Phono Cartridge Response Measurement Script

[JP]

Not looking at its peak - more at the curve profile and where it rises to its peak - the needle will reach it's minimum tracking ability at the resonance frequency of the cantilever, where the resonance itself will interfere with consistent mechanical contact with the groove, resulting in rising distortion - and after the resonance the distortion level drops off....

the cantilever resonance of course has its own multiple harmonics, so if you have the ability to measure high enough in frequency, and the needle profile is fine enough to track at very high frequencies, you should see the harmonics of the resonance both in the F/R and in the distortion.

And yet there’s nothing remarkable in the distortion profile of the plot compared to others, even those with lower ETM.

 

[dlaloum]

And yet there’s nothing remarkable in the distortion profile of the plot compared to others, even those with lower ETM.

That's why I engineered my spreadsheet to compare "modelled" vs "measured" performance - it allows me to deduct the pure electrical "modelled" performance, so I can then obviously see the rest of the frequency response - it exposes the resonance really obviously.

I am guessing with the HD.... and given I never measured HD at the time, I don't have a feel for how it interacts based on measurement... just theoretical musings.

 

[Balle Clorin]

A million thanks to @USER for sorting out my script problem A complete removal and reinstall of Anaconda7Python/Spyder did the trick. ,Everything works well now.
I cannot recall in what thread to post our frequency plots, so I have to celebrate with this plot here. A cheap OEM cart from AudioTechnica AT-3600 . I used it for troubleshoot something....

By the way my file is 48k but the scrips say 96K, Do I have to record all wav files in 96k? or should I change the script yo 48k somewhere?

 

[JP]

By the way my file is 48k but the scrips say 96K, Do I have to record all wav files in 96k? or should I change the script yo 48k somewhere?

That version does SRC to 96k regardless of what the source file is so that I only had to deal with one set of filters for RIAA, etc. The released version will SRC to 96k or 192k depending on the input file.

 

[JP]

Painful painstaking work... but excellent!!

Not bad once the script is adjusted to do what I want. After that it's just a matter of recording and editing the files, which for that plot takes about fifteen minutes.

 

[stereoplay]

Buckle up.

I'd mentioned here that the script can be used on 'contrived' sweeps. The script slices the audio and runs FFTs on the slices, grabbing the largest bin for each slice. Slices at the same frequency are averaged. This gives us a lot of leeway on test signals.

The big question has always been, how accurate are the test records?

If we playback a spot frequency and slow down the 'table, the amplitude of the signal amplitude will decrease 6dB/octave. Likewise, if we speed it up the amplitude will increase 6dB/octave. Thus, if we playback a 20kHz spot frequency at half-speed, we'd end up at 10kHz at -6dB. I've modified one of my SP-10MK3 to do +20/-50%.

By taking 20, 10, 5, 3, and 2kHz spots we can record each one while sweeping the turntable speed from -0% to -50% and stitch together a "sweep" from 1-20kHz. The exact frequency and amplitude of each spot doesn't matter as long as they're both consistent throughout the track. If they are, a 10-20kHz "sweep" made from a single spot frequency track, for instance, will be the true response of the cartridge.

We could then use a variation of the script to create a spot track "calibration" to correct any amplitude errors in the track within the dynamic consistency of the cartridge. This calibration could then be applied to the spot "sweep" to give a corrected frequency response sweep for the cartridge. In turn, this data could then be used to validate sweep tracks on test records, and in the creation of transfer functions to correct them.



View attachment 264170

View attachment 264171

A really great idea!

I thought about that and made two sweeps with the same record. One played at 33RPM and the other one with 45RPM. Difference in level is about 2.6dB.
I recalculated the RIAA equalization (poles and zeros) for the 45RPM version (this is important) and took the difference of both curves:

Grey is the "new" frequency response of the cartridge
Yellow 20...20000Hz 33RPM
Pink 27...27000Hz 45RPM

PS: The ups and downs below 300Hz are due to several tube modes

 

[JP]

Came across a JAES paper today "Absolute Calibration of Pickups and Records", by some folks at CBS Labs, where they describe a method they call VSTM (Variable-Speed Turntable Method):

"SEVERAL years ago a test record was developed with a sweep-frequency tone synchronized to a curve tracer for continuous and automatic measurement of pickup character istics. Such a sweep-frequency record cannot be calibrated by conventional methods, i.e., microscope measurements or light patterns. By synchronizing the turntable speed with the frequency coordinate of the curve tracer, a method was devised for calibration of the sweep independently of the frequency response of the pickup. This method also permits the calibration of a pickup independently of test record characteristics."

