noise is about the same at this gain using an NE5532, NJM2068 or OPA1612.

There's still some magnetically induced hum which is reduced by physical nulling.
It will undoubtedly be possible to actually pickup a fraction of a dB if this were in an actual box, not an open ProtoBoard so please don't degrade this into "where's the missing dB or Protoboards are shit."
I'm in no mood for it - it's time to move on with a layout.

I moved the servo sense point from the output of the cross-coupled stage to the outputs of the previous stage.
The reasoning for that is someone could run the board without having to install both THAT1240s.
In fact using the bridge circuit that defines CM voltage and sampling one stage upstream allows the board to run without any 1240s.
This also allows the servo op amp to be removed for troubleshooting or transistor matching.
Better that the differential Deboo, which has a common mode gain of "1" not be "blind" to the actual CM voltage and have to have its CM reference defined by the THAT1240s.
The disadvantage is that the servo does not correct the millivolt-level offsets of the THAT1240s themselves.
A jumper - providing options for both - might be worthwhile.

[billshurv]

With my two lowest output carts that would give me 28dB margin running straight into the miniDSP inputs, but should be spot on if my friend ever gets around to delivering the TI PCM4222EVM he has promised. Handy.

Noise looks good to me. We all love a phono stage noise drag racing w*nkfest. But for practical purposes I see nothing audibile there unless I score one of the legendary pathologically low output carts out there.

[mediatechnology]

Thanks Bill.

Phono preamps can be a w*ankfest indeed.

I think we should all order some ZTX851 from Mouser just to have in our stash.
They're the 2SD786s that we didn't buy 30 years ago when they were readily available.

Right now Mouser is out of stock but expect about 3600 pieces at the end of the month.
They're about $66 for 100 pieces.
Digi-Key have them in stock but are more expensive.

Out of that 100 pieces there are a lot of MC preamps that could be made.

[billshurv]

I did get a few for experimentation as I thought they might work in Richard Lee's common base headamp as well. But I will need more for some other things I now want to try. But a lifetime buy for me is only about a dozen depending on how many I need for matching pairs. It may end up cheaper for me to buy matched sets for this off you (if you end up offering that). I have a small stash of BCF862 but at least a couple of similar ones are available from on-semi that are only a nadge noisier. LSK170 I won't bother stocking up on.

Since I got a focusrite scarlett been also looking at phantom powered headamp options for that just to up my heresy score amongst audiophiles. That may use up my stash!

[mediatechnology]

I just made some measurements with a 10Ω Rsource.

Test Conditions:

Rs = 10Ω
Rg = 1Ω
Rfb = 316Ω (X2)
Gain = 56 dB
Cross-coupled output gain 6 dB
Total Gain = 62 dB

Results:

Output noise measures -78.96 dBu. (20 kHz BW)
Subtract 0.7 dB for converter loading and we get -78.26 dBu.
Referred to the input the Ein is -140.26 dBu.
That's an Rnv of 17.5Ω.
11Ω of that 17.5Ω is Rs and Rg.
Add another 4Ω estimated for 2X transistor rbb and that leaves 2.5Ω Rnv "unaccounted for."
NF is approximately 2.4 dB.
Noise density works out to be a about 0.53 nV√Hz.

I'll repeat that there is some hum contamination and these measurements were made on an open ProtoBoard.

I did a quick CMRR ratio test.

The bias resistors, 499Ω were not selected and were 1% tolerance.
The source resistors, 10Ω per leg, were matched to within 0.03% using a 4 wire measurement.
The CMRR is about 57 dB.
Most of what I'm seeing is pure resistor mis-match.

Also checked the servo sample point.

The servo can sample at the final balanced output or at the first stage outputs prior to common mode rejection.
In the MC preamp where gain isn't switched its better to have the final output servo'd so that the offsets of the THAT1240 are also servo'd.
When the final output is servo'd the differential offset, at the final output, is about 250 uV.

