Correcting HAECO-CSG Processing

I tried it, but did not notice much, if any, difference. I haven't had time to listen much though. I still can't figure out exactly what this vari phase rotation is supposed to do, since their documentation is so vague. If so, I could at least try to listen for whatever it is supposed to change.

One thing I may do (which I hate to do), is narrow the stereo image slightly. I noticed on playback that the center information is a bit weak. Narrowing the stereo image will create more energy in the center, which might balance things out a bit. It seems like vocalists, instruments, etc. in the center are slightly distant. I have a feeling they mixed the album with CSG in the chain, so they could hear the final mix as the end listener would hear it. I am only looking to do maybe an 80/20 mix of left/right (and vice versa) to see what happens. I also still notice that bass is a little on the weak side yet as well. Thing is, we are so used to hearing that "bloated" bass sound from CSG that we just may not be used to cleaner bass after the processing is removed.

I'll keep you posted. This has been exciting--it is like hearing these as new albums!
 
What I understand vari phase rotation does is go step-by-step gyrating the waveforms on any of the two channels to achieve a symmetrical waveform. This is why, since HAECO-CSG is futzed with beforehand, I corrected that and then went a step further asking my friend to run the vari phase rotation on the resulting file to see if there was anything that changed afterwards.

What I hear, as I mentioned above, is less reverb behind the voice. This might be due to the fact that the reverb itself is out of phase.

I wonder what vari phase rotation can do with songs such as Layla where I understand that the drums are mixed in out of phase.

Please do post any new experiments you make. I would like to, but I do not have the software to continue with my experiments.
 
This is interesting...

http://berk3.home.xs4all.nl/Info_24-96_Updated_26-08.txt

This is the interesting passage:

I have decoded the HAECO-CSG process by correcting the phase of the right channel using a nifty VST plugin that allows exact phase shifting. Then I synced the channels together up to the millisecond. With these two steps the phase between the left and right channel was restored as closely as possible (this certainly couldn't have been done in the analogue domain).

I did play around with a "azimuth" adjustment, thinking I may have heard a difference, but may not have.

The only reason I could see applying some sort of sync to the channels would be due to the CSG unit being a longer signal path, and the electronic components may have delayed the sound in that channel by perhaps a tiny fraction of a second (on the order of very few milliseconds). I wonder if this would bring the sound into focus more (although it is already quite good).
 
The first HAECO-CSG correction I did was over 2 years ago using a plugin. But the results were not as good as now.

If what you want to do is correct the HAECO-CSG I would use the 90º (+ or -) correction on the right channel and, if I wanted to correct the time-related phase problems, use vari phase rotation.

Of course that doesn't preclude playing around with other possibilities as I see you are. But the above is, AFAIK, the more to-the-point approach.

Have you tried the HAECO-CSG correction plus the vari phase rotation as second step? If so, have you noticed any difference between the results of only the HAECO-CSG correction, and of these and the vpr option?
 
No, I haven't noticed any difference with or without the vari phase option as of yet. I think what bothers me about that plugin is that it is not static--it varies its output. There is also an azimuth filter in the program that has an "automatic" setting, and the sound quality wavers all over the place as it adjusts itself. I am thinking that if there is time shift, it is going to be a static number of milliseconds.
 
Rudy, don't think that azimuth filter results will be the same as vari phase rotation. Before you said that analogue equipment tended towards variation to its basic nature. I suggested vari phase rotation because any variations that the electronics might introduce in the coding of HAECO-CSG would, theoretically, be solved by phase correction via rotation in a sequential time-based manner, which is what I understand that vari phase rotation does. Given this, it certainly varies its output, but it doesn't happen in real time. I think that the plugin must read the file some time before effecting a change and makes sure that both the waveform in the right and the left channel are phase identical.

As I mentioned above I do hear differences. For one, the reverb behind the voice loses presence. I suspect that this is because vari phase rotation eliminates certain out-of-phase elements present in the reverb, which must be what gives it depth. So, to me, the vari phase rotation treated file shows less reverb than the version I sent my friend to be processed. This must also be an indication that vari phase rotation does solve most, if not all, phase issues in the file. This might be good in certain cases, bad in others.
 
