I didn't. I was clarifying some gibberish. That has to happen every now and then in threads like this which alternate between performance and performance art.
Input shorted on MM is problematic. Bi-polar vs. FET inputs appear too much on an equal ground.
EDIT - On second thought you really short the inputs geez?!!
Last edited:
I think shorting the inputs to measure the intrinsic noise of an amplifier like this is valid. We know of course that connecting a cartridge will cause the noise to go up, but I think since every amp is tested like this by Stereophile, it's still a good reference to compare products. Maybe there could be a standardized input source defined like 300mH+200 ohms for MM and say 10 ohms for MC - but this would unfortunately disadvantage all subsequent technical reviews of similar products.
One thing I do like about the Stereophile approach is that equipment gets a subjective review and then it goes for a technical assessment, which by and large is very factual and objective (anyone remember that Zanden CD/D-A converter review in Stereophile from a few years ago?).
That said, if you are going to spend $10k or $60k on a preamp, you probably are looking for something else in your purchase - not just raw specs. They all look like very well built and designed products, even though I may not always agree with the design premises. Hell, if I had 10k . . . Naah, I am planning on making a trip to the South Pole in the next year or so, maybe after that!
One thing I do like about the Stereophile approach is that equipment gets a subjective review and then it goes for a technical assessment, which by and large is very factual and objective (anyone remember that Zanden CD/D-A converter review in Stereophile from a few years ago?).
That said, if you are going to spend $10k or $60k on a preamp, you probably are looking for something else in your purchase - not just raw specs. They all look like very well built and designed products, even though I may not always agree with the design premises. Hell, if I had 10k . . . Naah, I am planning on making a trip to the South Pole in the next year or so, maybe after that!
Last edited:
Scott, ok, I do not disagree that by shorting the input you are eliminating the current noise term, but in the absence of a standard you are going to end up with all sorts of results. (I suppose you could argue the same for shorting the input . . . ).
I did some simulation on this a while back and a c. 1nV/root Hz total input referred noise device (so about 1.4uV wide 20Hz - 20KHz noise input shorted) shows about 5.6uV noise with a 600mH/200 Ohm cart connected. Noise was just the max level so not converted to an equivalent w/band figure. Bipolar input device BTW. I suspect a fet input stage may have yielded better results at HF, but Vn of course probably higher.
Maybe John (Atkinson) would like to comment on the applicability of some sort of standard here.
I did some simulation on this a while back and a c. 1nV/root Hz total input referred noise device (so about 1.4uV wide 20Hz - 20KHz noise input shorted) shows about 5.6uV noise with a 600mH/200 Ohm cart connected. Noise was just the max level so not converted to an equivalent w/band figure. Bipolar input device BTW. I suspect a fet input stage may have yielded better results at HF, but Vn of course probably higher.
Maybe John (Atkinson) would like to comment on the applicability of some sort of standard here.
Pavel,
I measured it. NOTE that this is Ein (Input noise) specification referenced to 1V, a standard way of expressing noise in a system with widely variable gain (as SNR depends on gain and reference level).
For example, a "standard" MC Pickup has 0.5mV output at 14dBMax (this would be equivalent to 0dBFS on CD as we are nowadays used to full scale references, not a 0dB reference with 14dB Headroom) this level would be -66dBV.
So for this pickup and this reference level the (A weighted) SNR should be -145dBV -(-66dBV) = 79dB...
Actually, please read all of John's comments:
"AMR's website states that the PH-77 has a claimed level of input noise of –145dBV (0.056µV) .... Allowing for the difference in reference level, my figures confirm the manufacturer's specification."
To be honest, I do not.
From the above I notice that all you do with them is to misinterpret them (or the original specifications) and then make a big song and dance about this misinterpretation.
Ciao T
I measured it. NOTE that this is Ein (Input noise) specification referenced to 1V, a standard way of expressing noise in a system with widely variable gain (as SNR depends on gain and reference level).
For example, a "standard" MC Pickup has 0.5mV output at 14dBMax (this would be equivalent to 0dBFS on CD as we are nowadays used to full scale references, not a 0dB reference with 14dB Headroom) this level would be -66dBV.
So for this pickup and this reference level the (A weighted) SNR should be -145dBV -(-66dBV) = 79dB...
