Phono Cartridge Response Measurement Script

[Thomas_A]

@Thomas does this filter give the right answer?

(biquad-m *track* 0.73137993061956 -0.39063797071820 -0.27569578143734 1.0 -0.59106473412185 -0.39977762693709)

No:

Twice this one (biquad-m *track* 9.08612614639649000E-01 -5.22931473883012000E-01 -3.44913691685509000E-01 1.00000000000000000E+00 -6.04500910000000000E-01 -3.90945930000000000E-01) does:

 

[JP]

96kHz?

 

[Thomas_A]

Yes. I've never recorded any higher.

 

[JP]

Okay, so the question is going to be: what was the actual source of the error? Applying your filter twice is effectively squaring the filter, which would correct for if the record was cut with the filter applied twice. My filter is to correct for the record being cut at the wrong speed which shifts the poles.

 

[Thomas_A]

I can't say anything regarding the math and error but it surely aligns with the CA-TRS1007, when applying the filter squared. What Flo said makes your filter more logic though. My AI help said this, but have not checked it.

input → biquad1 → biquad2 → output

y[n]=b0x[n]+b1x[n−1]+b2x[n−2]−a1y[n−1]−a2y[n−2]

96 kHz sample rate​

Biquad 1​

b0 = 1.92104426
b1 = -1.79211512
b2 = 0
a1 = -0.87012987
a2 = 0

Biquad 2​

b0 = 1
b1 = -1.98048430
b2 = 0.98053743
a1 = -1.96613454
a2 = 0.96618729

---

 

[JP]

I went back and read what Flo said. My guess is that when he disabled RIAA it also disabled the half-speed compensation, so it was cut with standard bass de-emphasis at half speed, so the correction would be a first-order shelf with doubled turnover frequencies. @vinylflo

Perhaps you can e-mail him again?

My hunch is that's the scenario, so then you've a correction, and a correction that's not quite correct but matches expectations better.

 

[Thomas_A]

I went back and read what Flo said. My guess is that when he disabled RIAA it also disabled the half-speed compensation, so it was cut with standard bass de-emphasis at half speed, so the correction would be a first-order shelf with doubled turnover frequencies. @vinylflo

Perhaps you can e-mail him again?

My hunch is that's the scenario, so then you've a correction, and a correction that's not quite correct but matches expectations better.

I just tried the above two bi-quads, but it seems to be the reverse adjustment.

I have e-mailed him a couple of times, but no answer.

 

[JP]

It's interesting that TRS-1005 has that same ~9kHz/1dB dip unless I don't apply turnover.

 

[Thomas_A]

Can't stay up longer now. Work day tomorrow.

But I think the two bi-quads shows what the error would be on the Ortofon, so it should be applied in reverse. I used a CA-TRS1007 file (blue) and the applied the bi-quads on it (red). Looks quite similar to the Ortofon drop, but not quite. So maths should be checked...

 

[Thomas_A]

It's interesting that TRS-1005 has that same ~9kHz/1dB dip unless I don't apply turnover.

Yep. I'm quite sure there is some errors made related to cutting and RIAA settings in several records. Or something that is not told on the records, since they may have used special corrections during playback.

 

[JP]

I *think* with TRS-1005 the lathe was modified and it was cut without turnover. They make no mention of any recording or reproduction EQ, and not applying turnover gives expected results. It's weak, but it's what we have.

Still working through what combinations of error and EQ that make sense.

 

[JP]

I did a layback with a DL-1000A because it's the flattest cartridge I have. The telltale here is the response from 800 Hz to 1 kHz, with the double-pole filter being flat and the squared filter not. The tails of the LF corrections are fully settled by 3 kHz.

The dip to -1 dB around 12 kHz smells, but I don't think it's caused by the LF filter error during cutting. If I put filtering in the script for this record, it would be the doubled filter and not the squared.

 

[Thomas_A]

Hm, what is your double filter bi-quads? I might try it, or was it the one you showed?

 

[JP]

The one from yesterday.

 

[Thomas_A]

The one from yesterday.

Ok. I will try that again and also with the bi-quads that AI suggested but reverse. Later tonight.

 

[JP]

This is the filter inverted to show the recording characteristic vs. my layback with the DL-1000A. It tracks within 0.03dB to 2kHz.

 

[Thomas_A]

Checking a normalisation at 5 kHz with your filter, Ortofon original (blue) vs compensated (red). Feels that it should be in-between...

 

[JP]

Are you recording flat, such that the filter I provided is the only EQ?

 

[Thomas_A]

Are you recording flat, such that the filter I provided is the only EQ?

No, I can not disable RIAA from my preamp (and keep loading) so I apply the RIAA inverse correction in script. (I measured my RIAA to within ±0.1 dB between 40-20000 Hz.)

 

[JP]

Perhaps try by reverting RIAA via Audacity biquad first, then apply my double filter, and use no RIAA in the script. Your plot looks very similar to the effect of applying LF correction to a flat file.