Yes, so you need to move the red line up 2.6dB on the 45 RPM plot so at 20k the difference is closer to -6dB vs -5.6dB not -6dB vs -3dB.
Last edited:
Another way to explain, RIAA is flat at 1kHz so the red and green lines should always intersect at 1kHz reflecting only the RPM change.
It doesn't happen at all - it's an artificial construct aimed at helping understand what constrains limits of playback level at various frequencies.
Fortunately, no-one explained this to magnetic playback cartridges, which simply transcribe instantaneous stylus velocity into instantaneous cartridge emf or voltage output level. Nothing to do with frequency, at least in principle.
So if one considers playback of a sine wave f sweep with constant programme level and no RIAA emphasis, it would have constant stylus velocity across the band. If one changes rpm, only cart output level would change.
If one considers the same sweep made with RIAA 'on', f sweep velocity as recorded on the record is f dependent. So changes to rpm imply a mapping of 'old' RIAA level onto 'new' RIAA level, for playback to be flat, as well as scaling overall level.
LD
Last edited:
Thank you for this information, this is exactly what I did.
I adjusted Level in the same ratio as the change in rpm, resp a and b in the image below.
And Mapping the old Riaa into the new Riaa was done accordingly with the same a and b.
a= (45)/(33 1/3) and b = (60.75)/(33 1/3)
So I think I did everything the way it should have been done.
Hans
Attachments
I'm kind of burnt out on trying to figure out all the error sources in using these various test LP's. If you know that your cartridge is a perfect velocity transducer at these levels it's one thing, mine is not so I needed the extra correction.
You are applying Riaa in the digital domain, aren’t you
In that case it wouldn’t seem to be any problem for you to digitally remap the Riaa time constants when changing speed. After that, all that remains according to LD is a fixed level shift over the complete spectrum.
Or are you referring to something else with “extra correction”
In that case, what could be the reason for and the nature of this extra correction ?
Hans
But in reality cartridges are not perfect, despite whatever claims arise. This is obvious because in practice cartridges sound profoundly different, no matter what similarities they might possess under limited test conditions.
It just means one is measurng the wrong aspect of performance. I still expect to find a level sensitive answer, when eventually the secret is found.
LD
The cartridge is not a true velocity transducer. Your correction converges at high frequency to 0dB, this relies on the constant velocity behavior matching RIAA exactly, one up with f and one down with f at exactly the same rate.
This is why I think 500Hz to 50kHz constant velocity sweeps without RIAA at several different levels would be useful on a test LP. The STR series had one LP with only the 505Hz time constant a reasonable compromise that allows sweeps that span the whole range. Taking the RIAA conformance of the lathe and your pre-amp out of the equation is a good thing.
Last edited:
O.k. , I’m with you because that was exactly the reason why I questioned in #582 whether it could be a square law instead of a linear relation between speed and level, because I was missing 1.3 dB when speeding up and also when going down in both cases by a factor 1.35
But the answer in #584 was a definite No.
So it seems we are back at this point after all.
Question is, what can we do to come closer to finding answers with the materials we have, because flat recorded LP’s with different levels are not available.
I can have a look again at my recordings and measure the exact level change versus the calculated level change.
Would that bring anything ?
Hans
I have the CH precision disk so I can start getting confused about it too soon 🙂 (thank you Hans).
I am still tempted by the Ortofon test disk, even though it only has the sweeps at a single level and at the outside edge of the disk. Confusing that it is a constant velocity sweep and yet the liner notes say that all tracks should be played back with RIAA correction...
We may assume that mapping Riaa is 100% exact and so is the fact that PN has a 3dB/oct roll off, so the only variable is the level change when changing speed.
This isolates the search for level dependency to one parameter, right ?
Hans
This isolates the search for level dependency to one parameter, right ?
Hans
Constant velocity is 6dB per octave if your cartridge is 5.5dB per octave PN will be 3.5 dB per octave even with perfect RIAA.
That’s true, but on the record PN will go down with 3dB/oct, so when looking at levels, this will become obvious, unless the record is already compensated for such deviations, which seems a bit unlikely to me.
But tomorrow I will report my findings which hopefully will make us a bit wiser.
Hans
This is the first set of results, in this case taken from the CH Precision measurements.
What I have done:
1) Measured the 45rpm levels at 1/3 octave intervals starting from 2Khz up to 25.4kHz.
2) Calculated the new frequencies for 33 1/3 and 60.75rpm from the used frequencies at 1) by multiplying /dividing them with the factor 1.35.
3) Added to the measured level values in 1) the calculated change in level when going resp. a factor 1.35 to 33 1/3 and 60.75rpm.
In this change the exact Riaa has been mapped, plus the assumed change in level as a fixed number, in this case resp -2.6dB and +2,6dB.
The result is what can be called the "expected level".
4) Measured the levels at resp 33 1/3rpm and 60,75rpm, and calculated the difference with regard to the expected level.
Example: At 45rpm the level at 4kHz is -59.2dB.