Actually, please read all of John's comments:
"AMR's website states that the PH-77 has a claimed level of input noise of –145dBV (0.056µV) .... Allowing for the difference in reference level, my figures confirm the manufacturer's specification."
To be honest, I do not.
From the above I notice that all you do with them is to misinterpret them (or the original specifications) and then make a big song and dance about this misinterpretation.
Ciao T
Well everyone, I am told that I don't know what I am doing, but my JC-3 phono preamp got 'Phonostage of the year' with still un-approved (here) IC's and one of my other parent companies got 'Preamp of the year' and 'Power amp of the year'. Not designed by me, but by my colleagues. Maybe, just maybe, we know something about audio design that you might learn from. You never know! '-)
One thing complicating phono stage specs is that cartridges are frequently rated for output at 1KHz and 0VU. Here 0VU is taken to be 5cm/sec horizontal, which is 3.54 cm/sec to each channel's generator. And then, as Thorsten points out clearly, the maximum output is 20 or even (almost) 30Db larger, and this doesn't compare directly to modern full scale referenced numbers.
Thanks,
Chris
Thanks,
Chris
Well John I think you do and I know the time it takes to do class A , Golden ear level work. Congratulations!
I agree that -145dBV looks nicer than -66dBV or 79dB. Anyway, the question is how realistic is such number in evaluation of the system noise. If you compare the noise to 1V voltage at the input, you get absolutely unrealistic number. What would be the output voltage of the phono stage when driven with 1V/1kHz at the input? 10 000V of output voltage at 1kHz? Or 9000 V?
Usually we are interested to know how much noise is at the output of the audio system, compared to the output reference voltage. That is the noise we would hear. Switching specs to input noise referred to 1V for the component which has maximum input voltage of some 2.5 mV is just confusing the customers.
Agreed. Impossible to omit R + jwL nature of the cartridge impedance. There is a nice noise analysis of phono stages in the Vogel's book.
Hi,
Are you sure you understood the specification and my example?
-66dBV or 0.5mV was the output of the example cartridge at -14dBMax (dBMax indicating the maximum rated output/signal level).
0dB for LP is 5cm/S peak (3.54cm/S RMS) velocity and maximum specified velocity (which may be however exceeded by some LP's cut) is specified as 25cm/S Peak (17.7cm/S RMS) and would correspond broadly to the 0dBfs in digital systems.
By simply subtracting the signal level in dBV (a trivial conversion to make I may add, for those who do not do it routinely in their head, this may help: Analog Devices: Interactive Design Tools: Utilities :) on the input of the phono-stage from the noise in dBV (hence the use of Ein in dBV and not the more common Studio standard dBu/dBm) you get the SNR for that reference level.
As I pointed out for a 0.5mV pickup at the LP 0dB reference level the rated Ein of our PH-77 Phono Stage solves to -79dB. This is what JA measured on the quieter channel.
It seems a tube became a little noisy so the left channel was below spec. I did check the records at the factory, it was to spec when the reviewed unit went through our testing.
I must note that dBV is a widely used, accepted and fully understood measure in electronics, without any ambiguity:
Wikipedia - dBV
dB(1 VRMS) – voltage relative to 1 volt, regardless of impedance.
So with clear specifications as given by AMR there is clearly no ambiguity whatsoever and system noise can be fully evaluated. This is not the case with the kind of specification we often find for competitors products.
For example, a competing product retailing at appx. 1/2 of the price of the PH-77 that was reviewed in Stereophile but not measured by JA has the following Spec:
"S/N ratio >90dB “A” weighted"
Note that neither the gain at which this applies (which can be adjusted from 44dB to 74dB) nor any reference level is given, so while this figure seems seful, it is in fact completely useless, as for all we know this SNR may be taken with 10V RMS out and 44dB Gain!
I do not understand what is unrealistic about the number. The specification of Ein in Audio Systems as either dBV (measurement systems - reference level 1V) or dBu/dBm (audio - reference level 0.775 or 1mW into 600 Ohm) has a long history in electronics and is absolutely realistic and makes sure two items rated thusly are always directly comparable with a minimum of Math.
