Phono Cartridge Response Measurement Script

[Thomas_A]

It’s chasing two unknowns.

I don't see it...yet. The different cartridges with very different response give the same difference signal >300 Hz. It should not if cartridge error is a factor.

 

[Thomas_A]

I think I did this in post #532, but without shifting back. See the blue response.
I took 1KHz and compared it with 1,35KHz and so on.

The dip at around 230Hz is due to torsion of the tonearm.

Can you test another test record and analyse as in post 524?

 

[stereoplay]

I will do that this evening.

 

[JP]

I don't see it...yet. The different cartridges with very different response give the same difference signal >300 Hz. It should not if cartridge error is a factor.

Very little error for both the record and cart below 1kHz coupled with the FFT resolution.

Cart error has to be a factor. If the cart has an error that you know exists at 45 from the 33 level then you have to correct is at that F for 33 or you'll compound errors. I think this is why the error curve always looks very similar, or a contributor to that.

 

[Thomas_A]

Very little error for both the record and cart below 1kHz coupled with the FFT resolution.

Cart error has to be a factor. If the cart has an error that you know exists at 45 from the 33 level then you have to correct is at that F for 33 or you'll compound errors. I think this is why the error curve always looks very similar, or a contributor to that.

I'll have to think more. I was thinking if you have the perfect test record, all deviations should arise from the cartridge both at 33 and 45 RPM. Question: are the curves identical, i.e. do the peaks and dips occur at the same frequencies? If so, the difference signal should be constant vs fr (?)

 

[stereoplay]

Here are two plots with different test records. Again, I think the blue plot has to be inverted and represents more or less the frequency response.

 

[Thomas_A]

Here are two plots with different test records. Again, I think the blue plot has to be inverted and represents more or less the frequency response.
View attachment 280656
View attachment 280655

Thanks,

These are subtracted and aligned for 2.6 dB but not shifted 1.35x after subtraction?

 

[stereoplay]

Nothing is shifted. It is red-yellow as shown in the graph.

Another idea: If I change the capacitive load so the HF peak shifts, then it has nothing to do with the record, it's only the cartridge (and preamp of course). If I'm right with my assumption (blue is the test records' FR), then the graph wouldn't change.


What do you think about that?

 

[Thomas_A]

Nothing is shifted. It is red-yellow as shown in the graph.

I think it is needed but I might have it backwards A potential peak at 10k corresponds to the record level at position ”33 RPM 7407 Hz” (ie 10k at 45 RPM) minus level at position 10 kHz (reference @33 RPM). I think. So it has to be shifted back, 0.74x.

 

[stereoplay]

Which trace you think should be shifted back? I could do it in the software.

 

[Thomas_A]

Which trace you think should be shifted back? I could do it in the software.

The difference curve

 

[Thomas_A]

The problem with the method is that it is pairwise comparisons on the record. If there is a +2 dB peak at 740,7 Hz and +2 dB 1000 Hz on the record, the relative difference will be zero. So you will get errors. I guess the average difference curve of 33 vs 45, 33 vs 16, 33 vs 78 would take some of these errors out. Spin down methods as JP did is other options.

 

[stereoplay]

I got it now.

If we assume that the frequency response is raised by +1dB above 1KHz til the end of the sweep, we will see this for a short time in the crossover region, but not after that. The difference would then be 0dB above 1KHz. It will not work.

 

[Thomas_A]

I got it now.

If we assume that the frequency response is raised by +1dB above 1KHz til the end of the sweep, we will see this for a short time in the crossover region, but not after that. The difference would then be 0dB above 1KHz. It will not work.

Yep. The wiggles are the transient changes, pairwise comparisons. So it shows where there are changes but level returns to zero as soon they are the same. So it is the ”derivative” and tell you something about what is happening but not the record fr response.

 

[Thomas_A]

I guess you can recalculate it to the response, but simply doing subtraction just shows the derivative. Looking at the two graphs it is evident that the "change" in the DIN record means that it is decreasing in the top octave relative to the other test record. The plus change in the diff curve means just that you took the lower fr record point minus the higher fr record point (45-33 RPM), e.g. that the higher fr point is decreasing relative to the lower. Swap the subtraction.

 

[JP]

The problem with the method is that it is pairwise comparisons on the record. If there is a +2 dB peak at 740,7 Hz and +2 dB 1000 Hz on the record, the relative difference will be zero. So you will get errors. I guess the average difference curve of 33 vs 45, 33 vs 16, 33 vs 78 would take some of these errors out. Spin down methods as JP did is other options.

If you use one domain to update the other it *may* *kinda* work as in my example. You'd have to choose a starting point and assume it's accurate and error accumulation would likely be a meaningful issue.

This is something best explored logically rather than empirically. I'd consider creating a test signal dataset and a cart FR dataset so those are both known and then apply the method.

 

[Balle Clorin]

I use a record flattener like record pi or vinyl flat.

I wonder now if some of the oscillations may have to do with insufficient tracking force.

Regarding the wiggles I get om my ATPTG33 with ClearAudio TRS-1007, but not with ATOC9ML/II. I definitely occurs every rotation, I calculated the time vs frequency /roitation interval. The valley occurs at frequency intervals that match the rotation, and I can visually se the cartridge wiggle. Soit may be some kknd of "scrubbing effet .the cart goes up/down and the relative frequency/velocity changes. So maybe PTG has a disadvantage related to bumps that the OC9 does not have? ,puzzled. I will try to chage tracking force and VTA and see. Maybe the SME fluid damping will help?


But why is it so pronounced in right channel and not left so much?

My AT33PTG/2 wiggles in the right channel, but a cheap MM Thorens TPU257 or ATOC9ML/ii does not,,

ATOC9ML/ii no wiiggles

 

[stereoplay]

I have now modified a turntable in such a way that I can linearly change the speed between 0.5 and 1.4 times with a single start impulse.
Finely graded spot frequencies are recorded on the Elipson record, which I used to carry out the measurement. At the beginning of such a frequency, I changed the turntable speed. Here we see 23 steps to represent a frequency range of 20...20000Hz (yellow lines).
With an automated level adjustment I then get the true frequency response (white line).
Tonearm resonances etc. are of course included in the result.

It was a lot of work...

 

[JP]

Why bother with so much overlap?

 

[stereoplay]

Better have a greater overlap to align the levels.
Beside this it was no problem to change the speed over such a wide range.