The Truth and Fairy Tales of S\NR in Phono preamps (with multilanguage subtitles)
or
9 dB of deception from the the "input shorted" on FETs and 14 dB of fraud from the "input shorted" on bipolars
Nick Sukhov masterclass on Cy-XXI/TEAC PE-505/TEAC PA-B10/Accuphase C47/iFi ZENAirPhono/X-Altra/The Grail SE+
Full version with circuit diagrams and info files is here! [ https://www.patreon.com/posts/9-db-obmana-ot-i-90332386 ]
more about "noise input shorted" https://www.patreon.com/posts/68014370
more about highest quality phono preamps https://www.patreon.com/posts/73044002
or
9 dB of deception from the the "input shorted" on FETs and 14 dB of fraud from the "input shorted" on bipolars
Nick Sukhov masterclass on Cy-XXI/TEAC PE-505/TEAC PA-B10/Accuphase C47/iFi ZENAirPhono/X-Altra/The Grail SE+
Full version with circuit diagrams and info files is here! [ https://www.patreon.com/posts/9-db-obmana-ot-i-90332386 ]
more about "noise input shorted" https://www.patreon.com/posts/68014370
more about highest quality phono preamps https://www.patreon.com/posts/73044002
Attachments
Last edited:
Testing a phono stage without the proper input termination, is like testing a power amplifier without a load.
Meaningless.
Meaningless.
Your 1.1nV /rt Hz claim about the X-Altra in your video is not correct.
The X-Altra MC amplifier only input noise contribution is ~250 pV/rt Hz with a 12 Ohm Cart and <350 pV/rt Hz with a 30 Ohm cart. Below 12 Ohms, it's even quieter. See my sim results in the other thread. As I've explained a few times, on MC you can separate out the cart noise contribution from the amplifier since there is very little or no current noise contribution because the source resistance is very low. You are only going to get the noise performance you have with the X-Altra MC if you parallel many good JFETs - 10 or 20 devices with <1nV/rt Hz.
You are getting the figures you quote because you do not understand how the TIS input stage works. There is no 3.5 Ohm resistance to ground, which you insist on inserting into your models, thus getting erroneous sim results. As I have tried to explain numerous times, input resistance and resistance to ground are not the same when discussing noise. There is no divider action - the amplifier equivalent thermal input noise resistance appears in series with the cart thermal noise - and above ~3 Ohms, the cart will be generating most of the noise - certainly, this is the case with a 10 or 12-ohm cart resistance.
On MM inputs, the input noise current quickly becomes the dominant noise mechanism in bipolar input stages. On JFET input designs, current noise is normally not a factor, and only the JFET noise voltage has to be considered, but this is often higher than low-noise bipolar devices. As I explained in my audioXpress article, the LP797 noise with the input shorted was < 1nV/rt Hz vs 1.6nV/rt Hz for the X-Altra MM input. However, when a cartridge (1350 + 500mH) was connected the noise difference was 0.2 dB but the X-Altra has >30 dB overload vs 12 dB for the LP797. The thermal noise of the MM cart in parallel with the 47k (or whatever you use) limits the ultimate SNR to 75-76 dB. So, you cannot claim 86 dB ref 5mV for any MM amplifier.
The X-Altra MC amplifier only input noise contribution is ~250 pV/rt Hz with a 12 Ohm Cart and <350 pV/rt Hz with a 30 Ohm cart. Below 12 Ohms, it's even quieter. See my sim results in the other thread. As I've explained a few times, on MC you can separate out the cart noise contribution from the amplifier since there is very little or no current noise contribution because the source resistance is very low. You are only going to get the noise performance you have with the X-Altra MC if you parallel many good JFETs - 10 or 20 devices with <1nV/rt Hz.
You are getting the figures you quote because you do not understand how the TIS input stage works. There is no 3.5 Ohm resistance to ground, which you insist on inserting into your models, thus getting erroneous sim results. As I have tried to explain numerous times, input resistance and resistance to ground are not the same when discussing noise. There is no divider action - the amplifier equivalent thermal input noise resistance appears in series with the cart thermal noise - and above ~3 Ohms, the cart will be generating most of the noise - certainly, this is the case with a 10 or 12-ohm cart resistance.
