Rev 2/28/22 (see Notes)
Steely Dan has a reputation for being audiophiles, and Roger Nichols, their main engineer, was a proponent of digitizing analog recordings to preserve them. So, I find it hard to believe that they didn't digitize their analog masters (either the multi-tracks, the original stereo masters, the safety copies which Roger Nichols made when each album was new [1], or all of these), with a good ADC until 1998.
In fact, an article entitled Roger Nichols: Digital-To-Digital Transfers from the May 2006 issue of Sound on Sound indicates that Nichols used a 3M digital recorder to copy analog multi-tracks in 1982, and that he was transferring at least one of them to Pro Tools in 2006, which just happens to have been the 30th anniversary of Aja, when the legendary Cisco LP was released:
"I am currently transferring digital multitrack tapes from early (1982) projects into Pro Tools for surround mixing. I remembered some of the problems with early recordings that related to early A-D and D-A converter designs. During these transfers I wanted to correct any of the early shortfalls, if possible. Even though they were 16-bit recordings, transfers to 24-bit would help preserve the accuracy of the original recordings.
"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 [something more advanced than the Apogee AD-8000 8-channel 24-bit ADC which he obtained in 1997/8, I presume]. 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. [I gather that this means that 4 bits of the 8-bit DAC were used for controlling the reference voltage on a 12-bit multiplying DAC.] This required a very expensive HP spectrum analyser to set the tracking of all the converter elements.
[...]
Converter Tracking
"Let's compound the issue a little. A-D converters have the same problem. The error for each bit during the recording is added to the error for the bits during playback. In early digital machines they hand-matched A-D and D-A converters to match closely to get the best sound on each track. If you had to replace a converter, you were in big trouble unless you replaced both with a matched pair. SINCE A-D AND D-A CONVERTERS BASICALLY WORK THE SAME WAY, SOME MACHINES USED THE SAME CONVERTER FOR RECORDING AND PLAYBACK TO AVOID TRACKING PROBLEMS. [In other words, the linearity error of the recording was canceled out by playing it back through the same DACs which were part of the ADCs which were used in the A-to-D conversion-process. There would still be some quantization error of less than 1/2 LSB, but this would be way down in the noise, and dither might not have been required. The catch is that this required the same 3M deck which was used for making the recording to be used for playing it back, perhaps decades later, which would have been risky because converters can fail. So perhaps two or three recordings were made on separate decks running in parallel to ensure that at least one of the decks would survive for a few decades. It would have been an expensive approach, but it would have provided extra security because someone couldn't steal a tape and play it on just any deck without losing sound quality. It was if the ultimate sound quality was encrypted, and that the key to obtaining it was to play it on the same deck that was used for recording it.]
"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 [2], 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. [Well, actually, he wanted to make a 24-bit copy, and playing it back on the deck which was used for making the recording (i.e. each channel was played back through the same DAC which was used in that channel's ADC), and filtering it with the 3M deck's (analog) anti-aliasing filters [3] was the best way to precisely re-create the original waveform so that it could be digitized with a 24-bit ADC, and the resulting digital recording could be encrypted for security.]
So, as far as I'm concerned, the evidence is sufficient to conclude that Nichols made 3M digital copies of Dan's multi-track and original stereo analog masters back in the early 80's, and that they were converted to 24-bit copies which are being used for LPs. Any evidence or arguments against this conclusion can probably be explained away. For example, the fact that they made a surround-sound version of Gaucho, supposedly from the analog multi-track, and then never got around to the other albums, might have just been intended as a cover story for the existence of a 3M copies, and perhaps they concluded that it wasn't worth the effort to make surround-sound versions of the rest (not even their latest albums were released as surround-sound versions, as far as I know). If they want to retain control over those recordings, it's their right, and I'm satisfied with the 1999 Aja CD-release played on my Topping D10s (an awesome-sounding cheap DAC) and the knowledge that if I were to shell out $500 or so for a decent belt-drive turntable/preamp with a decent pivoting tonearm [4], about $100 for a good record-cleaner, and $50 for the LP, I could hear Aja in its original splendor, or even better, because the original recordings will last forever, and LP-recording and playback continue to improve.
