DIGITAL Deep Purple - Machine Head (UK Quad Mix on the 2003 EMI SACD)

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

steelydave

Super Moderator
Staff member
Moderator
Moderator
Since 2002/2003
Joined
Apr 21, 2002
Messages
3,451
Location
Toronto, ON
I've had this one sitting in my DAW and been meaning to look at it for nearly 18 months, ever since it was mentioned somewhere that the rear channels might be swapped - the recent chatter about this title in a couple of threads motivated me to finally get this one done, and the results were kind of surprising, though given the fact it was mastered by Peter Mew maybe I shouldn't have been.

I'll be covering the following things here:
  1. LFE issues
  2. Channel balance issues
  3. Channel assignments
  4. Observations on differences in track duration and instrumentation (just Lazy and Highway Star)

Before I start I just want to have a grumble that because this album is stuck on SACD that I had to (and you will have to) convert it to PCM in order to make these changes. because there are no consumer-grade options for working with channel volume levels of .dsf DSD files.

I converted the SACD to 88.2kHz/24bit WAVs with the 'PCM Volume' setting at +3dB - if you're going to replicate these tweaks I strongly recommend not going any higher than this, because one of the tracks needs a +2.5dB output boost and anything more than +3dB boost on the conversion will push this change into the red.

1. LFE Issues

The headline here is that (at least on my rip and PCM conversiont) hat the LFE is one of the most egregious examples of the delay issue I first identified in 2021 that I've encountered so far, approximately 17.5ms out of sync with the full range channels.

The severity of the problem seems to be partially due to the fact that there's a chunk (about 6ms) of the Front Left channel duplicated in the LFE channel before the actual LFE content starts, as you can see in this screenshot:

Highway Star LFE glitch zoom.jpg


Suffice it to say that fixing this unlocks a considerable amount of bass, especially if your front channels have bass management at all.


2. Channel Balance Issues

Loading this album into my DAW I was immediately struck by the fact that in the waveforms, the Rear Left (Ls) channel was consistently lower in volume than the other three channels, as you can see in this screenshot:

(LFE channel deleted just for visual clarity)
full album.jpg


Now, I know from my earlier forays into this stuff (like the diagonal pans on the O'Jays Ship Ahoy quad SACD for example) the folly of making arbitrary changes to things when you don't know what the engineer's original intention was, but with this album, there seemed to be a simple answer to balancing the rear (and front) pairs. Despite being an SQ LP release originally, it breaks one of the format's cardinal rules: it has elements mixed to the center rear position:
  1. Highway Star - lead vocals
  2. Maybe I'm a Leo - electric piano solo (2:24)
  3. Pictures of Home - bass solo (3:41)
  4. Never Before - organ solo (3:32)
  5. Smoke on the Water guitar solo (2:58)
  6. Lazy - lead vocals (this track was correct and didn't need adjusting in the rears)
  7. Space Truckin' - percussion solo (2:52)
For the vast majority of these tracks the rear-center panned instruments were pushed slightly to the right of the sound field, and adding exactly 2dB to the Rear Left (Ls) channel put them bang in the center. On most of the tracks, there's a rhythm guitar in the Left Rear, and rhythm keyboard in the right rear, and it seems like (visually in the waveforms, and audibly) that the keyboards are a bit louder than the rhythm guitars.So me the Occam's Razor 'simplest answer is probably the correct one' is that the rhythm instruments in the rears should be the same volume and the center rear placements should be in the center, not that the keyboards should be a bit louder and the rear placements should skew a bit to the right.

The front channel balances were all really close to perfect (I tweaked a couple of them by 1dB) with major outlier: Track 6, Lazy. For some reason, the Front Left channel was a full 5dB louder than the Front Right, heavily skewing all the phantom center placements (lead vox, organ solo, etc.) to the left.

(Lazy waveforms)
Lazy (waveforms.jpg


I pondered this while looking at the waveforms for a good long while - how could this one track alone be so off from all the others? Then it struck me: they balanced the Front channels based on the organ improv at the beginning, rather than the actual full-band song itself. The reason this led to the 5dB skew to the left is that for the organ improv, the Front left channel should be quieter than the Front Right, and the evidence is right there in the rears: the Rear Left (Ls), which would be the "twin" of the Front Left, is considerably quieter than the Rear Right (Rs), (which would be the "twin" of the Front Right) in the intro.

You can see the imbalance of the fronts in this Voxengo SPAN screenshot:

(Light green is the Front Right channel, darker green is the Front Left channel)
Lazy (incorrect).jpg


This is during the organ solo - you'd expect some part of the frequency spectrum to overlap, but as you can see the front right channel is just uniformly quieter.

This is with the front left channel reduced by 5db:
Lazy  organ solo (FR +5dB).jpg


Almost perfectly balanced.