If I have one Phonostage with -145dBV Ein and one with -135dBV Ein I know that the second Phonostage is 10dB noisier and thus will give me 10dB less SNR.
And as you see even with JA's measurements, as gain and reference levels are not fixed in Phono-stages there is no easy single number spec (such as in a CD-Player) that may be used, at least if a realistic and meaningful number is desired, other than Ein. Any other number is subject to massive levels of qualification and limits and two "output SNR" numbers for Phono Stages are rarely if ever directly comparable.
In our competitor analysis I spend a lot of time trying to get enough information to allow me to normalise for example the various SNR Specifications manufacturers give. Internally we use the same standard as Stereophile for AMR's R&D section - so I normalised to this where SP measurements where not forthcoming...
For a Phono-Stage the noise you will hear is actually the input referred amplified noise. The signal you will hear is the amplified input signal.
So the signal/noise ratio depends on the input noise and the input signal. Then Gain and output levels simply no longer form a part of the equation.
Again, this is really electronics 101 and nothing unusual or difficult. I would expect anyone who has worked professionally with Audio (design work not required) to know his decibels and volts (certainly I calculate stuff to the nearest dB routinely using not even pen/paper), but this of course is the 21st century when reasonable expectations are rarely if ever met...
To be honest, I am a bit peeved. Here a company actually gives a reliable, internationally recognised measure of the noise levels of their Phono Stage, which is instantly usable and meets those specs in independent measurements and as thank you gets beaten up over for misleading the consumer... Thank you very much.
Past that, to address some of the other issues raised, our input circuitry has extremely low current noise, so the cartridge impedance plays no role, other than for it's Johnson noise.
The high inductance in MM Cartrides makes that a bit of a problem, the old Pickerings with 1K/0.68H contributed considerable noise from their inductance, at 1KHz this already contributes 6dB excess noise over the 1K DCR. Though at -119dBV and with a -42dBV signal level this still leaves us 77dB SNR from the cartridge.
By comparison a 10 Ohm MC Cartridge (a common value for an 0.5mV output type, due to the physics and real limitations involved, though real cartridges vary considerably) for example has -145dBV Johnson noise for a 20KHz bandwidth.
Ciao T
Are you sure you understood the specification and my example?
-66dBV or 0.5mV was the output of the example cartridge at -14dBMax (dBMax indicating the maximum rated output/signal level).
0dB for LP is 5cm/S peak (3.54cm/S RMS) velocity and maximum specified velocity (which may be however exceeded by some LP's cut) is specified as 25cm/S Peak (17.7cm/S RMS) and would correspond broadly to the 0dBfs in digital systems.
By simply subtracting the signal level in dBV (a trivial conversion to make I may add, for those who do not do it routinely in their head, this may help: Analog Devices: Interactive Design Tools: Utilities :) on the input of the phono-stage from the noise in dBV (hence the use of Ein in dBV and not the more common Studio standard dBu/dBm) you get the SNR for that reference level.
As I pointed out for a 0.5mV pickup at the LP 0dB reference level the rated Ein of our PH-77 Phono Stage solves to -79dB. This is what JA measured on the quieter channel.
It seems a tube became a little noisy so the left channel was below spec. I did check the records at the factory, it was to spec when the reviewed unit went through our testing.
I must note that dBV is a widely used, accepted and fully understood measure in electronics, without any ambiguity:
Wikipedia - dBV
dB(1 VRMS) – voltage relative to 1 volt, regardless of impedance.
So with clear specifications as given by AMR there is clearly no ambiguity whatsoever and system noise can be fully evaluated. This is not the case with the kind of specification we often find for competitors products.
For example, a competing product retailing at appx. 1/2 of the price of the PH-77 that was reviewed in Stereophile but not measured by JA has the following Spec:
"S/N ratio >90dB “A” weighted"
Note that neither the gain at which this applies (which can be adjusted from 44dB to 74dB) nor any reference level is given, so while this figure seems seful, it is in fact completely useless, as for all we know this SNR may be taken with 10V RMS out and 44dB Gain!
I do not understand what is unrealistic about the number. The specification of Ein in Audio Systems as either dBV (measurement systems - reference level 1V) or dBu/dBm (audio - reference level 0.775 or 1mW into 600 Ohm) has a long history in electronics and is absolutely realistic and makes sure two items rated thusly are always directly comparable with a minimum of Math.
