Good song to check Azimuth?

I guess that begs the question how far off perpendicular is acceptable? I only ask because your previous post and link sent me down a rabbit hole with a DMM and a test record…

 

I would think if it's that noticeable to the eye it's needs to be addressed.


If the cartridge is level and crosstalk's so off it can be heard, it needs to be addressed. There must be a standard, somewhere. 15db or so?

I make my adjustments as exact as possible. This can take longer than the initial installation. That's where a fixed head shell arm comes in handy that has adjustable azimuth at the pivot.

 

Be very careful using the equal crosstalk method. Many test LPs don't have accurate crosstalk signals cut into them and the results will have you thinking you need to tilt your cartridge in an extreme way or that the cartridge is defective.

See this chart:



You can see how much results can vary across records and some of these are not suitable at all for this particular test. It's one of the hardest tests to get right on a test LP.

 

I would think the standard would be the stated crosstalk spec from your cartridge manufacturer, in my case 27db. I think the biggest problem with setup this way is that the voltage from the unmodulated channel is extremely hard to measure. When attempting to average the result I feel like there’s an extreme amount of opportunity to misinterpret the data, but maybe I am doing something wrong?

I quickly noticed that the measurements that I was getting from my Ortofon record seemed to suggest that something was wrong with the way the tracks were cut. It has 2 set of tracks, so 4 total, and one of these measures different than the other 3.

 

The Ortofon LP is the best in-print one for testing crosstalk, but it's far from perfect.

Now, I wouldn't mess with the DMM but would record those crosstalk tracks digitally and work from there. You'll get more accurate results that way.

 

You need to use a filter for the DMM, are you doing that? If not, you will be measuring the rumble and subsonic noise components too. A 1KHz bandpass would be the best (assuming the test track is 1KHz), but probably can get away with just a high pass. If you have an amp with tone controls and subsonic filter, use those. You can also just connect the DMM (set to AC voltage) between the L(+) and R(+) outputs to directly measure the difference, which is the crosstalk, or just take a reading on each side, which I guess is how you are doing it now?

 

While not perfect, I use a test record in conjunction with the meters on my phono preamp for level matching.

 

I suspect I will be using the software method in addition to the DMM, this is the part of there hobby that gets me excited. Is something like Audacity and plugging directly into the mic input of my MacBook going to get me there, or should I not bother without a ADC?

Currently I am using the method linked from post #23 in whichFremer does suggest a filter here but doesn't call it a requirement; however I am quickly seeing that this method is problematic without it. It now occurs to me that I could quickly set up one of my idle Mini DSP units for this task, would you suggest a high and low pass a little before and after the 1k mark?

 

Yes, or if there is option to implement a 1KHz bandpass filter use that.

 

Audacity or a similar program works fine. I don't know the quality of the soundcard on your Macbook. You can give it a try, but you'll need to experiment with setting levels in the software so you don't cause things to distort. A USB audio interface like a Focusrite or Motu and so on would be preferable.

 

You can easily implement a bandpass filter in Audacity via the Effect->Filter Curve menu, as below ... first image of Aldous Harding "Blend" track from her very good Party album, second image shows filter curve setup with about 60dB reduction at the 20-20K Hz extremes (could be a lot more or less but you want it fairly steep, this is only about 12dB per octave, a little less on the low side, a little more on the high side), centered around 1KHz, and third image shows the result of that track with filter applied ...

But I'd think it would be much quicker to do in realtime with a hardware filter and DMM unless you are all setup to record, and very proficient at it, which I'm not (though passive bandpass filters are difficult to implement with high attenuation slopes) ...

 

Since this thread is painful to read knowing what I'm sitting on, here you get an unfinished version of my azimuth-correction script for Windows to try out:

devinyl-steps.cmd

It uses the SoX audio processor. If you don't have it installed in the Windows path, you'll be given the choice to download the exe to the current directory.

Take a recording of your left channel/right channel separation test, edit to a 10 second file with half right-channel and half the left-channel separation track. Save as "input.wav" (the default file if you simply double-click on the script to run it).

Run script in the same directory. It will step through -5 to 5 degrees (clockwise looking from front) of its software azimuth correction technique, and save the output as a merged file, and a second wav in mid-side mode, which you can manually analyze.

View the devinyl-midsidesteps wav in your audio software. Switch to "panning view", Audition 3 (others may not have such features). You will see the channels converge at a center point of correct azimuth.



Additionally, when you see the point where the channels are balanced in mid-side panning, the amount above or below the center line is the amount of crosstalk. If below "0", there is actually out-of-phase crosstalk or the cartridge motor is not 90 degrees.

Then view devinyl-stereosteps.wav in "phase view", confirm that the phase has equal spacing from center at this amount of correction.



I'm looking to automate this analysis with real software instead of spaghetti scripts, but it seems I must still demand someone use a separation test for best results (although the analysis of random music still looks promising, just inconsistent).

When one knows the amount of electrical misalignment, you can re-examine how straight your azimuth setup actually is -- or if the diamond and cantilever all look perfect ... fix in software.

The "still plugging away at it" script otherwise performs a full workflow correction of vinyl needledrop, including azimuth correction, when the setting for "forceazimuthanalysis=yes" is disabled. Docs and features to come some other time.

 

That's the trick. I set up a Butterworth crossover at 48db to high pass 800 and low pass 1200, this seemed to be about the sweet spot to not loose any signal strength. The readings I got this time were absolutely stable, no variation to speak of, and when I went to make my adjustments I could immediately see the results on my meter. It did leave my cartridge slightly off of perpendicular to the record surface by about .0325 which seems a little more than I would have expected, however my measured crosstalk is about 31db like this and the voltage output between the two channels is within .002.

 

Hmm... somehow, it appears as if we’re overcomplicating azimuth adjustment

 

Nothing is overly complicated in the pursuit of better sound…

 

Wow. I marvel at all you guys go through.
I eyed balled my Soundsmith cartridge.
I threw on a test record followed by Sgt. Pepper.
Sounded real good. I picked up a VTA Azimuth cartridge alignment tool.
Made a minor adjustment. And now the sound/separation is even better.
I’m good. Carry on. Cheers.

 

I found a still sealed copy of that John Sebastian LP on Discogs for $12 including shipping. Those steel drums at the beginning of the track are hard to track. I had my azimuth set visually using its reflection, so with a tiny bit of barely breathing on my azimuth-on-the-fly dial, I was able to get his voice centered, way behind the steel drums as described in the early posts here. That was the ez part.

The a/s setting was more challenging. I kept getting scratchy noises out of both channels, didn’t really tell me which way to adjust. I tried a bit more a/s, a bit less a/s, no real improvement. I pretty much gave up for a while.

So today I’m playing Heartless Bastards/The Mountain, side B/track 2. The singer has a very resonant voice. If your speakers have any loose screws on the crossover boards or on the drivers, this will reveal that! Anyway, the voice was giving me some scratchy sounds out of the left channel so I increased a/s. Scratchiness from the left went away, but now I was hearing the voice leaking through from the previous groove. That told me I added too much a/s. So I backed off on the a/s to about where the scratchy sounds in the voice from the left channel were still mitigated, with no previous groove vocal bleed through on the right channel.

I put John Sebastian back on. I was still getting a tiny bit of scratchiness from both channels from the steel drums right at the beginning, but it was much much less than before, almost unnoticeable. Probably a limitation of my stylus.

All in all, it was very educational to learn what to listen for when getting everything dialed in just right. Now I have to make a promise to myself not to touch anything anymore