Strangely enough, Lazy was the only track where the rears didn't need any tweaking, something I re-checked about 5 times in different places because all the other tracks needed the Rear Left boosted by 2-3dB


So having said all that, these are my adjustments for the whole album:

(LFE -17.5ms)
  1. Ls +2dB
  2. Ls +2dB
  3. FR +1dB, Ls +2dB
  4. FR +0.5dB, Ls +2dB
  5. Ls +2.5dB
  6. FL -5dB, Master bus +2.5dB (you need to boost the overall volume of this track because of the reduction in volume to the front channel to get the same vocal level as the other tracks)
  7. FR +1dB, Ls +3dB

3. Channel Assignments

As I said off the top, the reason I first dug into this album was someone mentioned that the rear channels were swapped compared to the original SQ LP release. Short of having a photo of the master tape box, the only way to figure out if the channels are swapped is by ear (and by using my medium-sized intellect) and looking for things that either occur simultaneously in the front and rear speakers, or that move (pan) from one speaker to another.

I listened to the whole album on headphones, soloing various channels as I went, and the only usable occurrence of any of this on the whole album is, once again, the keyboard intro to Lazy.

It appears that they used 4 mics to capture this: one on the treble rotor of the Leslie speaker, one on the bass rotor, and a stereo pair some distance away to capture reverb or 'room sound'.

When the track starts, the treble rotor is in the front right, the bass rotor is in the front left, and the stereo room mics are in the two rear speakers. During the first bit of organ noodling, the two front mics slowly switch positions, crossing each other in the soundfield, until you now have bass rotor in the front left, treble rotor (with increasing distortion) in the front right, and the room mics behind you.

The next movement sees the two front channels slowly move to the rear to join the 'room sound' mics: the front left (bass rotor) moves to the rear left and the front right (treble rotor) moves to the rear right. This tells me that the channel assignments are correct, because if you swapped the rears, the treble rotor (the more audible of the two) would move from the Front Right to the Rear Left, which is impossible to do using a joystick panner because if you were doing that, at the point you were halfway between front right and rear left you'd also be halfway between front left and rear right, ie quad center, so you'd be hearing the sound coming out of all 4 speakers, not just the Front Right and Rear Left.

Now that isn't to say that the original SQ LP didn't have the rears the other way around, or maybe that it had both the fronts and rears the other way around (I don't know, I've never heard it) but it's my opinion, having checked my fair share of quad channel assignments, that the left/right relationships between the front and rear channels on this SACD are fine, as-is.

Simply put, as well, because (outside of that organ intro) there are no around the room swirls, diagonal pans or any other front/rear relationships in this mix, if you want to swap the rears because it makes the album sound better to you, you can with very little detrimental effect.

4. Observations on Differences in Track Duration and Instrumentation

First off, the Discogs entry for the SACD has the wrong runtimes for the multichannel (quad) layer - they've just been copied from the stereo layer. These are the actual quad runtimes:
  1. 6:15 Highway Star
  2. 4:57 Maybe I'm a Leo
  3. 5:07 Pictures of Home
  4. 4:01 Never Before
  5. 5:44 Smoke on the Water
  6. 7:11 Lazy
  7. 4:39 Space Truckin'
The only notable differences are Highway Star (6:05 Stereo vs. 6:15 Quad) and Lazy (7:22 Stereo vs. 7:11 Quad).

I'll get Highway Star out of the way first because it's just minor: starting around 19 seconds in, there are several extra bars of the intro groove (totaling about three seconds) before the distorted keyboard and then cascading screams come in. Presumably the stereo version had this edited out after it was mixed down and the quad mixers just didn't notice.

I think it's pretty well known that some of the final guitar solo on Lazy was edited out ont he quad mix, but the specifics are a little more interesting than that:

Starting at 5:59 there are a few seconds of lead guitar fills present in the quad mix that aren't in the stereo mix. Then at 6:06 (for about 14 seconds, through 6:20) there's an entirely different guitar solo than the one in the stereo mix, lower in register and not quite as aggressive. Maybe this was the one cut live with the band and then later discarded in favour of the overdubbed one that appears in the stereo mix?

Finally, the section from 6:19.5 through 6:32 (12.5 seconds) in the stereo mix is completely missing, edited out of the quad mix. The two mixes join back up during a big drum roll behind the end of the solo. I was kind of surprised to discover that only twelve-and-a-half seconds of solo were excised - in my mind for some reason it was 30 or 40. I think the combination of getting to hear an alternate solo, and only losing that small amount of music makes the UK quad mix not nearly as much of a problem as it may have seemed, at least in my mind anyway.

