i can only say this.......those early steely dan CD's on MCA without the bar codes on the back try sure do sound good, i mean really good.
I was listening to the first CD pressing of Gaucho, the 1984 CSR Japan4US and noticed some big dropouts at 5:02 on the right channel of track 4. Examining this glitch revealed that it is the result of digital copying errors and not of dropouts on an analog master tape. Interestingly, this error is also present, albeit in less severe form, on the 1989? secret remaster, the 1993 Citizen box and the 2000 Nichols remaster. On the 1991 MFSL and the reissue of the CSR (mastering 3), which stayed in print in Europe until the release of the 2000 remaster, the error was 'repaired' by pasting in respectively 1.2 and 2 seconds of audio from a different source. This, and the post above by @JoshM about different track times inspired me to do some additional investigation. Some audiophoolery coming up... Comparing the waveforms of the JVC Japan Katy Lied, the CSR Gaucho and the CSR Aja with those of their 1989?/1993/2000 remasters revealed that, apart from track 1 on Gaucho, the samples consistently drift out of sync at a speed of about 33 or 63 samples (less than two thousandth of a second) per minute. Now, there's no chance in hell that a new transfer from an analog master tape could consistently lead to a drift this small, so the consequence of this must be that all these masterings have been derived from the same digital transfers. And that source could very well be those notorious early eighties 3M digital tapes. If these tapes were played back in 1984/1985 to create new digital masters a question is whether they re-digitized the analog output signals of the 3M machine or whether they succeeded in retrieving the undisturbed digital signal directly from the pins of the appropriate IC inside the 3M machine and converted the 50 kHz signal to 44.1 kHz, before it could be fed to a Sony PCM machine to create redbook masters. Since this must have been rather complicated to do in 1984 the transfer with an analog step seems to be more likely to me. However, a few years later a full digital transfer could certainly have been done by a decent technician teaming up with a good programmer. Or maybe 3M or Sony provided an adapter. If they, analog or not, transferred the 3M tapes again in 1989? for the secret remasters this could explain the slight drift of the samples compared with the earlier discs and it could also explain the different digital errors between the batches. Those old digital tapes are known to be not very reliable with weak error correction routines. My conclusions: The 1984/1985 Aja and Katy Lied CDs don't contain Steve Hoffman masters unless they were re-used for the 1989?/1993/1999/2000 remasters. This seems very unlikely to me. I guess the hunt for one of those 5000 copies can kick off. Nope for Gaucho. The MFSL Gaucho is a clone of the 1984 CSR with the same EQ. The only discs I encountered that seem to be made from unique transfers of analog tapes are the MFSL Aja, the two Japanese VDP discs of Aja and Gaucho, track 1 on the 1989/1993/2000 remaster of Gaucho and the JVC Japan4US of, again, Gaucho. (Why was this last one commissioned by MCA?)
@strippies - sone notes from Roger about the 3M decks that might be helpful - if you haven't seen them already: These early 3M digital 32-track machines did not have digital outputs, so the transfers were to be made via analogue cables into new 24-bit converters. There was no such thing as a 16-bit converter when the 3M machine was designed, so they used a unique combination of a 12-bit converter with an additional four bits of an 8-bit converter for gain ranging. This required a very expensive HP spectrum analyser to set the tracking of all the converter elements. also There were linearity problems with the 3M machines, but you could set the D-A tracking to match the A-D tracking so that the throughput of each track was linear unto itself. This meant that what you recorded on a track was what you played back on that track. This would be good enough for the transfers I needed to make. I did have a digital interface board that I built for the 3M, but if I'd used that, the digital transfers would have had the non-linearity of the A-D converter without the correction of the D-A converter. So, analogue transfers it was to be. From Roger Nichols: Digital-To-Digital Transfers Cheers, Paul
It’s amazing after decades and decades of discussing these SD things there are still vitally ripe rabbit holes to mine. Thanks @strippies for the research. How does the Gaucho mastering 8 fit in? It feels different enough somehow to me without any skills but I’m guessing you’d ferret anything out one way or another in five minutes flat.
and still, to this day, Robjohns's questionable "Disclaimer" still appears at the front end of the article. What a shame.
What do you find "questionable" about it? Robjohns is basically saying that the point of dithering is to get the DAC to produce smoother analog output, and I think that's essentially correct. Nichols' argument basically assumes that there is no dithering, and that the DAC produces an analog waveform that precisely matches the current sample at any instant; he never even mentions dithering as a factor.
