Correcting HAECO-CSG Processing

I've been playing around with this plugin called PhaseBug that works with Audacity and other audio software I believe. It allows one to literally dial in the amount of phase adjustment one wants on either or both channels of stereo. And since FOOL ON THE HILL is a particular favorite of mine, it was worth as much time and effort as I am willing to give the effort.

I had thought this issue was put to bed a few years ago when our Rudy found a way to twist the right channel of the album back the 90° phase shift that was altered by the HAECO-CSG system all those years ago. Initially, I was thrilled to hear Lani and her vocal cohorts much more centered in the mix than ever before. But there was something that was still bugging me. It still wasn't perfect. I was hearing both a slight imbalance in the channels and some kind of muddiness in the bass.

So then this PhaseBug thing was pointed out to me and I thought "Here's my chance to play around with this album and see what I can come up with." I initially thought that maybe it wasn't a pure 90° phase shift. Was it 89°? 85°? 120°? With PhaseBug, all such changes are possible and one can instantly listen to the results. With my sound setup on the computer, I can set the output to do a canceling of the center channel - what is known as OOPS (out-of-phase-stereo) - or a Karaoke setting. And while this is in effect, I call up the PhaseBug in Audacity and dial the red and yellow circles to any degree I want. Sure enough, 90° in the right channel seemed to be the best. Maybe that was the best we could do?

Then I wondered if altering the left would work. Well, I found that it sort of didn't matter if the left was altered as long as the right was 90° away from it on that one side. So I could dial the thing around the circle anywhere I wanted as long as the right and left were 90° apart in the correct direction. And that got me to wondering what would happen if I altered the left to a +45° phase shift and the right channel to a -45° shift. They were 90° apart, and the OOPS setting was working, eliminating the vocal pretty much. So how would it sound.

My impression was that this resulted in a somehow cleaner adjustment of the phases. Things sounded more balanced. The bass didn't sound smeared like it had before. So, I continued tinkering with this adjustment as the basis for the correction.

As each track was adjusted, I then went into the waveforms, looking to see if the phasing between channels was aligned correctly. This entails finding a section that is supposedly centered - like a solo vocal with no other instruments playing. It's not always guaranteed that you'll find that kind of section in every track, but you can usually find something to key in on. This is where you want to assure that the very height of a wave is in exactly the same horizontal position as its corresponding wave in the other channel. They need to line up exactly to be in perfect phase.

The HAECO thing, I don't believe was responsible for the discrepancies I found on this album. Truth be told, there aren't too many recordings out there that don't need any phase adjustments. There are a lot of recording heads out there, and not all of them are (were) in perfect phase when the masters were created. I can't tell you the number of times I've run across tracks out of phase by three or four "samples" in one direction or another. The fix to this is to find where the peaks or valleys line up, split the tracks into two independent tracks and using the back/forth arrow tool to slide one track in the direction it needs to go. Once this is done, the OOPS effect should be even more impactful, eliminating the center vocal almost entirely. Every now and then, a slight adjustment of the volume of one channel is needed to get that vocal to really disappear as far as it can.

And all of that is what I've done to FOOL ON THE HILL. Track by track, examining samples, adjusting phases 45°, adjusting channel balance, and I think it sounds really good now. Is it THAT much better than what our Rudy came up with? Hard to say, but I think so.

One more factor, that I forgot to mention was the choice of the files to use for this experiment. Way back when, I believe Rudy used the Verve remaster of the album. I tried that too and found that it was a bit too peak limited - and it got worse as I adjusted the phases. So I checked an old Japanese 24-bit remaster. It was pretty well brickwalled too and when I tilted the phase, the brickwall distortion got worse. Finally I went back to the old A&M CD (same as the Rebound version), and found that the waves weren't anywhere near maximized. They were nicely lower in level and the phase adjustment didn't move too many of the samples into the "red" territory.

I played the result of my correction on FOOL ON THE HILL as a CD-R in my car and it just sounded wonderful to me.

I'm going to play around with other albums too, just to see if they improve. If anyone wants to hear these tracks, let me know and I'll upload them somewhere.
 
The HAECO thing, I don't believe was responsible for the discrepancies I found on this album.
It actually could be responsible, though.

The CSG ran a signal through one channel only for its phase processing. The other channel had to use some external components to "delay" it electronically, as the phase processing had a slight delay going through all the components that made it work and they wanted to be certain the unprocessed channel matched it in the time domain.

So to find a phase issue at this late date? Doesn't surprise me, really, as something I mentioned before is that electronic components are never an exact value--they are within a range of values, called a tolerance. So a capacitor may be rated nominally at 100µF but depending on the grade of capacitor, that tolerance can be 2%, 5%, 10%, or more commonly 20% in most general-purpose capacitors. Resistors, same thing. There are expensive resistors with tight tolerances, but they also have a similar tolerance of values. Coils/transformers, too, can vary slightly as well.

According to one site I read, it was thought that CSG divided up the signal into distinct frequency bands and applied the phase shift to all of them, then combined the output. So that's even more of an anomaly if a component's value is off. Who's to say all of the frequency bands are working exactly in tandem? In theory, phase issues could end up being present in some of the frequency bands and not others.

So the point is, how exactly can we be guaranteed that an exact 90° phase shift in one channel, along with the phase of both channels eventually lining up, can be accomplished with analog equipment? It's close...very close, actually...but it's not exact like digital can be. So that goes back around to having to tweak the phase adjustment and make other edits to help sort it out. A straight 90° can work well but there's always room for improvement.
 
I'm going to play around with other albums too, just to see if they improve. If anyone wants to hear these tracks, let me know and I'll upload them somewhere.
I would greatly enjoy hearing what you've come up with.

While all of this shop talk is over my head (I trained as a chemist, not an electrical engineer), from the grandstands, I wonder if any difficulty in finding a definitively quantitative corrective action is related to this being an early -- if not the first -- HAECO LP A&M issued, which in itself would suggest numerous on-going immediate actions in response to numerous excursions with the new technology. I read that Ry Cooder's Bop Till You Drop -- you'll recall the hype, "rock's first digital album..." -- was met with continuous engineering issues owing to the then-newly developed digital recording technique.
 
Back
Top Bottom