Turns out I'm not so clever after all. Though today we don't need to synchronize the turntable speed.

 

[USER]

More examples of test record difference consistency.

 

[Balle Clorin]

What does the third plot show? Defference beweeen Signet snd AT cartridge?

 

[USER]

What does the third plot show? Defference beweeen Signet snd AT cartridge?

ATN150MLX stylus on Signet body. I am playing around with that body type (AT150, AT7V, AT120, etc.) and various styli to see how inductance affects the frequency response. Essentially Signet cartridges are premium AT cartridges with higher inductance. These Signets spec at 550mH while the AT150 line specs at around 350mH. As inductance mostly affects the high end, my bet is that with a lot of AT cartridges you don't need to match the bodies and styli, and in fact you may be better off with a unique and possibly cheaper combination. People do this all the time on other boards, which makes sense as AT is notorious for terminating and replacing cartridge lines, but, of course, they do so blindly so I sure as hell don't trust what they say works.

Edit: here is a NON-FINAL, WORKING example:

Edit 2: I want to redo the AT7V measurement as it is quite old, but I am hoping I can end up with something like this:

EDIT: ORIGINAL AT150 MEASUREMENT IS WRONG AND PEAKS CLOSER TO THE OTHERS

 

[levimax]

ATN150MLX stylus on Signet body. I am playing around with that body type (AT150, AT7V, AT120, etc.) and various styli to see how inductance affects the frequency response. Essentially Signet cartridges are premium AT cartridges with higher inductance. These Signets spec at 550mH while the AT150 line specs at around 350mH. As inductance mostly affects the high end, my bet is that with a lot of AT cartridges you don't need to match the bodies and styli, and in fact you may be better off with a unique and possibly cheaper combination. People do this all the time on other boards, which makes sense as AT is notorious for terminating and replacing cartridge lines, but, of course, they do so blindly so I sure as hell don't trust what they say works.

I thought lower inductance was considered "better"?

 

[USER]

I thought lower inductance was considered "better"?

I used to think the same. But the just-added second graph above is making me question general wisdom. This project is aimed right at it, after all. I'd need to wait and see what a corrected CBS-based graph looks like before I make any final decision, but I don't think I prefer the original, lower inductance body in this particular example. (It also shouldn't go unsaid that inductance is but one variable. This is simply a unique opportunity to look at it.) Those Signets are premium ATs and the TK7LCa is a nice cartridge with a beryllium cantilever and line contact diamond so the high inductance is deliberate. I can see impedance affecting different cartridges in different ways as some like AT have a particular house curve from what I can tell. This one seems to be made to play around with.


Edit: As I have mentioned earlier I have 2 different ATN150MLX styli and they give me different results on the same body, so I am posting this as work in progress. (I am pretty sure the difference is not wear. More, I get different results than JP. So there's more for me to sort out.)

EDIT: ORIGINAL AT150 MEASUREMENT IS WRONG AND PEAKS CLOSER TO THE OTHERS

 

[morillon]

measurements made with what overall capacitance seen by the cell?
can be considered the impact on the response curve after adaptation of this capacitance...
we will observe these 3 cases quite differently, for example, right?
i talk after 12k.... like if we use a low capacitance with signet etc(?)
perhaps why low inductance is pointed as preferable?

 

[USER]

measurements made with what overall capacitance seen by the cell?
can be considered the impact on the response curve after adaptation of this capacitance...
we will observe these 3 cases quite differently, for example, right?
i talk after 12k.... like if we use a low capacitance with signet etc(?)

120pF. All.