If I have one Phonostage with -145dBV Ein and one with -135dBV Ein I know that the second Phonostage is 10dB noisier and thus will give me 10dB less SNR.
And as you see even with JA's measurements, as gain and reference levels are not fixed in Phono-stages there is no easy single number spec (such as in a CD-Player) that may be used, at least if a realistic and meaningful number is desired, other than Ein. Any other number is subject to massive levels of qualification and limits and two "output SNR" numbers for Phono Stages are rarely if ever directly comparable.
In our competitor analysis I spend a lot of time trying to get enough information to allow me to normalise for example the various SNR Specifications manufacturers give. Internally we use the same standard as Stereophile for AMR's R&D section - so I normalised to this where SP measurements where not forthcoming...
For a Phono-Stage the noise you will hear is actually the input referred amplified noise. The signal you will hear is the amplified input signal.
So the signal/noise ratio depends on the input noise and the input signal. Then Gain and output levels simply no longer form a part of the equation.
Again, this is really electronics 101 and nothing unusual or difficult. I would expect anyone who has worked professionally with Audio (design work not required) to know his decibels and volts (certainly I calculate stuff to the nearest dB routinely using not even pen/paper), but this of course is the 21st century when reasonable expectations are rarely if ever met...
To be honest, I am a bit peeved. Here a company actually gives a reliable, internationally recognised measure of the noise levels of their Phono Stage, which is instantly usable and meets those specs in independent measurements and as thank you gets beaten up over for misleading the consumer... Thank you very much.
Past that, to address some of the other issues raised, our input circuitry has extremely low current noise, so the cartridge impedance plays no role, other than for it's Johnson noise.
The high inductance in MM Cartrides makes that a bit of a problem, the old Pickerings with 1K/0.68H contributed considerable noise from their inductance, at 1KHz this already contributes 6dB excess noise over the 1K DCR. Though at -119dBV and with a -42dBV signal level this still leaves us 77dB SNR from the cartridge.
By comparison a 10 Ohm MC Cartridge (a common value for an 0.5mV output type, due to the physics and real limitations involved, though real cartridges vary considerably) for example has -145dBV Johnson noise for a 20KHz bandwidth.
Ciao T
Dear Scott,
Agreed, we do need to account input current noise, however, if using J-Fets, and typical cartridge impedances the current noise contribution to system noise is negligible.
If anyone is misguided enough to use Bipolar Transistors on the input to an MM Cartridge then we have all sorts of trouble, not the least the actual base current of in input transistors which means means we must use a coupling Cap to keep that DC current out of the cartridge...
Ciao T
Agreed, we do need to account input current noise, however, if using J-Fets, and typical cartridge impedances the current noise contribution to system noise is negligible.
If anyone is misguided enough to use Bipolar Transistors on the input to an MM Cartridge then we have all sorts of trouble, not the least the actual base current of in input transistors which means means we must use a coupling Cap to keep that DC current out of the cartridge...
Ciao T
It's not true, Thorsten. IF you keep the betas high and the Iq low, the problem is minimized. THEN, if you use a lowish inductance cartridge like a Grado, or a SOTA Shure MM cartridge, the inductance is not so high. Third, the INTEGRATION due to the RIAA curve removes the slight increase in current noise that might be generated with frequency.
I don't think you can say there are no good RIAA amplifers using bipolars (MM input). Good JFETS perform very well in this role, but I would not count bipolars out. I also don't think a good quality coupling cap will trash an otherwise good design - besides, a failed input device could spell the end for an expensive cartridge - with static it's not unheard of.
Back to the earlier noise discussion, the best way forward I think is to use an agreed cartridge inductance and series resistance.
Back to the earlier noise discussion, the best way forward I think is to use an agreed cartridge inductance and series resistance.
Last edited:
Yes, I am very sure that I understand the specification and your example.
I also can make you sure that I know what is a dBV (re input). But is rarely used for the system with 1mV operating range (input). dBV re input is not widely used.
The proper definition of S/N I am attaching now. -145dBV re input only confuses the readers. Why not re 1kV for power amplifier? the same logic,
Attachments
Last edited:
- Status
- Not open for further replies.