Fun with vinyl measurements

[undeakyetonde]

Maybe @Michael Fidler would be so kind to expand on dynamic range of records? How could one verify the below quote?https://michaelfidler.com/products/pro/mc-pro-moving-coil-phono-stage/

Many transfers from high quality vinyl records have shown a signal-to-noise ratio relative to 5cm/s that reaches 72dB, so it's very important the the phonostage beats this figure by a further 6dB. This makes sure that when the noise of the phonostage sums with the surface noise of the record, the overall SNR doesn't drop by more than a decibel, the smallest perceptible change.

Let's say I have a reference 5 cm/s 1 kHz sine, for example -20 dB RMS in digital realm. I also have a record digitised (same setup). What do I compute to get a record's SNR relative to 5cm/s?

 

[Balle Clorin]

 

[Michael Fidler]

Maybe @Michael Fidler would be so kind to expand on dynamic range of records? How could one verify the below quote?https://michaelfidler.com/products/pro/mc-pro-moving-coil-phono-stage/


Let's say I have a reference 5 cm/s 1 kHz sine, for example -20 dB RMS in digital realm. I also have a record digitised (same setup). What do I compute to get a record's SNR relative to 5cm/s?

These are based on some A-weighted and 220Hz to 22kHz measurements of LP surface noise that I made I while ago, referenced to peak levels -3dB to approximate RMS. Some very good ones could produce 72dB or so.

If you wanted to make a measurement referenced to 5cm/s, then if it's set to -20dB I would take the noise floor, either A-weighted or 220Hz to 22kHz and deduct the -20dB reference point.

 

[undeakyetonde]

These are based on some A-weighted and 220Hz to 22kHz measurements of LP surface noise that I made I while ago, referenced to peak levels -3dB to approximate RMS. Some very good ones could produce 72dB or so.

If you wanted to make a measurement referenced to 5cm/s, then if it's set to -20dB I would take the noise floor, either A-weighted or 220Hz to 22kHz and deduct the -20dB reference point.

Does this mean that you had an LP capture where noise floor RMS (silent groove?) was whole -72 dB lower than a reference 5 cm/s RMS track from the test LP played on the same setup [with both signals band limited to 220 Hz to 22 kHz]? Wow.

Are details of 220Hz-22kHz bandlimiting of any importance? Would a pair of brickwall LP/HP filters in an audio editor do?

---

(Ortofon Test LP #5 track appears to be 5 cm/s *peak*, so in the Waxwing VM95ML post the -20 dB *RMS* digitised 1 kHz sine should correspond to 3.54 cm/s RMS 1 kHz sine)

 

[Michael Fidler]

Does this mean that you had an LP capture where noise floor RMS (silent groove?) was whole -72 dB lower than a reference 5 cm/s RMS track from the test LP played on the same setup [with both signals band limited to 220 Hz to 22 kHz]? Wow.

Are details of 220Hz-22kHz bandlimiting of any importance? Would a pair of brickwall LP/HP filters in an audio editor do?

---

(Ortofon Test LP #5 track appears to be 5 cm/s *peak*, so in the Waxwing VM95ML post the -20 dB *RMS* digitised 1 kHz sine should correspond to 3.54 cm/s RMS 1 kHz sine)

Not quite that, I'm afraid - just peak -3dB of one of the tracks was 72dB above the noise floor. I need to do some proper measurements in the future, but noise tends to vary depending on the pressing itself as well. Some hiss noticeably, others not so much...

As LF noise increases below 200Hz, it's important to weight the measurement to filter it out. We're not really very sensitive with our hearing below this point in any case, so if we make a flat measurement, it will be distorted in the negative direction.

 

[undeakyetonde]

Not quite that, I'm afraid - just peak -3dB of one of the tracks was 72dB above the noise floor.

Sorry, I don't understand, where does the highlighted part come from then?

Many transfers from high quality vinyl records have shown a signal-to-noise ratio relative to 5cm/s that reaches 72dB, so it's very important the the phonostage beats this figure by a further 6dB. This makes sure that when the noise of the phonostage sums with the surface noise of the record, the overall SNR doesn't drop by more than a decibel, the smallest perceptible change.

To my best knowledge, peak velocity (e.g. during a transient) is likely to be higher than 5cm/s...

 

[Michael Fidler]

Sorry, I don't understand, where does the highlighted part come from then?