In a mic preamp where gain is switched the 1240 errors can become large relative to the Vbe*Gain error at low gains.
It's better to sample at the outputs of the first stage, before CM rejection to minimize the DC voltage across Rg.
When the first stage outputs are servo'd the differential offset, at the op amp outputs is about 500 uV.

When the servo is in either position, at 62 dB gain, I measure <<20uV offset across Rg.

[billshurv]

I was mulling this last night and the lowest output of the non-insane cartridges out there are 0.09mV and between 1.5R and 3R. I have one of these, current on long term loan to a cartridge and mulling getting one more as and when one pops up. Looking at your numbers I reckon you are still well below vinyl surface noise on the proto board with this but just about on the limits of hearing noise with the arm raised.

[terkio]

The servo can sample at the final balanced output or at the first stage outputs prior to common mode rejection.
In the MC preamp where gain isn't switched its better to have the final output servo'd so that the offsets of the THAT1240 are also servo'd.
When the final output is servo'd the differential offset, at the final output, is about 250 uV.

When the first stage outputs are servo'd the differential offset, at the op amp outputs is about 500 uV.

When the servo is in either position, at 62 dB gain, I measure <<20uV offset across Rg.

I would expect much lower measured servo errors.
With the 2277 specs at 10 uV voltage input offset , your circuit has the potential of a 10 uV error, where you measure 250uV or 500uV.
This can come from not enough gain in the servo loop and unperfect servo resistors matching.
You can make a servo with gain.
Presently, you have gain 2 with 100K/100K and 20K/10K.
You can make gain 100 with 1K/100K and 2K/100K. This makes a servo loop 100 times more accurate. Speed will be 100 times faster too. Can be slowed down with a higher cap.
Have a look to servo resistors to see how the error is sensible to resistor matching.
The errors you measured are certainly good enough, but there is something fishy.

[mediatechnology]

I think it's good enough as-is.
We don't obsess over little things here which is why I avoid consumer-oriented fora altogether now.
Particularly consumer-oriented fora discussing phono preamps...

The overall servo loop gain is indeed two so half the delta-Vbe error of the transistors is reflected in the output.
(There is a gain of 2 in the 1240 cross-coupled stage. That's why moving the sample point to the first stage doubles the Vos.)
Note that the uncorrected offset error is not subject to deltaVbe * Gain; It is deltaVbe/2.
That particular un-selected pair (in an M7 nut) has about 500uV offset.

As it currently is the servo forms a high-pass response at about 7 Hz.
I don't think I'd want to listen to it if the -3dB point was 700 Hz.
Though I could add DC gain (lowering servo Fc to match) and bury that DC in 1/f noise to reduce Vos further why bother?
Would the high-gain servo then require a passive pole on it's output to reduce it's noise?
It might. Then you have a 2 pole servo response which can peak.

The op amps in the A/D that follow it are going to have more offset than transistor Vbe mis-match.
That's what the converter's HP filter is for or maybe something novel like a coupling cap could be used to shave off the sub-millvolt Vos.

I was considering moving the servo about an octave higher though because it then provides a nice first-order high pass.

I did add a couple of resistors to the sch for stability with slower op amps such as the NJM2068.
More later.

Now if you really want something to worry about I can shift the Vos by about 50 uV by shining a flashlight on the transistors.

[billshurv]

I think it's good enough as-is.
We don't obsess over little things here which is why I avoid consumer-oriented fora altogether now.
Particularly consumer-oriented fora discussing phono preamps...

My Epiphany came when someone pointed out that, by the time you have added RIAA de-emphasis and A-weighting to the noise measurements then you requirements are considerable less than people copy/paste around the internet. CMRR to kill hum is far more important than the last 0.1dB on NF in real use.

[mediatechnology]

CMRR to kill hum is far more important than the last 0.1dB on NF in real use.

Exactly.
Or the last mV of offset.
I'd rather have the CMRR.

I'm thinking about a way for the builder to Vbe match using the preamp as a jig.
Considering two rows of pads for each transistor's CBE connections.
One pair of rows would have machined pin sockets installed for test.
The other pair of rows would be to hard-solder them once selected and glued/"heatsinked."