Awesome work. Finding similar digital clicks on Aja is what helped me figure out which CDs were derived from the digital tapes. I also hear the dropout on "Gaucho" around 5:03 on the Citizen box, which is definitely from the digital tapes, as well as on the 2000 remaster. I don't hear it on my 37220 DIDX56 CINRAM, the 1991 2102 Japan CD, the MFSL, the 2003 SACD, or the 2010 SACD. (I don't think I own the 37220 CSR, so I guess that's on my "to buy" list.) You're also definitely right that the 37220 DIDX56 CINRAM and the MFSL have equalization that's incredibly, incredibly close. I am curious, though, what makes you think the CINRAM, MFSL, and other discs that lack the dropout have pasted audio? Given the difference in speed, I'd tend to think they're actually a different transfer of the analog tapes. The L/R balance differences between them seem larger than I expect if that weren't the case. For example, "Babylon Sisters" from the '00 remaster and Citizen box differ from each other by only .02 dB (which I'm guessing has to do with the '00 being more compressed). On both, the left channel is approximately .20 dB higher (.19 vs .21) than the right. On the MFSL, the left is .81 dB louder than the right, and on the CINRAM the left is .46 louder than the right. For the heck of it, I also normalized the volume of the CINRAM, Citizen, and 2000 to the MFSL's volume and put all three through MusicScope:
I prefer the remaster of Royal Scam to the old MCA CD. But in the case of Aja it's just the opposite. The remaster is brighter.
Does anyone else wonder why there has not been a remastering effort for all of the albums and released on vinyl and CDs? The last remastering of most of the albums occurred in 1999. It seems they are overdue for a major overhaul and updated release campaign. Many other bands have done this during this century so why not Steely Dan?
Check out the bump around 5:02 on the spectrogram of the Gaucho 37220 mastering 3. It's clearly visible that an edit was done. On the MFSL spectrogram it's a bit more difficult to spot but it's there.
I can confirm that I do not see a bump like that on my copy of mastering #8 (matrix MCAD-37220-2-S4C14, made in Japan, referred to above as "JVC Japan4US"), which makes sense if it's a unique analog transfer.
That's interesting. I'm not quite sure what that says about if/how it was edited, particularly an analog splice versus a digital paste. I'm not saying you're wrong or anything, just leery of speculating since there's been so much confusion about all the Steely Dan mastering stories. The one thing I did want to add about the Aja masterings is, as I wrote about here, the ones that we know to be from the '82 digital tapes all have the same clicks in the same spots. What's been commonly called Hoffman's mastering (and its clones) don't have that. Neither do ’84 Japan, MFSL, Takiguchi, and MQA CD masterings. (But the MFSL has its own unique error.) It definitely seems -- and this was the main point of Nichols's writing about the Steely Dan CDs, even if all of the specifics still aren't clear -- that different releases used different sources without a lot of consistency or clarity.
Assuming your Cinram 37220 Gaucho contains mastering #3 there's no need to buy the earlier CSR. They're digitally identical, except for that DIGITAL 'fix' on track 4, a 0.04 dB level shift on track 5 and a little inaudible copy error that unfortunately was introduced at the start of track 5. Like I said above, for whatever reason Babylon Sisters on the 89?/93/00 remasters is sourced from a new analog tape transfer, unlike the other six tracks. That's also the reason for the speed difference with the track sourced from the 3M tapes on the earlier discs. The MFSL Gaucho and CSR/mastering #3 are both made from the same digital source, presumably the 3M digital tapes but they are not digitally identical. That's because the MFSL is made from a new transfer of those 3M tapes. The question is whether MFSL had the technical expertise to do a full digital transfer, unlike the Nichols team, judging from his comments quoted above by @Paul P. MFSL didn't touch the EQ but it seems there indeed was some juggling with the gain between the two. Again, like I said above, the 1984 Aja was made from a different transfer of the 3M tapes compared with the 1989?/1993/2000 discs. Apparently the latter transfer turned out worse, with more copying errors, than the first. Quality control must have been lacking. On top of that the treble was boosted and an unhealthy amount of NR was applied in the digital domain. Generally more so on the 1993/1999/2000 discs resulting in even more treble than on many of the 1989? secret remasters. Nichols must have been a digital NR fan. Luckily for early CD collectors that technology was not yet available to him in 1984/85. For your amusement. Another graph of the Gaucho track 4 at 5:02 digitally pasting job. This time in waveform view between the CSR and the MFSL. I've put a vertical line directly on the sample were the paste edit begins. Check out the difference in shape between the waveforms to the left and to the right of that line.
Two possible reasons come to mind: Last pass was done by Don & Walt's fave engineer Roger Nichols. Since Roger passed, they may not want to engage with anyone else. Master tapes may be gone in the Universal backlot fire - possibly not much left but the 2000 digital masters anyway. Cheers, Paul
It’s interesting that the waveforms prior to the edit/paste look almost identical between the MFSL and CSR, three waveforms of the edit/paste look quite different.