It's just a spectrogram of surface noise - you can see how there is a disproportionate amount below 300Hz or so. A-weighting, therefore, is unfortunately required if we want to get a subjectively proportional measurement.

To my best knowledge, peak velocity (e.g. during a transient) is likely to be higher than 5cm/s...

I would concur with that. But I have also seen in a 'one-off' situation how a phono preamp with an SNR of only 64dB/ref 5mv/5cms can degrade the A-weighted measurement by 3dB on one transfer alone, which would tend to suggest at least 64dB. I will have a look through my hard-drives and see if I can the example. At some point in the future I will acquire some very high quality pressings and make some real measurements against a 5cm/s reference disc level, but I still stand by the 72dB figure for many high-quality pressings - especially Direct Metal Mastering.

So it looks like my homework is to get hold of a mint DMM, make a transfer and reference it against 5cm/s.

 

[undeakyetonde]

It's just a spectrogram of surface noise - you can see how there is a disproportionate amount below 300Hz or so. A-weighting, therefore, is unfortunately required if we want to get a subjectively proportional measurement.

No, I was to referring to something different

To me, statement on the website says "noise floor of best records is at least 72 dB, so noise floor of the preamp should be even lower". With all numbers referenced to output of 'typical' cartridge tracking a 5 cm/s rms groove, to make numbers comparable. Noise floor of my best records is much less than that, so I was wondering whether I'm doing something wrong

The quote from your website specifically says "72 dB relative to 5cm/s", whereas your previous post says "72 dB relative to -3 dB peak of a recording". I don't understand how you could get a referenced-to-5cm/s measurement from "just peak -3dB of one of the tracks", assuming the tracks are music. It is also not clear to me whether you compared "just peak -3dB of one of the tracks" to noise floor of a preamp or noise floor of a record.

Could you please could sort out this confusion?
Maybe you would be so kind to write everything out explicitly (peak/rms of signal X, compared to noise floor of Y, measured with Z) so there could be no chance of misreading?

---

You're probably aware of this, but the "Sound of Silence" book by Burkhard Vogel features calculations of DMM noise floor, but as far as I can tell these are based on writing down values from some published noise floor graphs, not by first-hand measurement.

2nd edition, page 215:
Table 11.4 Maximal SNs for various types of records
Vinyl 33 1/3 LP: -70.5 db(A)
Vinyl Single or Maxi 45: 73.5 dB(A)
Cu 33 1/3 LP: -72.5 dB(A)
Cu 45: 75.5 dB(A)

 

[Michael Fidler]

No, I was to referring to something different

To me, statement on the website says "noise floor of best records is at least 72 dB, so noise floor of the preamp should be even lower". With all numbers referenced to output of 'typical' cartridge tracking a 5 cm/s rms groove, to make numbers comparable. Noise floor of my best records is much less than that, so I was wondering whether I'm doing something wrong

The quote from your website specifically says "72 dB relative to 5cm/s", whereas your previous post says "72 dB relative to -3 dB peak of a recording". I don't understand how you could get a referenced-to-5cm/s measurement from "just peak -3dB of one of the tracks", assuming the tracks are music. It is also not clear to me whether you compared "just peak -3dB of one of the tracks" to noise floor of a preamp or noise floor of a record.

Could you please could sort out this confusion?
Maybe you would be so kind to write everything out explicitly (peak/rms of signal X, compared to noise floor of Y, measured with Z) so there could be no chance of misreading?

This would be more appropriate for a technical article than a product description, though. Although the preamp is referenced to 5cm/s, we are talking about the dynamic range of the record, which should still be around 72dB best case, or maybe even more. These do not have to be exact measurements, as we are measuring the level of a random signal after all. If you make the measurement from the inner groove, the peak level is usually in the region of 5cm/s.

Another way to reference to 5cm/s is if you know the electronic noise voltage exactly and can reference against that based on the cartridge nominal output level. I still believe 72dB A-weighted/220Hz+ is possible based on some inner-groove measurements I made a few years ago.

 

[undeakyetonde]

This would be more appropriate for a technical article than a product description, though.