On MM inputs, the input noise current quickly becomes the dominant noise mechanism in bipolar input stages. On JFET input designs, current noise is normally not a factor, and only the JFET noise voltage has to be considered, but this is often higher than low-noise bipolar devices. As I explained in my audioXpress article, the LP797 noise with the input shorted was < 1nV/rt Hz vs 1.6nV/rt Hz for the X-Altra MM input. However, when a cartridge (1350 + 500mH) was connected the noise difference was 0.2 dB but the X-Altra has >30 dB overload vs 12 dB for the LP797. The thermal noise of the MM cart in parallel with the 47k (or whatever you use) limits the ultimate SNR to 75-76 dB. So, you cannot claim 86 dB ref 5mV for any MM amplifier.
1) I know what is TIA [ https://en.wikipedia.org/wiki/Transimpedance_amplifier ] , but yours X-Altra MC pre is not a TIA. Because it does not have feeback loop and not 0 virtual input. X-Altra MC is simply common base preamp with low (but not null) input resistance, not TIA.
2) I do not connect 3.5 resistance to ground, this is an ordinary common base input resistane of ZTX itself.
3) МС head EMF produce a voltage, and only then it translates to currents. This does not change the requirement that RTI noise be brought to the EMF input in the form of voltage rather than current.
4) my MM preamp has "passive cooling" 150 kOhm input resistor, not 47 kOhm, and therefore is much better than as you claim "The thermal noise of the MM cart in parallel with the 47k (or whatever you use) limits the ultimate SNR to 75-76 dB. So, you cannot claim 86 dB ref 5mV for any MM amplifier." . Measured SNR better than 84 dB is here: https://www.patreon.com/posts/metrologiia-s-sh-89604378
Last edited:
A TIA does not exclusively require feedback to be called a TIA. Its simply a circuit that converts a current input to a voltage output either directly or by using a suitable load resistor. But if you are more comfortable calling it a common base input then that's fine. Some call it 'current injection'.
Here is the stepped plot of the X-Altra MC 3 Ohms to 18 Ohms. The total RTI noise with an 18 Ohm Rcart is 640 pV/rt Hz. The thermal noise of just the 18 Ohms is 545pV/rt Hz. With a 12 Ohm Benz cart and set for 5mV output, the total RTI noise is 530 pV/rt Hz. Even with the test input resistance of 47 Ohms (880pV/rt Hz), the total system noise is 918pV/rt Hz. Where do you get 1.1nV/rt Hz?
Your 84 dB SNR claim does not state what your source is (L + R), whether 'A' weighting is used, and what the resistor cooling contribution is. Without these figures, a performance comparison is not possible.
Here is the stepped plot of the X-Altra MC 3 Ohms to 18 Ohms. The total RTI noise with an 18 Ohm Rcart is 640 pV/rt Hz. The thermal noise of just the 18 Ohms is 545pV/rt Hz. With a 12 Ohm Benz cart and set for 5mV output, the total RTI noise is 530 pV/rt Hz. Even with the test input resistance of 47 Ohms (880pV/rt Hz), the total system noise is 918pV/rt Hz. Where do you get 1.1nV/rt Hz?
Your 84 dB SNR claim does not state what your source is (L + R), whether 'A' weighting is used, and what the resistor cooling contribution is. Without these figures, a performance comparison is not possible.
Nikolay Sukhov is very much concerned about the commercial promotion of his phono-preamp. That 's why he publishes all his topics on this forum . 😉 And he is trying to fight with competitors .
It would be wonderful if he did that without making claims about other people’s efforts that were completely untrue.