 

[morillon]

low inductance is associated ""historically"" with higher bandwidth, right?
example:

 

[dlaloum]

ATN150MLX stylus on Signet body. I am playing around with that body type (AT150, AT7V, AT120, etc.) and various styli to see how inductance affects the frequency response. Essentially Signet cartridges are premium AT cartridges with higher inductance. These Signets spec at 550mH while the AT150 line specs at around 350mH. As inductance mostly affects the high end, my bet is that with a lot of AT cartridges you don't need to match the bodies and styli, and in fact you may be better off with a unique and possibly cheaper combination. People do this all the time on other boards, which makes sense as AT is notorious for terminating and replacing cartridge lines, but, of course, they do so blindly so I sure as hell don't trust what they say works.

Edit: here is a NON-FINAL, WORKING example:

View attachment 267830
Edit 2: I want to redo the AT7V measurement as it is quite old, but I am hoping I can end up with something like this:
View attachment 267834

Totally with you on that - and playing with the bodies, in addition to the loading, provide further flexibility in achieving best results.

The ATN150MLX is an excellent stylus, but you definitely need to adjust C & R loading to control that peak and get closer to a flat F/R...

Having said that - these plots are within a range where some not overly draconian digital EQ could get it to where it needs to be!

My measurements of the ATN152lp and ATN440MLa show quite similar profiles....

 

[morillon]

I was indeed expecting that with a 120 pico it would not be so marked...
will have to observe with less...
but the scale of the measurement presented is also important...
(even if the has high frequencies rather in the sibilance ... joins the fact that the at are presented a-t cartrige "clear" see a little sharp often)
;-)
(it's a bit funny ... because for me it's rather the observation of different mounts of cantilever - diamond sizes on a given cartdrige body that seems to me the most interesting in what audiotechnica allows)

 

[JP]

I used to think the same. But the just-added second graph above is making me question general wisdom. This project is aimed right at it, after all. I'd need to wait and see what a corrected CBS-based graph looks like before I make any final decision, but I don't think I prefer the original, lower inductance body in this particular example. (It also shouldn't go unsaid that inductance is but one variable. This is simply a unique opportunity to look at it.) Those Signets are premium ATs and the TK7LCa is a nice cartridge with a beryllium cantilever and line contact diamond so the high inductance is deliberate. I can see impedance affecting different cartridges in different ways as some like AT have a particular house curve from what I can tell. This one seems to be made to play around with.


Edit: As I have mentioned earlier I have 2 different ATN150MLX styli and they give me different results on the same body, so I am posting this as work in progress. (I am pretty sure the difference is not wear. More, I get different results than JP. So there's more for me to sort out.)

View attachment 267840

I've not spent much time on it, but being a system the mechanical resonance needs to be taken in to consideration as well. If you look at a V15-V at low Cl there's a considerable in-band droop as the mechanical resonance is around 33kHz. The higher Cl for that cartridge uses the electrical resonance to compensate.

And then you have the true low-inductance designs, like the EPC-(P)100CMK4 at 33mH.

 

[dlaloum]

I've not spent much time on it, but being a system the mechanical resonance needs to be taken in to consideration as well. If you look at a V15-V at low Cl there's a considerable in-band droop as the mechanical resonance is around 33kHz. The higher Cl for that cartridge uses the electrical resonance to compensate.

And then you have the true low-inductance designs, like the EPC-(P)100CMK4 at 33mH.

With mechanical resonance at 70kHz + they could afford to go for a low inductance design - they didn't need the LCR circuit to do EQ to achieve a flat F/R - with the Shure and the resonance around 33kHz - it still needs a bit of LCR EQ to keep things under control....

 

[JP]

With mechanical resonance at 70kHz + they could afford to go for a low inductance design - they didn't need the LCR circuit to do EQ to achieve a flat F/R - with the Shure and the resonance around 33kHz - it still needs a bit of LCR EQ to keep things under control....

It doesn't really read that way. At 350mH and 350pF (where the V15-V is relatively flat) the electrical resonance is 14.4kHz. At low Cl it has a droop right around there that the electrical resonance compensates. "Under control" infers the opposite.