Yes but it would be superb to have it written out at least in a post on ASR

I also believe that product description should be technically correct. More specifically, for comparing numbers in dB they should be referenced to the same thing. Currently I do not understand if your product description is correct in this sense, sorry. It clearly says "72 dB relative to 5cm/s" but here you wrote something else

 

[Michael Fidler]

Yes but it would be superb to have it written out at least in a post on ASR

I also believe that product description should be technically correct. More specifically, for comparing numbers in dB they should be referenced to the same thing. Currently I do not understand if your product description is correct in this sense, sorry. It clearly says "72 dB relative to 5cm/s" but here you wrote something else

It is a little somewhat reckless, I would agree, but I'm deliberately making it a bit extreme to make the point that the noise figure of the phonostage against the record will still be 1dB with a surface noise level this low. The broader context matters here.

Perhaps 'many' should be replaced with 'some can be as high as', but in my defence I'm picking a very hard number to be conservative with the following noise figure claim, as this is auxiliary information. If I reduced this estimate, I could liable to the eventuality that whatever system noise figure estimate made would be less accurate, so overestimating the surface noise makes the final point more conservative, if that makes any sense. Therefore, I would like to argue that making this figure more optimistic makes the description of the product itself more accurate/conservative, which is my intention when I write the descriptions.

I'm skeptical of the noise floor/dynamic range figures published in the 1970s and in other legacy papers, as we generally don't know what the noise figure of the phono preamp used to make the test was at the time.

 

[undeakyetonde]

Re: conservative estimate & overestimation - fair enough, but I’m still unsure if specific statement about 72 dB rel 5 cm/s SNR from the website is true or not. Sorry for mentioning this technicality again and again but it bothers me

Re: older measurements
I remember Mr Vogel mentioning (or at least speculating on) phono preamp used for the measurements and taking its noise into account. (Too lazy to re-read)

But yes, it would be great to see updated and thoroughly described measurements. Wish you success in obtaining a mint DMM

 

[Michael Fidler]

Re: conservative estimate & overestimation - fair enough, but I’m still unsure if specific statement about 72 dB rel 5 cm/s SNR from the website is true or not. Sorry for mentioning this technicality again and again but it bothers me

Re: older measurements
I remember Mr Vogel mentioning (or at least speculating on) phono preamp used for the measurements and taking its noise into account. (Too lazy to re-read)

But yes, it would be great to see updated and thoroughly described measurements. Wish you success in obtaining a mint DMM

It looks to be the case that Vogel is making his calculation close to 8cm/s, so I may only be off from him by a few dB with my result-conservative auxiliary figure.

If you look at the 73.5dB(A) figure, it's the raw surface of the record (which is against 8cm/s with a Shure V15 putting out 5.12mV RMS), so if we compensate for 5cm/s we get 69.4dB - not too far off! So perhaps I should revise down to 70dB and replace 'many' with 'some of the best' and still be safe with the noise figure estimate. Of course, a great many cartridges put out considerably less than 5mV at 5cm/s, so I may still be quite technically correct with the website as it is for a compensated SNR calculation if we reference the vinyl surface noise to 5mV with a 3.5mV cartridge.

It looks like a few mint DMMs are on eBay for not too high a price. So a good plan for me to get round to would be to take an MM PRO, DV-10X5 mk2 (high output 2.8mV) for a 5cm/s noise figure of over 80dB and make a few reference measurements.

 

[morillon]

8cm 1khz us tipicly europeen

It looks to be the case that Vogel is making his calculation close to 8cm/s, so I may only be off from him by a few dB with my result-conservative auxiliary figure.

If you look at the 73.5dB(A) figure, it's the raw surface of the record (which is against 8cm/s with a Shure V15 putting out 5.12mV RMS), so if we compensate for 5cm/s we get 69.4dB - not too far off! So perhaps I should revise down to 70dB and replace 'many' with 'some of the best' and still be safe with the noise figure estimate. Of course, a great many cartridges put out considerably less than 5mV at 5cm/s, so I may still be quite technically correct with the website as it is for a compensated SNR calculation if we reference the vinyl surface noise to 5mV with a 3.5mV cartridge.

It looks like a few mint DMMs are on eBay for not too high a price. So a good plan for me to get round to would be to take an MM PRO, DV-10X5 mk2 (high output 2.8mV) for a 5cm/s noise figure of over 80dB and make a few reference measurements.

8cm 1khz is typically a european usage, din standard...
;-)

 

[undeakyetonde]

It looks to be the case that Vogel is making his calculation close to 8cm/s, so I may only be off from him by a few dB with my result-conservative auxiliary figure.

