Most moving-magnet phono amplifiers have a 47 kohm resistor across their input, which adds 0.5869 pA/√Hz of thermal noise current at 20 degrees Celsius. On top of that, there is the thermal noise associated with the iron losses of the cartridge, although those are fairly low in the cartridge PMA usually uses.
I too had expected the high-frequency noise of the PMA OP27 to be a little louder.
Certainly my measured value is higher than the DS value (0.4p) even with common mode cancellation, but I don't think it is too high, especially for a voltage noise density of 3 nV/√Hz with bias cancellation.
If I have time, I will actually attach it to a phono EQ circuit at a later date and measure the noise spectrum.
However, my MM carts are old V15 type III & IV (both almost the same 1.38kΩ + 500mH), so I think they have a higher impedance than the M-35. So the difference in current noise may be more significant.
That's clear.
Just for curiosity it would be interesting to see mason_f8's and PMA's test set-up of DUT's, to see why results are not inline. Here I mean op27 as thread is calling for, in masons test current noise is not fancy, in PMA 's test it's overall performance is almost inline with opa637...
I'm believer that it's proof of puding to test dut in real intended circuit to grasp whole thing that matters, even admitibly testing one specific item also has merit, just it is not giving whole picture of intended performance.
Just for curiosity it would be interesting to see mason_f8's and PMA's test set-up of DUT's, to see why results are not inline. Here I mean op27 as thread is calling for, in masons test current noise is not fancy, in PMA 's test it's overall performance is almost inline with opa637...
I'm believer that it's proof of puding to test dut in real intended circuit to grasp whole thing that matters, even admitibly testing one specific item also has merit, just it is not giving whole picture of intended performance.
Hi, sorry, as I posted my comment yours just arrived.
Great if you do that, but don't do it for me, I'm just curious.. Whenever is perfect 😊
The Shure V15-III has more iron losses, though. The phase of its impedance doesn't go above about 72 degrees, according to the measurements of a former colleague of mine.
I took the measurements when I had the time.
The gain of the Phono AMP is 40dB (@1kHz), but it was subsequently amplified by 1000x and measured, so mV/√Hz on the scale should be read as μV/√Hz. (In the Analogue Discovery graphs, the /√ is omitted and the value is written as VHz).
Cart used is V15 type-III (measured DCR=1.403kΩ, inductance 555.3mH) Cart load is 47kΩ//190pF (including cable)
*It would be boring if I broke the stylus because of this, so I removed the stylus and took measurements. Inductance was reduced to 550.9mH.
Orange is OPA 627.
Blue is LT1028.
Brown is AD797.
Green is OPA27
As expected, as with the PMA results, they are close to OPA627.
I added yellow because I thought that the one with a bit more noise is LME49710 (DS value 1.2 pA/√Hz, actual measurement by my LM4662 1.25 pA/√Hz)
Other additional measurements from my interest.
The orange OPA627 is left as a reference for comparison.
Cyan is NE5534 (TI); Slightly noisier than the FET input at high frequencies, but very good for a bipolar.
Grey is OPA134; the high noise in the mid-range is probably due to the relatively high voltage noise; above 10kHz it almost overlaps with the OPA627. (It may just be that the GB product is small and the response is reduced.)
Purple is the OPA1677; OPA1678 is one of my favourite OP-AMPs because it is inexpensive and I am glad that a single has been released.
Red is the OPA1655; which has the characteristics of a MOS input with a large 1/f but the lowest noise in the mid to high frequency range.
Yellow-green is OPA828; almost overlapping with OPA627.
Magenta is OPA1641;
The table at the top right of the graph shows noise levels measured at the output of a phono amp at 40 dB (@1 kHz). This is measured with A-weighting + 20 kHz LPF.
Taking the output of the cart as a 5mV reference (although the Type-III is 3.5mV), the output is 500mV (-6dBV), so for example, for an OPA627 (-84.6dBV) the effective S/N with the cart attached is 78.6dB.