Notes
Revisions
2/24/22 - Revamped after finding the 2006 Sound on Sound article by Roger Nichols using 3M decks to copy analog multi-track masters, which eliminated the need for a lot of the circumstantial evidence and speculation in the previous revision.
2/25/22 - Corrected a few conceptual errors, added some insights, and smoothed out some rough spots.
2/28/22 - Revised Note 4.
[1] Nichols stored the safety masters (15 ips analog copies of the original stereo analog masters) in his personal archive, and never played them, except for Aja, once, to make the 1982 digital master with a non-oversampling 16/44.1 Sony deck with the notorious dry/smeared-sounding input filters. (Some tracks from this version of Aja can be heard on the Very Best Of Steely Dan CD, a compilation made from the 1981/82 digital recordings. The Greatest Hits CD was apparently made with Apogee AD-1000 20-bit ADCs from a good analog master, and sounds good, although the analog master was created in 1978 and it was apparently digitized in 1993 when the AD-1000 was introduced, so it's not as clean as the LP-copy I had in the mid-80's.) Nichols could have used the 44.1 kHz oversampling deck which JVC introduced in 1982, and which made great recordings, but Dan probably didn't want to let such good digital copies out into the wild. The 1985 digital masters were apparently made with a Sony 1610 with Apogee input filters from the same analog tapes, which by then had seriously deteriorated so that the high end is little but a dry sheen, except for Katy Lied (dBx-encoded) and Gaucho. The Gaucho digital master has a slightly high speed, so that the slow cuts sound rushed, and the Aja CD has a peaky high-end boost which sounds dirty and artificial. The MoFi Aja and Gaucho CDs (which I haven't heard) were made from the same digital masters, and according to Nichols, in an article in Metal Leg 18, Gaucho has the same speed error as the commercial release. However, MoFi might have used different EQ on Aja. MoFi got their own oversampling ADC in 1988 from Theta Digital. It used BB ADCs in parallel to sample the input at presumably 176 kHz, and Motorola DSP chips to implement the digital filter that reduced the sampling rate to 44.1 kHz.
[2] Nichols obviously broke out the digital signals so that he could duplicate digital recordings without converting them to analog. 3M probably sold accessories for this purpose, but Nichols, who was famous for creating recording-related gear, might have found a better approach, or at least a less expensive one. If the duplicates were played back on the original recorder, their errors would be canceled out.
[3] It was easier to design anti-aliasing filters with a linear phase characteristic for converters with a 50 kHz sampling rate than for those with a 44.1 kHz sampling rate. But in about September of 1985, Apogee introduced aftermarket filters with linear phase response for non-oversampling 44.1 kHz decks. Oversampling 20-bit studio-grade converters were introduced by Apogee and others in 1993, and 24-bit units were introduced in 1997.
[4] Linear-tracking tonearms are expensive gimmicks designed to relieve rich people of their excess money, which I realized after designing a cheap one and wondering why the design hadn't been used previously. Even if a linear-tracking design manages to reduce tracking error, it won't sound significantly better, and it will have higher lateral inertia, which causes problems when playing records with off-center holes and/or warps. (There are tangential pivoting tonearms which reduce tracking error by rotating the headshell, but there's no concensus among audiophiles that they sound any better, and their prices are astronomical.) I also prefer pivoting arms because they allow a phono preamp to be placed near the pivot to minimize the length of the tonearm wiring. With modern op-amps, excellent phono preamps can be built into turntables. Well-Tempered Lab turntables, which are based on original thinking, use pivoting arms, although long ones in some cases. Unfortunately, they are considerably more expensive than what I would want to spend on a TT.