Here's a screenshot of the quad mix lined up with the stereo mix and split to show where the missing solo section is, which I think also illustrates how small an amount of music it is:
lazy quad vs stereo.jpg


Some might ask "why remove 13 seconds of a track at all?" The answer lies in the fact that when these mixes were done, EMI UK hadn't decided to release matrix quad LPs at all, so like the early RCA quad mixes of 1970, they were done entirely with Q8 release in mind. This meant often swapping tracks between the two sides (or programs) of the Q8, and in cases like Machine Head, editing out entire sections of music to try and get the running time of the two programs roughly equal. Other EMI Q8 releases also suffered from this - possibly the most egregious was Dark Side of the Moon, which had Money fade out halfway through the song at the end of Program 1 and fade back in to finish at the beginning of Program 2.

The (frankly, bonkers) tracklist for the UK Q8 of Machine Head shows you just how far they went in service of balancing the runtime of the sides, on top of editing Lazy:

Program 1
6:15 Highway Star
5:44 Smoke on the Water
7:11 Lazy (Original quad running time as per German LP: 6:40)
----
19:10 (or 18:40 with 6:40 Lazy)

Program 2
4:39 Space Truckin'
5:07 Pictures of Home
4:57 Maybe I'm a Leo
4:01 Never Before
----
18:44


If you made it this far I hope you enjoyed this deep dive into the quad mix. Despite its issues I love it - it has a lot of the flavour of the original stereo mix, but is still more than aggressive enough to please the discerning quad enthusiast. I think it's superior to the US quad mix, which sounds like it was mixed in Hollywood in 1974 (because it was) and the not-daring-enough 5.1 mix from the DVD-A.
 
Dave, many thanks for the sterling work in analysing this great SACD, and that's before the corrections you suggest. (I have owned the SQ LP since it was released). As this is a quadraphonic album I see that there is an active LFE channel that many other SACDs do not have, and that is also true of the Quadio releases. Is this SACD unique in being 4.1 or are there others. I have a vested interest in this as my car system is compromised without the LFE channel on quadraphonic albums. This is because the rear speakers do not have full range bass relying on the sub-woofer. As a result it sounds generally great in 5.1 but can be a little disappointing when the bass is predominately from the rear in 4.0. The Average White Band quad album being a good example of this. I am not grumbling about this compromised performance, just curious about which SACDs may be 4.1.

I shall try your suggested tweaks later in the year when I shall be mostly retired. Thanks again.
 
Thanks so much for your analysis, Dave. I'm wondering if the LFE and channel balance adjustments can be made on FLAC files from a rip of this SACD. I also wonder if some of the same issues are present on the three bonus tracks.
 
Look at them in MMH’s Channel Volume tool. You will see the bonus tracks look suspect so I also tweaked them a fraction Ls. Look at the rms value not peak. Rms is the average level for the entire track, all those channel volume adjustments move the rms closer to the fronts.
 
@ForagingRhino I understand you are not releasing any quad mixes of Machine Head anytime soon, but the person above talks about alleged issues regarding channel levels on the UK quad version, which may be important if that release ever comes down the pipeline.

I'm available on a consultancy or full-time basis if anyone looking in here needs someone to do QA/QC/project coordination for any type of legacy quad reissue/modern surround project! :giggle:
 
Dave, many thanks for the sterling work in analysing this great SACD, and that's before the corrections you suggest. (I have owned the SQ LP since it was released). As this is a quadraphonic album I see that there is an active LFE channel that many other SACDs do not have, and that is also true of the Quadio releases. Is this SACD unique in being 4.1 or are there others. I have a vested interest in this as my car system is compromised without the LFE channel on quadraphonic albums. This is because the rear speakers do not have full range bass relying on the sub-woofer. As a result it sounds generally great in 5.1 but can be a little disappointing when the bass is predominately from the rear in 4.0. The Average White Band quad album being a good example of this. I am not grumbling about this compromised performance, just curious about which SACDs may be 4.1.

I shall try your suggested tweaks later in the year when I shall be mostly retired. Thanks again.

I can't provide an exhaustive list off the top of my head, but there are some quad reissues that have .1 LFE tracks - these include some of the early '00s Sony SACDs of the O'Jays Ship Ahoy, The Isley Bros. 3+3, and Jeff Beck's Blow by Blow. Many of the more recent Sony Japan ones are too, like Beck Bogert & Appice and some of the Santana albums.

Automotive surround is really outside of my experience - you should post this query in our Automotive Surround section as it will get much more visibility, but regardless of whether an album is 4.0, 4.1, 5.0 or 5.1, you should (ideally, theoretically) enjoy the same amount of bass response from your system.

This is surely the point of bass management after all, to route the low frequency signals to the speakers capable of handling them, isn't it? Like I said, I don't know much about automotive surround implementations, but in a home theater setup, if you tell your AVR you don't have a subwoofer, it routes the LFE content to your main front left and right speakers, and similarly if you set up the crossover frequency for your rear speakers to reflect the fact that they can't reproduce low-frequency content, it should throw that audio information to your subwoofer, or front main speakers.
 