I agree--it could be that it is eliminating a few artifacts of the CSG process, but also getting rid of good phase differences, the ones that convey depth and location of the sounds within the soundstage.

The one remaining process I'm curious about was the time delay added to one channel. I had read elsewhere of someone who had done a similar 90 degree phase correction, but then applied a slight delay to one of the channels (only by a millisecond or two), which is pretty much what that azimuth filter does. By only a few milliseconds, though, I do not hear any difference in the sound. I know having a filter circuit on one channel and not the other could lead to some delays, but...by how much? Would capacitors and inductors really contribute any significant delay?

I also just thought of something. To undo CSG, why didn't they just use it on the opposite channel? Think about it. If you bumped the left channel ahead by 90 degrees, couldn't you just bump the right channel by the same amount to sync them back up again? The only drawback is that the components inside any existing CSG processor would probably have deteriorated, and would need work.
 
Both channels would probably in sync by changing the left channel, but then both waveforms would be rotated to a 90º angle. I do not know if that would cause havoc with the peaks or not.

My philosophy here is that the simpler the tweak the better. Once one starts piling too many tweaks or doing strange things it might be fun, but it would probably lead into a labyrinth of secondary issues. It's much like the old audiophile adage, "the shorter the signal path, the cleaner, clearer, the sound." In the same way, the least strange tweaks the purer the results should sound. The issue at hand is which sole movement solves the problem best. Since what the literature I've read about HAECO-CSG, IIRC, mentions that the tweak was applied to the right channel, then that is where the tweak should work the best with the least file manipulation.

To end up with a clearer-sounding file is the same reason behind the importance of increasing the bit depth to 32 (or, at least 24) bits, before tweaking the file. And I would add increasing the sampling frequency too to, say 96kHz. My ears tell me that the results are more transparent when one has ended the tweaking and returns the file to 16/44.1. And, since you have RX Advanced, you've got the best, most transparent SRC and dither (Ultra MBit+ dither) for that purpose.
 
Here's a weird one: While doing today's AOTW for Jimmie Rogers' WINDMILLS OF THE MIND (SP-4187), I was referring to both a CD issue on Collectors Choice and the original LP. My LP is a white-label promo and I noticed a gold sticker on the back heralding the CSG processing. Sure enough, a quick glance at the disk itself shows the telltale heavy-looking grooves of HAECO-CSG. The white-label even had a CSG number on it for the matrix number.

Yet the CD has none of this nonsense on it. It sounds like it should, with everything in proper balance. The vocal track disappears when I use the karaoke setting on my sound card's software.

But I put the LP on the turntable and listened to a couple of tracks and sure enough, the uncomfortableness of the center channel reared its ugly head.

My suspicions are, that stock copies of the LP may not have the CSG processing. I'd need someone who owns one to verify that. It may be that A&M CSG'ed the thing for radio stations but left the main stock copies alone, thus when Universal lent out the masters to Collectors Choice, they gave them the original non-CSG'ed tracks.

Harry
 
It is possible they pressed some promo-only records for radio use. :agree: I don't think I have Windmills here though.
 
Hello everybody,
I am new to this forum. My interest in HAECO-CSG is twofold. First of all I really like Sergio Mendes and brasil 66. I also have some Iron Butterfly MONO 45's made with this system. It is a bit of a mystery to me why anyone would want to apply this system to mono, but it really happened and the result is awful. That's unfortunate because the recording is the very rare edited version of the album tune In The Time Of Our Lives. Fortunately I also have a mono Italian 45 of the same edited song, and it sounds superb.

I recently bought a 1-dollar A&M compilation LP called Magic Box SP 19006 and the rear sleeve stated that CSG is used. I am going to have some fun using your suggestions to reverse the effect and judge the results. What I have understood is to apply a 90 degree (plus or minus) phase change to the right channel.

I use Adobe Audition CS6, stand-alone version. It has the graphic phase editor. I will let you all know what happens, but it may be a couple of weeks.
Never mind, hobbies cannot be hurried.

Bas Möllenkramer
Amersfoort
The Netherlands
 
I have an interesting addendum to the CSG subject and A&M. Easily one of the most popular albums suffering from CSG is Sergio's FOOL ON THE HILL. To the best of my knowledge, there are no known stereo versions worldwide that DON'T have the CSG processing, and to date, all CDs of the album continue to be issued with the CSG intact. Even Sergio himself, I'm told, is dismayed over that damnable processing on that album.

The only way, short of using the software methods described in this thread, to not have CSG processing is to resort to the somewhat rare mono version of the LP, or the mono versions found on 45 singles.

One way of getting the stereo version of the title track - without CSG - and without software tricks - is to find a white-label-promo of the single. I have such a single. It's numbered 961-S with the matrix number of S-1521. It was pressed by Monarch and has "So Many Stars" on its b-side, also in stereo. It's possibly the only stereo issue of the song without the CSG processing and it sounds amazingly like Rudy's de-CSG'ed version.

FoolHillStereo45.jpg

I do not know if there's a counterpart to this for "Scarborough Fair". I don't have one in my collection like that - my only "Scarborough Fair" single is a stock mono copy.

Harry
 
This is the only promo I could find on Discogs for "Scarborough".

Sérgio Mendes & Brasil '66 - Scarborough Fair / Canto Triste »

And since the photo does not specify stereo, I would guess it is mono. And I take it the "-S" after "961" above denotes stereo?

I might be heading to Jerry's Records in mid May with a group of friends who are meeting up there. He stocks over 750,000 45RPM singles (and actually is trying to sell them off as a lot--seems he wants to get out of the 45RPM biz). Chances are this title could be among them.
 
I'm beginning to think that with promos, A&M might have done special runs for AM radio stations and FM radio stations in mono and stereo and because of their rarity, don't always populate Discogs. The one I have for "Fool On The Hill" for example - Discogs lists the title as having the same song on both sides, yet mine has "So Many Stars" on its b-side - and that is not listed on Discogs.

Harry
 
If it's not there, then it's up to us to add it in. The data is primarily user-contributed. For promos without listings, they may be out there, but someone who is a Discogs member has to take the couple of minutes to create the listing for it. But even with the missing data, they are about the only site around that lists so many variations of each album or single. I know that some release differences even get down to differences in the deadwax (which may indicate a different mastering or simply a different pressing plant). Just listing items in a discography isn't sufficient these days--it's that in-depth release information that makes Discog a growing valuable resource.

I know I have encountered some variations myself that I need to add when I get some free time. There is a way that you can make a duplicate of an existing release, modify it, then submit that as a separate release. That saves a lot of time.

I think I have a "Scarborough Fair" single here but I probably only played it once. It's nothing I would have bought.
 
Seems CSG appeared on singles sporadically--would it be on promo singles only, or stock copies? (Although I know of one or two stock copies that had CSG.)

It was a dumb move, really--ruining the stereo sound to improve the folded-down mono...
 
Being a 40-plus-year music business veteran, I can tell you that during my tenure in the business (from 1969 until around 2013), reel-to-reel was never considered the "definitive" or "highest quality" format for major label releases -- maybe it was earlier, since most hi-fi nuts from the '50s had a nice reel-to-reel deck. In the era leading up to the CD, and even when cassettes were outselling records, the LP was still the headline format and it received the most attention with things like Quad, "direct to disk" mastering, heavier vinyl, things like that. Of course LP was king due to its popularity and to a lesser extent, its size; that's why it had the most deluxe packaging of all the formats.

A format's list price has nothing to do with its quality, though. List price is calculated based on what the item costs to manufacture, and consideration is also given to the cost to produce the recording itself, how many defects there will be, how many returns are expected (on average and based on past history), and a bunch of other smaller things like shipping, promotion cost, and so on. The only reason reels cost more back then is because they didn't sell as well as LPs, and they cost more to make. Packaging costs were probably less with reels than with LPs, due to reels not having elaborate sleeves and occasional extras like LPs did.