If you look at the 73.5dB(A) figure, it's the raw surface of the record (which is against 8cm/s with a Shure V15 putting out 5.12mV RMS)

Yes, I looked it up, those SNR estimates are definitely referenced to 8 cm/s peak (Table 11.1). Bandwidth 20-20kHz.

DMM/vinyl noise floor information is from Neumann VMS-80/DMM cutting lathe service manual, which apparently contains a spectrum plot of a silent groove. It also says that this plot is provided by Teldec (inventor of the DMM process).

Mr Vogel takes great care to account for rumble, preamp noise floor (Neumann PUE 74), and even noise voltage density of the cartridge used to create that plot (Shure V15V).

So those should be good estimates

2nd edition, page 215:
Table 11.4 Maximal SNs for various types of records
Vinyl 33 1/3 LP: -70.5 db(A)
Vinyl Single or Maxi 45: 73.5 dB(A)
Cu 33 1/3 LP: -72.5 dB(A)
Cu 45: 75.5 dB(A)

For anyone else interested - Burkhard Vogel, "The Sound of Silence", 2nd edition, 2011, ISBN 978-3-642-19773-4. Chapter 11.

 

[Michael Fidler]

Yes, I looked it up, those SNR estimates are definitely referenced to 8 cm/s peak (Table 11.1). Bandwidth 20-20kHz.

DMM/vinyl noise floor information is from Neumann VMS-80/DMM cutting lathe service manual, which apparently contains a spectrum plot of a silent groove. It also says that this plot is provided by Teldec (inventor of the DMM process).

Mr Vogel takes great care to account for rumble, preamp noise floor (Neumann PUE 74), and even noise voltage density of the cartridge used to create that plot (Shure V15V).

So those should be good estimates


For anyone else interested - Burkhard Vogel, "The Sound of Silence", 2nd edition, 2011, ISBN 978-3-642-19773-4. Chapter 11.

OK - looks like I just need to change that 5cm/s to '5mV with many contemporary cartridges' and I'll be more accurate + 'many discs' to 'some of the best discs'.

I think I was still about right there making a reference point for the noise figure based on transfers of very nice discs with a conservative margin of a few dB thrown in for good measure, but it's still not quite right so needs re-phrasing and adjusting.

 

[undeakyetonde]

@Michael Fidler maybe you could be so kind to answer a couple more questions related to phono perhaps and their noise measurements?

There is this expensive hi-end phono preamp for which noise performance is specified as

Input noise (MC): 0.35nV/rtHz
Input noise (MM): 0.9pA/rtHz

1) Is it a good way to describe noise performance of a phono preamp?
2) Is it possible to compare your products to this unit?

Thanks

 

[Michael Fidler]

@Michael Fidler maybe you could be so kind to answer a couple more questions related to phono perhaps and their noise measurements?

There is this expensive hi-end phono preamp for which noise performance is specified as


1) Is it a good way to describe noise performance of a phono preamp?
2) Is it possible to compare your products to this unit?

Thanks

Although there is no SNR figure, voltage noise can be a good way to describe MC inputs, as the input impedance is so low.

Current noise is only part of the equation for MM. 0.9pa/sqrtHz is very high indeed for an MM input - an NJM2068 will give you a third of that.

It's my view that all SNR measurements should be made with a cartridge load connected. These numbers on their own are not very useful, and would tend to obfuscate far more than they describe.

 

[Balle Clorin]

t 72 dB rel 5 cm/s SNR from the website is true or not. Sorry for mentioning this technicality again and again but it bothers me

These Numbers does not make sense at all
At 1000hz the highest modulation in vinyl is 35cm/s or 7 times the reference level , that is 17db above the reference level. 72db above re reference level is simply not possible. To have a 72db span from +17 the groove noise have to be -55 db, That is maybe not an unreasonable number even if 40-54 db is what I see on mine records

 

[undeakyetonde]

These Numbers does not make sense at all
At 1000hz the highest modulation in vinyl is 35cm/s or 7 times the reference level , that is 17db above the reference level. 72db above re reference level is simply not possible. To have a 72db span from +17 the groove noise have to be -55 db, That is maybe not an unreasonable number even if 40-54 db is what I see on mine records

My best reading was “noise rms is 72 dB *below* reference of 5 cm/s rms 1 kHz sine”, which I found a bit optimistic and surprising. But I guess you could get a similar number with A-weighting or bandlimiting

I agree that 72 dB *above* that level is not possible