Would this be a factor that would not be much improved by changing from 47kΩ to an electronic load?
The gain of the Phono AMP is 40dB (@1kHz), but it was subsequently amplified by 1000x and measured, so mV/√Hz on the scale should be read as μV/√Hz. (In the Analogue Discovery graphs, the /√ is omitted and the value is written as VHz).
Cart used is V15 type-III (measured DCR=1.403kΩ, inductance 555.3mH) Cart load is 47kΩ//190pF (including cable)
*It would be boring if I broke the stylus because of this, so I removed the stylus and took measurements. Inductance was reduced to 550.9mH.
Orange is OPA 627.
Blue is LT1028.
Brown is AD797.
Green is OPA27
As expected, as with the PMA results, they are close to OPA627.
I added yellow because I thought that the one with a bit more noise is LME49710 (DS value 1.2 pA/√Hz, actual measurement by my LM4662 1.25 pA/√Hz)
Other additional measurements from my interest.
The orange OPA627 is left as a reference for comparison.
Cyan is NE5534 (TI); Slightly noisier than the FET input at high frequencies, but very good for a bipolar.
Grey is OPA134; the high noise in the mid-range is probably due to the relatively high voltage noise; above 10kHz it almost overlaps with the OPA627. (It may just be that the GB product is small and the response is reduced.)
Purple is the OPA1677; OPA1678 is one of my favourite OP-AMPs because it is inexpensive and I am glad that a single has been released.
Red is the OPA1655; which has the characteristics of a MOS input with a large 1/f but the lowest noise in the mid to high frequency range.
Yellow-green is OPA828; almost overlapping with OPA627.
Magenta is OPA1641;
The table at the top right of the graph shows noise levels measured at the output of a phono amp at 40 dB (@1 kHz). This is measured with A-weighting + 20 kHz LPF.
Taking the output of the cart as a 5mV reference (although the Type-III is 3.5mV), the output is 500mV (-6dBV), so for example, for an OPA627 (-84.6dBV) the effective S/N with the cart attached is 78.6dB.
I measured the impedance and it is almost right at 73° at 3 kHz. (This is measured with a stylus attached)
Would this be a factor that would not be much improved by changing from 47kΩ to an electronic load?
Attachments
Last edited:
Very nice set Mason!
Funny to see how dinosaur Op27 is close to mighty OPA627 in real circuit, but also that newish OPA828 lives to the promise. Actually very nice guide what to use.
Of subject, but couldn't not to notice, huge bump at 50hz like in all single ended preamps, go differential input and get rid of that for good 😊, just couple of db random noise penalty to erase this and any other external common mode interference.
Funny to see how dinosaur Op27 is close to mighty OPA627 in real circuit, but also that newish OPA828 lives to the promise. Actually very nice guide what to use.
Of subject, but couldn't not to notice, huge bump at 50hz like in all single ended preamps, go differential input and get rid of that for good 😊, just couple of db random noise penalty to erase this and any other external common mode interference.
Your measurements are pretty close to what my former colleague Richard Visee measured in the 1990's, although in his case, the self-resonance was higher, around 67 kHz and 124 kohm.
In any case, the smaller the phase of the cartridge impedance, the lossier the impedance, the more thermal noise the cartridge generates and, relatively, the less advantage you get from a low-noise termination impedance. (All without a record playing, of course.)
Richard's measurements and my calculations can be found on page 43 of the October 2003 issue of Electronics World, see "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.
The table below shows the impedance as measured by Richard and the impedance as calculated for 1338.8 ohm in series with an ideal 460 mH inductance. The magnitude is close, but the phase is not.
R is the real part of the measured impedance, G the real part of the reciprocal of the measured impedance and F is how much you worsen the thermal noise density when you terminate the cartridge with an ordinary 47 kohm resistor instead of a noiseless 47 kohm resistor.
Last edited: