OK, everyone, I don't get it. FM does not exist, frequencies, 40 Hz off are OK. Of course, higher order number fitting can be employed, if 40 Hz is too much, even for people who are familiar with the test methods, to maybe an error of +/- 5Hz. +/- 1 Hz or so should be practical with a retest, EVEN with the old, original equipment.
I found this interesting article about how much dynamic range is posible in measurements.
High Dynamic Range ? Fact or Fiction? | Prosig Noise & Vibration Measurement Blog
High Dynamic Range ? Fact or Fiction? | Prosig Noise & Vibration Measurement Blog
Scott,
I hope you'll let us know the results of your discussion with Pavel about the ADA4898-1.
I am sure many others would also want to know, and none of us would ever accuse AD of making a bad part.
It is in itself a good learning opportunity to understand why this happens at all.
Many thanks,
Patrick
I hope you'll let us know the results of your discussion with Pavel about the ADA4898-1.
I am sure many others would also want to know, and none of us would ever accuse AD of making a bad part.
It is in itself a good learning opportunity to understand why this happens at all.
Many thanks,
Patrick
So far i have used it an inverted configurations at the input of a phonostage. Actually i played an MC pre-pre with this chip for Jan Didden and it worked fine. What i know is, that when i used a resistor bigger then 300 Ohm from the output to the negative input ( Pin 2 ) i needed a small value compensation cap not bigger the 10pF or it starts to oscilate. It is a very unusual design.
The article conveniently overlooks such physical realities as thermal noise and linearity. I was talking to a designer of a "32" bit dac who admitted instantly that the thermal noise limited his design to 24 bits or less. You may get there with Josephson Junctions and liquid helium but would need the resources of a national laboratory to try it. Here is an example JVS 7000 PREMA: Primary Josephson Voltage Standard System that should be good to .1 ppm (better than 24 bits) or better but I think the max sample rate is in the samples per hour range.
I thought you had already agreed that a photocopier has funhouse optics and cannot be trusted to give any sort of linear graph, just a couple of pages back.
If there is anything to be found that graph is definitely not a trustworthy source, lets move on until an original can be scanned properly or the experiment can be repeated with digital measurement.
Wrinkle
Do you mean that the source switch is implemented into the phobo stage?
I'm talking about an output signal ~ 2Vrms for the phonostage - that's equal like from a CD-Player / DAC. Then, in theory you need only a passive pre - like a TVC - or buffer stage as line stage.
Advantage: better habdling of the signal voltage from the source
Or how do you lay out the input sensitivity of active linestages when you're dealing with input voltages around 2Vrms from line level sources? Do you attenuate the signal at the input?
In my system a gain of 60dB in the MC phono section is enough to play my tube amps loud so i rarely use a linestage but sometimes a buffer. The tube amp has a potmeter at the input so i do it the Japaneese way. Sofar i have not heard a linestage in my system that did not do SOMETHING to the sound. I do not say that it sounded bad, but ultimately transparency was chalenged in most cases. For example diffences in my various phonostages where swamped to a certain degree when they reached " very good quality". They sounded somehow all the same when they where good enough. Taking out the linestage restored diffences between them.
How do you qualify what a good phono stage should sound like , or do you do this subjectively with listening tests not comparing it with what is going into it.
I ask this because the gain stage in your phono stage has the potential also challenge the transparency, or is your previous statement only with regards to valve line stages.
Regards
Arthur
First and formost is RIAA acuracy. I check that with an anti RIAA i made with 0.1% toleranced components. I have a 0.1% bridge with Kevin leads and i have send 6 caps i measured in house to a Swedish institute that has a calibrated Wayne-Kerr bridge in a temperature controled room. My measurements have been confirmed to be accurate.
I do not implement the Neumann pole because i think it is a cutting artifact and there is no agreement where this pole is set. Some claim 2.2usec, 6dB octave, others claim a shorter time constant with a second order ( 12dB octave ) rolloff. Not all records are cut on Neumann lathes but rather on Ortophon or Westrex to further complicate the issue.
Most MC cartridges have a rising treble anyway and that is also dependent if you measure the cart on the outter radius of the record or the inner. Usually, measured more to the inner of the record, the peak gets smaller or disappears.
I check RIAA acuracy with a 1kHz suarewave fed over this Anti RIAA and measure level over a wide frequency range with an accurate wideband True RMS meter. I also measure digitally. I think that even small errors are audible and both channels should be as similiar as posible.
Second i design for lowest noise. 0.3nV is state of the art. With Fets you have a higher 1/F frequency though but there are ways to design around that too.
Third i design for low distortion although i am not convinced that a stage sounds better that has -100dB THD then a stage with -80dB distortion that has only second and third order harmonic distortion, preferably less third.
I also do a variety of IM and transient measurements that i have described in detail on my MPP thread.
Forth i design for a decent overload margin. Especially clicks and pops on the record can be very rich in transients.
Of cause i care about PSU, layout, crosstalk, component choices, etc.
Third i do subjective listening tests when this basic requirements have been met and of cause other people listen to my designs too. I participate in "listening contests" as John may call it. That happens on shows worldwide and also when a product is given for review. I regulary listen to as much of the competition as i can. Less now when my designs and i got more mature and i have found my way to interpret the world.
The only way to access a phonostage for transparency whould be to feed in a signal of known quality, maybe a good quality SACD player or a Master Tape, then Anti RIAA the feed, then adjusting for level differences and then listening compared to the straight bypass. That has being done in the past ( See Denis Collins in AudioXpress ) and transparency has been claimed. I doubt if a test like that mimics the real situation "When the Stone hits the Platic" as Salas used to say.
I do not implement the Neumann pole because i think it is a cutting artifact and there is no agreement where this pole is set. Some claim 2.2usec, 6dB octave, others claim a shorter time constant with a second order ( 12dB octave ) rolloff. Not all records are cut on Neumann lathes but rather on Ortophon or Westrex to further complicate the issue.
Most MC cartridges have a rising treble anyway and that is also dependent if you measure the cart on the outter radius of the record or the inner. Usually, measured more to the inner of the record, the peak gets smaller or disappears.
I check RIAA acuracy with a 1kHz suarewave fed over this Anti RIAA and measure level over a wide frequency range with an accurate wideband True RMS meter. I also measure digitally. I think that even small errors are audible and both channels should be as similiar as posible.
Second i design for lowest noise. 0.3nV is state of the art. With Fets you have a higher 1/F frequency though but there are ways to design around that too.
Third i design for low distortion although i am not convinced that a stage sounds better that has -100dB THD then a stage with -80dB distortion that has only second and third order harmonic distortion, preferably less third.
I also do a variety of IM and transient measurements that i have described in detail on my MPP thread.
Forth i design for a decent overload margin. Especially clicks and pops on the record can be very rich in transients.
Of cause i care about PSU, layout, crosstalk, component choices, etc.
Third i do subjective listening tests when this basic requirements have been met and of cause other people listen to my designs too. I participate in "listening contests" as John may call it. That happens on shows worldwide and also when a product is given for review. I regulary listen to as much of the competition as i can. Less now when my designs and i got more mature and i have found my way to interpret the world.
The only way to access a phonostage for transparency whould be to feed in a signal of known quality, maybe a good quality SACD player or a Master Tape, then Anti RIAA the feed, then adjusting for level differences and then listening compared to the straight bypass. That has being done in the past ( See Denis Collins in AudioXpress ) and transparency has been claimed. I doubt if a test like that mimics the real situation "When the Stone hits the Platic" as Salas used to say.
There is a Heisenberg limit that works out to 20bits @ 1GHz or so, I have the graph somewhere. Were you talking to Martin M.,a friend from back in the old days? I saw him last year and this year I will take them up on the offer to hear 32 bits in their listening room.
Last edited:
What is your opinion of the RIAA accuracy of the sources? Do you think any of the RCA's and Mercuries that the audiophiles love were better than 1%.? As a measurement tool alone you can create a digital inverse-RIAA that is better than .001 dB.
Joachim
let's assume your cartridge delivers an signal output of 0.4mV, this results with your 60dB gain of your phono section an output voltage ot 400mV.
Still we have a difference to the 2V of a line source. How do handle this, with your volume pot or so you use a trimpot at each input?
Ah, RIAA EQ. I use 1% or better components, and I commend you Joachim for making such an accurate inverse RIAA. I use the network approved by both Lipshitz and Jung, in the 30 year old derivation. Works for me.
For everyone else, it is important that SERIOUS DESIGNERS get the RIAA correct. This is because we don't want whatever sound quality that we get, to be attributed by a 'critic' as being an aberration in the frequency response. And they will, from experience, unless you get it right.
High frequency compensation, above 20KHz, is a long considered debate, and more has to be discussed than has been done here, to get a better 'compromise' as to how to handle it. For example, disc cutter resonance and Q, phono cartridge phase, and damping, phono stylus scanning losses, as well as pre-boost HF shelf. [Is it 40KHz, 50KHz, etc? You can't boost forever. ;-) ]
For everyone else, it is important that SERIOUS DESIGNERS get the RIAA correct. This is because we don't want whatever sound quality that we get, to be attributed by a 'critic' as being an aberration in the frequency response. And they will, from experience, unless you get it right.
High frequency compensation, above 20KHz, is a long considered debate, and more has to be discussed than has been done here, to get a better 'compromise' as to how to handle it. For example, disc cutter resonance and Q, phono cartridge phase, and damping, phono stylus scanning losses, as well as pre-boost HF shelf. [Is it 40KHz, 50KHz, etc? You can't boost forever. ;-) ]
Last edited:
- Status
- Not open for further replies.