I don't really think any one format got more consideration than any other, soundwise. Occasional tweaks were added to all the formats at one time or another - Dolby NR on cassettes, higher quality tape, etc. But in the boom days of stereo and new formats like 8-track and cassette, music companies were mainly just cranking out as many copies as possible. Quality was not the largest consideration in any format -- quantity was, which is what led to massive griping from audiophiles about LP quality starting in the late '70s. (I remember one guy going through five copies of Heart's Dog and Butterfly before finding one he was happy with.)
 
For me its always been Quality over Quantity. And until CDs Vinyl was my first choice and cassettes were my secondary ( as well as experimental for recording the vinyl making mix tapes as well as my radio Djing etc.) What ever format is more solid and durable as well as Best sound quality. And will last a long time. Is the one i generally go with and since the early 90s for me its been the CD And CD-R Formats and now with downloads and Cd Ripping to computers ( which i am a newbie to) i plan to keep everything i have i cannot justify replacing my library again and spending a ton of money i was fortunate just to replace the albums i used to have on vinyl with the Cds ( and not to mention the lps i still have which will never be on CD or availiable as digital files) so in my opinion i say Enjoy what you have and take the best care of it as possible regardless of format
 
After finding out about a plug-in for Audacity to allow channel phase manipulation (called PhaseBug), I've been playing around with it and it seems that not all CSG albums and songs are created equal. That is to say that *many* of them simply use a +90° shift on the right channel (corrected with a -90° shift). But I'm finding that some albums or tracks work better at different settings. Right now I'm working with the YE-ME-LE album from Sergio and I'm finding that the right channel shift works better at -60°, indicating that it was originally shifted +60° in the right.

In the post-FOOL Sergio era, it appears that Lani and Karen each had slight shifts to one side or the other. After correction, I'm hearing Lani coming from slightly left of center, and Karen coming in slightly to the right of Lani. This all sounds clearer on YE-ME-LE after a -60° phase correction.*
 
Last edited:
* Except "Masquerade" and "What The World Needs Now", which sound better with a 90° phase correction!
 
Last edited:
60° wasn't an option with CSG--only 90° and 120°, per their instruction sheet.
 
Wouldn't +60° in one channel be equivalent to -120° in the other channel? I'm still trying to work the math in my head...
 
Wouldn't +60° in one channel be equivalent to -120° in the other channel? I'm still trying to work the math in my head...
I don't think so. Using the circular display in PaseBug, if I go +60 in the right channel on say "Some Time Ago", it pretty well neutralizes the lead vocal - moreso than at +90.

PhaseBug Plus60 Right.jpg

Now, if I move the red(right) channel to 0°, and set the yellow (left) channel to -120°, there's no cancellation effect at all when OOPS'ed.

PhaseBug Minus120 Left.jpg

You can also see that the angle between left and right as if hands on a clock is way different. My experiments seem to indicate that any angle made by the "clock hands" will yield the same result if shifted around the dial. So my +60 right would give the same result as +30L and -30R. Or +60L and 0right.
 
I'm looking over the instruction manual again. The center channel buildup control has three settings, and only the +3 position (+3dB) was recommended for "all compatible tape and disc mastering." 0 and -3 were only for combining stereo into mono, such as when creating a mono LP or tape. (Cutting a record with more than 90° of phase shift would result in too much vertical information in the groove, making cutting and playback difficult.)

The thing that bothers me about CSG is that 90° seems like a moving target. Did the engineers really adjust the azimuth of the tape head prior to creating every master tape with CSG processing? And were the phase shift networks inside the CSG unit calibrated properly? They were calibrated properly when the units left Holzer Engineering, but it bothers me that the circuitry was analog and would have been prone to component values drifting and, if there are indeed different phase shift networks for different parts of the frequency spectrum, what happens when they all start slowly drifting in value as they age? (There is a note in the instructions to return the CSG processor to Holzer if it needs to be recalibrated.) With digital, any phase shift would be set in stone.

The out-of-phase check will only work if a signal is of exactly equal electrical amplitude in both channels. I know on some recordings, vocalists are not quite centered.

The CSG processor also had a phase invert switch for one channel, which was used for setting azimuth.

It'd be interesting to see this on an oscilloscope. Only I currently don't have access to one at the moment.
 

Attachments

  • HAECO_csg_instructions.pdf
    447.7 KB · Views: 1
Back
Top Bottom