4kHz becomes resp 2936Hz and 5400Hz when going to 33 1/3rpm and 60.75rpm.
Expected level change because of A) Riaa and B) 2.6db in speed caused level change will be resp -0.7dB for 33 1/3 and 0.4dB for 60.75rpm.
Measured level at 2936Hz is -58,5dB and -60.3dB at 5400Hz.
Level difference from expected at 33 1/3 Hz is: -59.2dB-0.7dB+58.5dB =-1.4dB
Level difference from expected at 60.75rpm is: -59.2dB+0.4dB+60.3dB =1.5dB
In the image below, these differences between measured and expected are visualised.
The only variable in this is the level change. The measured values and the Riaa mapping are as exact as they can be.
It becomes obvious that the expected level change of 2.6dB when changing speed 1.35 times, is not correct.
The two images with an accuracy of within +/-0.5dB, are almost complementary to each other, making the probability that they are correct quite high IMO.
For the most part of the frequency range, deviation is ca 1.5dB.
Is this part the constant amplitude part of the spectrum ?
And from above a certain frequency, where deviation drops rapidly to zero
the constant velocity part ?
Hans
P.S. the results from the Adjust+ record still to come.
What I have done:
1) Measured the 45rpm levels at 1/3 octave intervals starting from 2Khz up to 25.4kHz.
2) Calculated the new frequencies for 33 1/3 and 60.75rpm from the used frequencies at 1) by multiplying /dividing them with the factor 1.35.
3) Added to the measured level values in 1) the calculated change in level when going resp. a factor 1.35 to 33 1/3 and 60.75rpm.
In this change the exact Riaa has been mapped, plus the assumed change in level as a fixed number, in this case resp -2.6dB and +2,6dB.
The result is what can be called the "expected level".
4) Measured the levels at resp 33 1/3rpm and 60,75rpm, and calculated the difference with regard to the expected level.
Example: At 45rpm the level at 4kHz is -59.2dB.
4kHz becomes resp 2936Hz and 5400Hz when going to 33 1/3rpm and 60.75rpm.
Expected level change because of A) Riaa and B) 2.6db in speed caused level change will be resp -0.7dB for 33 1/3 and 0.4dB for 60.75rpm.
Measured level at 2936Hz is -58,5dB and -60.3dB at 5400Hz.
Level difference from expected at 33 1/3 Hz is: -59.2dB-0.7dB+58.5dB =-1.4dB
Level difference from expected at 60.75rpm is: -59.2dB+0.4dB+60.3dB =1.5dB
In the image below, these differences between measured and expected are visualised.
The only variable in this is the level change. The measured values and the Riaa mapping are as exact as they can be.
It becomes obvious that the expected level change of 2.6dB when changing speed 1.35 times, is not correct.
The two images with an accuracy of within +/-0.5dB, are almost complementary to each other, making the probability that they are correct quite high IMO.
For the most part of the frequency range, deviation is ca 1.5dB.
Is this part the constant amplitude part of the spectrum ?
And from above a certain frequency, where deviation drops rapidly to zero
the constant velocity part ?
Hans
P.S. the results from the Adjust+ record still to come.
Attachments
Hans, this would be a lot easier with spot frequencies.
The theoretical basis here is that the pink noise is really pink, the RIAA pre-emphasis is exact, the pre-amp's RIAA is exact, and the cartridge is perfect. There are too many degrees of freedom to isolate where the problem resides.
Here are the results for the Adjust+ record.
This time going resp from 33 1/3 to 45 and 60.75rpm.
Procedure is the same as for the CH Precision record, but now with 1/2 octave steps.
The same thing happens here, for most part of the frequency range, level doesn't go up by 2.6dB and 5.2dB for both speed steps, but only half that amount with ca 1.3dB and 2.6dB.
Only at the high end of the spectrum. the deviation in expected versus measured level goes to zero.
Hans
This time going resp from 33 1/3 to 45 and 60.75rpm.
Procedure is the same as for the CH Precision record, but now with 1/2 octave steps.
The same thing happens here, for most part of the frequency range, level doesn't go up by 2.6dB and 5.2dB for both speed steps, but only half that amount with ca 1.3dB and 2.6dB.
Only at the high end of the spectrum. the deviation in expected versus measured level goes to zero.
Hans
Attachments
I did use spot frequencies, 1/3 octave apart, didn't I ?
To my opinion Pink Noise plays no part at all in this case.
I just pick a frequency, measure the level, map that frequency to a different rpm, map the Riaa and adjust level with the factor of the speed change and finally compare this calculated level with the measured level.
With white noise the procedure would have been exactly the same.
I would be glad to hear if and where my assumptions are wrong.
Therefore. the only variable that remains to my best knowledge is the level, where it becomes clear that the level does not change as assumed as a fixed value to the same amount as the factor in speed change, but over a large part of the frequency range it's closer to the Sqrt of the speed change factor.
Hans
- Status
- Not open for further replies.