Thanks so much for your analysis, Dave. I'm wondering if the LFE and channel balance adjustments can be made on FLAC files from a rip of this SACD. I also wonder if some of the same issues are present on the three bonus tracks.

Yes, definitely, this was how I did mine, more or less - buried in my original post somewhere was the explanation of how I converted the SACD .dsfs to .wav using foobar - the same would go for SACD to flac, you just have to change the output file format. . I only used .wav because the DAW software I was using at the time I did the conversion didn't allow for direct importing of flac files. You should be able to do all the adjustments in my newly-crowned assistant @HomerJAU 's Music Media Helper software - do the LFE offset first (you can't do minus numbers so rather than -17.5ms the subwoofer, +17.5ms the other 5 channels) and then do the channel volume adjustments afterward.

I didn't bother with the "bonus tracks" as they're from the Warner US DVD-A - I have this whole disc ripped separately and if I was going to tackle those tracks I'd probably do them along with the whole 5.1 album mix. Also, the purist in me would rather work on the original 24/96 PCM files, instead of 24/96 transcoded to DSD for the SACD and then transcoded back to 24/88.2 in order to work on.
 
Yes, definitely, this was how I did mine, more or less - buried in my original post somewhere was the explanation of how I converted the SACD .dsfs to .wav using foobar - the same would go for SACD to flac, you just have to change the output file format. . I only used .wav because the DAW software I was using at the time I did the conversion didn't allow for direct importing of flac files. You should be able to do all the adjustments in my newly-crowned assistant @HomerJAU 's Music Media Helper software - do the LFE offset first (you can't do minus numbers so rather than -17.5ms the subwoofer, +17.5ms the other 5 channels) and then do the channel volume adjustments afterward.

I didn't bother with the "bonus tracks" as they're from the Warner US DVD-A - I have this whole disc ripped separately and if I was going to tackle those tracks I'd probably do them along with the whole 5.1 album mix. Also, the purist in me would rather work on the original 24/96 PCM files, instead of 24/96 transcoded to DSD for the SACD and then transcoded back to 24/88.2 in order to work on.
Yes, I already have the FLAC files, even the bonus tracks, which I wasn't aware are from the Warner US DVD-A, so it should be easy enough to make the adjustments you posted.

I actually listened to the album on a recent road trip (Highway Star, natch), and it sounded like something was off.
 
I remember that Machine Head was one of the few albums released in both SQ and CD-4 (I owned both). The mixes were very different between the two formats, not just in instrument placement but in bass depth, echo, etc. It would be interesting for another contributor to make his own comments.
 
So these are the results I got after applying the channel volume changes in MMH:

MHSACD Adj Lev.JPG


I tried to apply steelydave's changes to tracks 1-7 and apply similar changes to tracks 8-10. First, it appears I may have misunderstood the recommended changes to track 6. I lowered FL by 5 db and raised the other channels by 2.5 db. Now the FL and FR are unbalanced and lower than the FL and FR on all the other tracks except for 9 and 10.

I could go back to the previous version of track 6 and apply the correct changes, or there might be changes I could make to this version that would accomplish the same thing. Reading the MMH documentation about this tool and looking at these results led me to consider a couple of things:

  1. Should the RMS values of FL and FR be more or less consistent across the ten tracks?
  2. Should the RMS values of SL and SR be more or less consistent across the ten tracks?

If the answer to one or both questions is yes, then I might proceed with using some of the sub-tools within Channel Volume to achieve desired results.
 
I tried to apply steelydave's changes to tracks 1-7 and apply similar changes to tracks 8-10. First, it appears I may have misunderstood the recommended changes to track 6. I lowered FL by 5 db and raised the other channels by 2.5 db. Now the FL and FR are unbalanced and lower than the FL and FR on all the other tracks except for 9 and 10.

Yeah, you have misunderstood slightly - you need to -5 the front left, and then boost the whole mix (all 6 channels) by 2.5. I guess the way to do this in MMH's channel volume editor would be this:

_FL: -2.5
_FR: +2.5
__C: +2.5
LFE: +2.5

_SL: +2.5
_SR: +2.5
 
Yeah, you have misunderstood slightly - you need to -5 the front left, and then boost the whole mix (all 6 channels) by 2.5. I guess the way to do this in MMH's channel volume editor would be this:

_FL: -2.5
_FR: +2.5
__C: +2.5
LFE: +2.5

_SL: +2.5
_SR: +2.5
Thanks, I like easy fixes!
 
@ForagingRhino I understand you are not releasing any quad mixes of Machine Head anytime soon, but the person above talks about alleged issues regarding channel levels on the UK quad version, which may be important if that release ever comes down the pipeline.
When the time comes, the US quad mix will be offered up.
 
Back
Top