Electrical Azimuth adjustment- aka crosstalk

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

I have been mulling over the role of cross-talk in the vinyl experience for a few years now, triggered by this thread http://www.diyaudio.com/forums/analogue ... -self.html . After a bit of meandering Douglas didn't reach his goal of a vinyliser but did do an AES paper on a devinyliser which is very similar to a mastering elliptical filter to remove Side information below 100Hz. What was interesting in this implementation was the deep notch you could get by tweaking the all pass delay. AES paper is here http://www.douglas-self.com/ampins/Pari ... yliser.ppt
devinyliser.JPG
Anyway parked that as one for later then recently took a serious look at the MTC-MS-II board and realised that it would be a useful part of any vinyl front end and not just for mastering. Plus mops up some odd bits and pieces I need.

Then recently saw a review for the Jolida Foz XT-R http://www.jolida.com/product/foz-xt-r . This claims to be able to adjust for azimuth misaligment electrically and improve cross talk to 'up to 40dB'. To do this there are two knobs one for left and one for right and an uncalibrated meter that you adjust for minimum. What it appears to do is add some in phase cross-talk to cancel the out of phase. I'm not sure how much need there is to have individual channel adjustments, even though you generally get different cross-talk measurements left and right out of a cartridge.

Electrical cancelling of azimuth errors is at least in theory good because what you really want to optimise in the mechanical alignment is stylus azimuth, but I think Jim Fosgate missed a couple of tricks in getting this box down to $300 retail (also possible he is much cleverer than me). Firstly cross talk is frequency dependent. Secondly it might be level dependent.

Then the penny dropped. A stereo width control as per the MTC-MS-II Is exactly the same but it works on both channels equally. With the tilt control I think it handles the frequency selective part of the adjustment. Level dependence we still can't fix, but that's small beer.

Of course there is one school that this cross talk is part of what makes the 'vinyl sound' and so fixing it is taking away the magic, but I reckon it's got to be worth a play :). Unless of course I have completely missed the bleeding obvious, which often happens.

Thoughts?
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Electrical Azimuth adjustment- aka crosstalk

Post by mediatechnology »

Looks like we have two different things to discuss here: Douglas' Vinylizer and the Jolida Azmith Corrector.

I think I participated in Douglas' thread. What Douglas failed to explore was the effect of his HP filter slope on the LP crosstalk curve. The steeper HP slopes he shows will have severe peaking in the crosstalk curve. It pays to look at what happens on the other side of the "hill."

If you want to HP filter Side and maintain separation without peaking its better to use an Elliptic EQ on playback and have the Allpass correct the HP response. That's what the EEQ-12 does: viewtopic.php?f=7&t=911 http://ka-electronics.com/shop/index.ph ... search=EEQ

My meandering thread is here: viewtopic.php?f=6&t=828

The HP (Vertical) and LP (Crosstalk) slopes on the EEQ-12 are both second-order. Stepper slopes aren't needed for playback and the cutoffs can be set higher since there's unlikely to be any Vertical LF recorded information in Side to begin with.
FWIW Sterling Sound was one of my first customers for the EQ-12: They bought 8 channels for two lathes.

Though the EEQ-12 only removes the vertical LF component warp is mostly vertical anyway.
Simpler first, second or third order filters set below 20 Hz can be used to remove the lateral LF info from L and R which I've found to be much smaller.

If it were me I'd use the EEQ-12 to remove LF vertical warp. I've tried it and it removes the "wateriness," "Leslie" doppler effect and on 'phones "out-of-head" queasiness that LF warp creates.

I'll address the Jolida in another post after I find a couple of links.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

well the de-vinyliser was really a bit of history as to how I meandered down this path. I'm still in the camp where I would rather damp the LF wibblies but it slowly made the penny drop on how important cross-talk management might be. I'm also not sure if we have two separate issues or just two sides of the same coin. Right now open mind and fear for Wallet. At least I have plenty of real estate in my case :)

Aside: Haven't found time to work out exactly what is different between your elliptic approach and his. On a late night squint they look very simlar but I need to re-read your EEQ thread with his powerpoint open next to it.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Electrical Azimuth adjustment- aka crosstalk

Post by mediatechnology »

Douglas' approach uses two second-order sections plus a single-order allpass.
Three filter networks for a total of five single-order networks.

Elliptic EQ for vinyl mastering usually requires four channels: Two for modulation: two for preview.
Douglas approach, if adapted for EEQ, would be a nightmare to tune: 10 switch poles.

The EEQ-12 uses only two second-order sections which are "easier" to externally-switch requiring only 8 poles.
I still recommend relays however.
The EEQ-12's fc is inaudible enough that three choices are almost always enough.

The vinylizer as well as the Elliptic EEQ-12 are both derived filters.
The HP curve does not tell the entire story.
Those 18 dB/octave slopes he gets by monkeying around with the Allpass are going to cause peaking in the crosstalk curve.