😊
😊
You are wrong, and you are not clairvoyant
. I just try not to violate Ohm's, Kirchhoff's and Nyquist's laws 😉L+R are standard 500 mH 1 kOhm, IEC-A weghting output (and also ITU-R weghting) ref 5 mV 1 kHz @ MM EMF node, passive cooling input 150 kOhms. Full Microcap circuit file and PCB gerbers is here [ https://www.patreon.com/posts/skhema-gerbery-i-86997687 ] . Real measurement metrology is here [ https://www.patreon.com/posts/metrologiia-s-sh-89604378 ]. Real listening is here [ https://www.patreon.com/posts/proslushka-cy-v-89311755 ]
In my opinion, extremely high signal-to-noise ratios in phono RIAA pre-amplifiers no longer provide any advantages above a certain value (maybe from -55 to -60dB).
check out this thread:
https://www.diyaudio.com/community/...e-noise-i-want-to-know-all-approaches.296358/
Of course, this is different with a microphone pre-amplifier because every additional db improvement in SNR also increases the dynamic range.
check out this thread:
https://www.diyaudio.com/community/...e-noise-i-want-to-know-all-approaches.296358/
Of course, this is different with a microphone pre-amplifier because every additional db improvement in SNR also increases the dynamic range.
The old IHF standard A-202 (now CEA-490) required that the preamp noise be measured with a standard cartridge terminated with a standard load. The standard cartridge was 500 mH in series with 1 kOhm, (Shure M-75) and the standard load was 47 kOhms paralleled by a 125 pF capacitor. A noise simulation of the standard cartridge and load gives an A weighted noise of about 430 nV over the 20-20kHz bandwidth. That is about 78 dB below the standard reference level of 5mV at 1 kHz. That is about the best you can do with a noiseless preamp. Now Nick claims that with a non-standard cartridge and load he can get an SNR of 85 dB and it is the best. If I recall correctly, the SNR for a quiet record grove is about 60 dB. So what do you get for that extra 25 dB of SNR?
I'm getting 82db from my preamp.
20db lower then groove noise.
Bjt front end op amp second stage.
20db lower then groove noise.
Bjt front end op amp second stage.
Resistor cooling will give you 2 dB best case with a bip input. I suspect this is less with JFETs because there is little or no current noise component to cool out
(Self explored resistor cooling in ‘Small Signal Analog Design’ and you can see the cooling contribution for both bipolars and JFET devices in his write up)
With A weighted measurement, you can generally assume about 4 dB. I don’t usually bother with weighting as I prefer dealing in absolute numbers.
(Self explored resistor cooling in ‘Small Signal Analog Design’ and you can see the cooling contribution for both bipolars and JFET devices in his write up)
With A weighted measurement, you can generally assume about 4 dB. I don’t usually bother with weighting as I prefer dealing in absolute numbers.
Shouldn't you also be giving that noise an RIAA deemphasis? I think I'm more interested in that than the A weighted result.
Yes, you are right. All noises must be equalized by both RIAA and IEC-A (or ITU-R) weighting, otherwise the result will be greatly distorted and not true. the truth is that:
"Shouldn't you also be giving that noise an RIAA deemphasis? "
I did. The A weighting is just because the IHF standard requires it.
I claim that you just select the load for a Q of 0.6 and forget the frequency response and noise. Vinyl records only have an SNR of about 60 dB, maybe 70 dB if they used metal mastering. There will be more attenuation at the high end, but there isn’t much music out there so I don’t care. Why a Q of 0.6? That’s because it’s almost phase linear. If you look at the Q of a two pole phase linear Bessel filter, the poles require a Q of 0.57, so 0.6 is close enough. You have to select the load for more than FR and SNR.
I did. The A weighting is just because the IHF standard requires it.
I claim that you just select the load for a Q of 0.6 and forget the frequency response and noise. Vinyl records only have an SNR of about 60 dB, maybe 70 dB if they used metal mastering. There will be more attenuation at the high end, but there isn’t much music out there so I don’t care. Why a Q of 0.6? That’s because it’s almost phase linear. If you look at the Q of a two pole phase linear Bessel filter, the poles require a Q of 0.57, so 0.6 is close enough. You have to select the load for more than FR and SNR.
Moving on, this is the wideband raw input noise that the RIAA preamp will see - this does not include any amplifier noise voltage or noise current contribution. For a typical input stage (JFET or low noise bip), the total input noise will be around 3.1uV. Obviously, this noise is then severely shaped by the RIAA EQ.
This severely shaping is -20 dB @ 20 kHz (ref 1 kHz) thanx to RIAA and -10 dB by IEC-A wtng (or even -20 by ITU-R). Total -30...40 dB is too much to ignore.