Phono Cartridge Response Measurement Script

[JP]

Hmm, what do you mean by "both files"? In the "advanced" section I see an option for a second file but it's not clear what will be done with it. I tried using trimmed L and R there but it gave a garbage output. I don't see any option to turn off auto extraction.

Familiarize yourself with the script options, the majority of which are surfaced on the Advanced tab: https://sjplot.com. And yes, the "Sweep Extraction" control is on the Advanced tab, in the "Sweep Extraction" section.

Your specific issue was caused by the HF noise in your recording. In the ~35ms of "silence" between the end of the pilot tone and the start of the sweep the script picked up the ~37kHz noise in the ordersignal function, which is higher than the ~20kHz sweep end, so the function assumed the sweep was running backwards. The correction was misapplied due to a transposition. That's been fixed.

You didn't see the HF garbage in the plot because the Standard configuration automatically applies end_f depending on the record, so no data beyond 20kHz was plotted. If you use Advanced this parameter is not automatically applied. I shifted the start sweep index by 35ms and shifted the end detect filter to not be centered which gives a tighter trim on the extraction.

I made these changes as there are legitimate use cases where better noise immunity is beneficial, however the downside is that it masks setups that aren't fit for purpose. You really need to spend time on your setup and validating your chain.

 

[JP]

Version 18.7.0 released:

• Standardize audio data arrays.
• Standardized arrays mitigate ordersignal transposition.
• Constrain crosstalk fft data to the bounds of the fundamental frequency (f0).
• Re-order DSP functions (RIAA, XG-7001 correction) to occur prior to extracting sweeps. This means save_sweeps output will be post DSP.
• For extract_sweeps, shift sweep start by 35ms, and shift filter from centered to edge for sweep_end_detection. This trims a tiny bit of audio but is benign.

 

[ray_parkhurst]

The new version works fine with the 312HEP. I'll try it on my other cartridges that had this problem, including the 310MC.

I assume the 37kHz noise is due to the MC resonance but I'll look further into it to see if there is some other culprit. I believe this only happened with MCs, but I did see it both with and without the MC Pre-preamp in the SL10. The HOMC plus Pre-pre together certainly don't help.

 

[ray_parkhurst]

I made these changes as there are legitimate use cases where better noise immunity is beneficial, however the downside is that it masks setups that aren't fit for purpose. You really need to spend time on your setup and validating your chain.

I'm having trouble finding the "pathological" 37kHz noise in the recording. The Audacity spectrum analysis looks fairly clean and shows nothing at 37kHz. I'm not sure how to solve a problem I can't find.

 

[JP]

Audacity's spectrum analysis is easily one of the worst implementations I've ever seen, so I'd be looking for another tool, though it can give good hints. Also oddly good for LF stuff like quickly determining resonance.

Anyway, it would appear that you neglected to apply inverse RIAA.

 

[Balle Clorin]

what is wrong with my excel here, i try to import script cvs file ,-it look normal- but stupid Excel put a blank lines in for every row an makes the import useless

AI helped
️ Option 1 — Use Windows Notepad (the built‑in one)


Notepad can convert line endings, but only indirectly. The trick is to force it to rewrite the file.


Steps

1. Open the CSV in Notepad
2. Press Ctrl+A to select all
3. Press Ctrl+C to copy
4. Open a new Notepad window
5. Press Ctrl+V to paste
6. Save the new file as clean.csv

Why this works:
Notepad rewrites the file using Windows CRLF endings when you paste text into a new document.


This fixes the issue in many cases.





️ Option 2 — Use PowerShell (very reliable)


If you want a guaranteed clean file:




(Get-Content input.csv) | Set-Content -Encoding ASCII output.csv





This forces Windows‑style CRLF endings.


You can also use UTF‑8 if needed:




(Get-Content input.csv) | Set-Content -Encoding UTF8 output.csv








️ Option 3 — Use Excel’s “From Text/CSV” importer


This avoids Excel’s automatic parser and normalizes line endings.

1. Open Excel
2. Go to Data → From Text/CSV
3. Select your CSV
4. Click Load

Excel will clean up the line endings during import.





Quick check


Open the CSV in Notepad and turn on word wrap (Format → Word Wrap).
If each line looks doubled or spacing looks odd, you definitely have mixed line endings.





If you want, paste 5–10 lines of your CSV here and I can tell you exactly what’s causing the blank rows.

 

[JP]

None of that tells me what's wrong with your line endings. Without that, or better, the actual file that's causing you issues, I can't help you.

I'll be back in a couple weeks.

 

[Balle Clorin]

None of that tells me what's wrong with your line endings. Without that, or better, the actual file that's causing you issues, I can't help you.

I'll be back in a couple weeks.

.Not a problem any more, I got a workaround, If I copy the file content, cut it CTL X, and paste back it works. Something to do with line ends that becomes corrected that way

 

[JP]

That’s great but it’d still be good to fix the problem.

 

[Balle Clorin]

None of that tells me what's wrong with your line endings. Without that, or better, the actual file that's causing you issues, I can't help you.

I'll be back in a couple weeks.

Is mentioned, I got a workaround. I just have to open the csv file in Notepad and save it once. explanation here

 

[JP]