EDIT: The other big difference is that, in my estimation, the devinylizer's crosstalk curve will be 6 dB/octave or less. The EEQ-12's is 12 dB/octave for both the HP and LP slopes. For the mastering engineer the difference is a smaller sonic footprint in the midrange.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

Interesting. As I said, not currently sure how much this is needed on replay once mechanical damping is optimised. But might be an extra cherry on top and I like Cherries:). I'm more interested in what happens higher up the band.

As a sort of on-topic aside. Have you seen any analysis of how Behringer do their width control in the DEQ2496? The manual tantalizes but falls short on detail. It has a seperate low frequency control but without detail other that 'twiddle this knob' hard to tell what it's up to without putting one on the bench. Given second hand prices it's almost a cost neutral option to buy one to try things out before deciding what is actually needed in the analog domain as you lose nothing if you then sell it. Negative WAF sadly.

Back on track. We know that channel seperation is usually 10dB worse at 10k compare with 1k. I've never seen any measumet below 1kHz though. I assume seperation drops with frequency.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Electrical Azimuth adjustment- aka crosstalk

Post by mediatechnology »

I do see the common link here between the de-vinylizer and the Jolida. The common thread is crosstalk; the difference is in what band.

This is a thread I started at lathe Trolls regarding vinyl polarity and crosstalk: https://www.lathetrolls.com/viewtopic.php?f=1&t=6661

I think we came to the ultimate conclusion that there is no ultimate conclusion regarding crosstalk polarity.

What I found was that for my test record/cart combinations (one record, two carts) both had out-of-polarity crosstalk.
Making it "more mono," with the Stereo Width Controller providing blending, I could increase the measured separation at the single test frequency by at least 10 dB.
In this case, not introducing crosstalk resulted in a wider image because the crosstalk polarity was reversed.
Its mostly a midrange thing and I wasn't able to listen long enough to decide which I liked better.

"YPMV." (Your polarity may vary.)

The Jolida crossfeed has independent L>R and R>L crosstalk with variable level and polarity for each channel.
That seems more universal than blending both channels equally.

Wouldn't azmith correction require a frequency-dependent adjustment?
I think that's where the Tilt EQ comes in...
What does Jolida do?

While you're at it have a look here: viewtopic.php?f=6&t=110&p=890
I think something like the above could be re-purposed particularly the +/-1 polarity-flipping gain stage.
A single order filter could be used to introduce a frequency-dependent effect.

And how do you calibrate the thing?
If it were me I'd want a mono lateral cut HF tone or band-limited noise disc and adjust playback azmith, either mechanically or electrically, while peaking the mono sum for maximum output.
That's how we used to final peak azmith adjustment on 2" multitracks.
The mono sum of 24 tracks provided a "high-Q" adjustment.
For two track machines a Lissajous display was best and that might be the case for cart alignment.

I'm not sure how well anyone would be able to increase separation outside of the midrange but its worth trying.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

Much to digest there thanks. And glad to see this has been discussed this century! I'll come back once it's sunk in :)

Calibration. Yes. tricky one that. I need to think about that some more. Not least because test records are in a lot of cases not that well calibrated as they did what they could in the day.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

My thoughts on calibration. I had started off thinking that a left channel only sweep followed by a right channel sweep then analyse off line would be the starting point. With that you could see the difference vs frequency across the band and also how different L-R and R-L cross talk were. Given we can only polish this t**d so far I would expect to see a half decent cartridge not requiring much adjustment. But I have no numbers on this beyond factory specs. I do hope that you could do a single adjustment rather than each channel seperately as gets complex otherwise.

Random notes on alignment. The usual method of azimuth alignment is to play a vertically encoded track and adjust for a null. Of course most test tracks only have a 300Hz vertical track on them so this doesn't necessary help much and also doesn't address stylus azimuth, which gets critical for certain stylus shapes. But vertical adjusted for null and horizontal adjusted for max should give the same answer, so that certainly seems a good way for live adjustment electrically.

So the homework for me is to actually record some single channel sweeps and see what the cross talk actually looks like then I will know the sort of solution needed, and then get my head around whether there is 'correct for cart' followed by 'correct for mastering' followed by 'season to taste'. if it's all three then the initial calibration only needs to get close. However that lathe trolls link suggests that mastering cross talk is down in the weeds in comparison.

A rapid prototyping capability would really help here. I would drop £150 on a second hand Behringer if they actually explained how it worked for width control. Might email them and see if they have tech support who know what they are doing. Will get me burned at the stake but having a rapidly adjustable digital eq to try things would seem to be a good thing.


Aside: was last night of the proms this evening. I have to say the BBC OB team do a superb job.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

Another idea for calibration here http://korfaudio.com/blog38

On a first read seems to have useful tests in for checking stylus azimuth but not sure it does anything for generator azimuth measuring.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

Re: Electrical Azimuth adjustment- aka crosstalk

Post by billshurv »

Have started fishing around for examples of recordings of left and right sweeps from people I know. I then realised that I had missed a very important consideration in my initial musings. Which is that, for a generator azimuth error it will be +ve phase on one channel and -ve phase on the other, so a simple blend will not correct for the case of a generator azimuth error with respect to the stylus.

Hopefully will start to get an idea of the scale of the corrections needed soon.
Post Reply