1/ Nichols is examining the effect/benefit of 24- vs 16-bit sampling (i.e. A/D conversion) of the 3M digital recorder analog output. As he explained, the boundary conditions on his transfer project required that he use the 3M analog output. (as noted by Paul P in post #728 above) --- 2/ Note the pre-emptive caveat by Robjohns: "While I wouldn't want to discount Roger Nichol's [ and, N.B., a vast majority of audio engineering professionals's ] perceptions of the sonic improvements gained from 24-bit conversion over 16-bit..." OK, then where do perceived improvements in bass reproduction (and soundstage, and...) come from? --- 3/ Robjohns paras 2 & 3 contain... freshman mistakes. --- 4/ see /1/ above. Quantization noise is related to quantization step size. Time-varying audio signal (high-level) to input of high-resolution ADC > low correlation of qn to the audio signal. Time-varying audio signal (low-level) to input of high-resolution ADC > inherently higher correlation of qn to the audio signal. Correlated qn yields errors which are evident in frequency domain as harmonically-related components in ADC output spectrum. These would be undesired harmonics related to the bass signal's actual frequency components; such undesired harmonics can both corrupt the actual frequency components, as well as add completely false ones. Lower qn > more accurate bass reproduction, as Roger presented.
approx two yrs ago (in course of litigation, and notably after the original New York Times feature "The Day The Music Burned" named a lengthy list of artists), UMG finally revealed a list of those artists whose masters had been destroyed in the 2008 fire; artists mentioned did not include Steely Dan (one would think based on name recognition alone that they might have been mentioned if affected): "In new court filings..., Universal Music Group gave a basic account of which artists lost their master recordings in the blaze. The 19 artists on the list include Slayer, White Zombie, Soundgarden, Nirvana, Sonic Youth, Elton John, Beck, Y&T, Sheryl Crow, R.E.M., Bryan Adams, …And You Will Know Us By The Trail Of Dead, Jimmy Eat World, David Baerwald, Les Paul, Peter Frampton, Susanne Vega, Michael McDonald and Surfaris. The account details to what extent the damage of each of the artists incurred. For some artists, such as Slayer, Sonic Youth, Les Paul, Michael McDonald and Peter Frampton, masters were lost and no replacement copies were mentioned, inferring that those recordings could possibly be gone for good. Other artists, like Beck, had some of their masters destroyed, but high-quality replacements were available. Still others, like Elton John, were told that they had some masters affected and Universal is still working with them on figuring out just what was lost." Much more info in a similar Rolling Stone article about the Universal vault fire. Probably no clarity until litigation grinds forward...
A couple Steely Dan US analog masters were used circa 2014 for the SHM-(SA)CDs, if we're to believe the credits.
I believe the 2014s of Gaucho anyway are a reboxing and priced down reissue of the 2010. There is at least a decent chance it is a Japanese copy tape. The discog entry notes "New DSD transfer from Japan original master tapes by Manabu Matsumura." Steely Dan - Gaucho I seem to remember it is all in Japanese, but CDJapan is pretty good with translations and they note "Features the 2010 DSD mastering based on Japanese original analog tape." The likely one that might be new, which Universal did a fair amount of new remasters on Platinum SHM-CD around 2013 for other artists, but for Gaucho it also states the 2010 transfer. The 2010 SACD, if you look at the second picture in its discogs entries of its book does say in English "analogue master tapes"; but then the discogs entry does again reference Japan original master tapes Steely Dan - Gaucho which seems to come from the translation of the top of the back of the shm-sacd. Gaucho [SHM-SACD] [Limited Release] Steely Dan SACD
Steve Hoffman Info Site | Discography states that the Steve Hoffman remaster of Aja is MCAD-37214. It shows up here as Mastering 3 and 4 but not 7, which is stated to be Steve's. What am I to believe is correct? It's probably hard for Steve to keep track of his remastering releases.
You'll have to read thru the thread for the proof, but masterings 3, 6 and 7 are identical, bar a DC offset on 3 that was corrected on 6 & 7. Unfortunately, there's no way of knowing if this is Steve's or Roger's. We know Steve did the first pass, then the tapes were taken away by Roger who wanted to re-do them. Since these are the same base mastering, only MCA knows whose tape they used. The story goes there were 5000 Hoffman Aja discs pressed, but so far EAC has not found any evidence of them. So we have to go with the masterings we've found. Hope this helps! Cheers, Paul P.S. Mastering 4 is most likely by Glenn Meadows, which was re-used for Citizen Steely Dan. It's another post in this thread, but Glenn carefully added a touch of additional NoNoise and EQ to Citizen, which makes the EAC peaks slightly different
Earlier in the thread I made a case for mastering 6 being the same as mastering 7, only with a slightly different TOC layout. See second post on page 26. So that one should also be ”correct” if you’re looking for the mastering that may or may not be the Hoffman.
