SH Spotlight CD & Digital flaws: Our technique for finding A/D & D/A converters for DCC Golds in 1992

Discussion in 'Music Corner' started by Steve Hoffman, Jun 6, 2008.

  1. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    Not in mastering. You could insert. It just, well, didn't sound totally wonderful.

    The real drag with the system is those turkey U-Matic 3/4" tapes. I doubt any of them still play.:shake:
     
  2. Plan9

    Plan9 Mastering Engineer

    Location:
    Toulouse, France
    We still used them in my video classes 3 or 4 years ago! :laugh:
     
  3. I had my first "OMG the SOUND" experiences when helping install the new SSL console here at UNC Asheville. Our teacher used the new patch bay to A/B between the Digidesign 96 I/O and the new Lynx Aurora unit, and I couldn't believe the difference. Going from the 96 to the Lynx, the bass tightened and cleared up, and a blanket came off the midrange; we all were completely blown away.

    It's equal parts fun and infuriating to hear such vast differences in converters.
     
  4. emmodad

    emmodad Forum Resident

    Location:
    bay area, ca
    and it's what's inside that counts....

    "The little 16/44.1, silver disk can kill giants. It just needs the right vitamins."


    it's fun reading this thread, bring back some (geek) converter design memories for me... and here's another tidbit wrt 16/44.1 performance if "fed the right vitamins"... relating today to 1984.

    in today's Great Debates about "which DACs sound best" or "why DAC X sounds different from DAC Y" there will often be discussion about different bit depth / resolution, different output sample rates, different DAC chips, current-output versus voltage-output DAC chips, different analog reconstruction filters, different analog output circuitry... ALL have an impact on resulting sound. IE it's not just one individual facet, but the overall D-to-A system and electronic signal path which contributes to whatever end quality you perceive from the DAC "box."

    In 1980 I was fortunate to start my career in the group at Bell Labs which designed some of the very earliest single-chip DSP integrated circuits (programmable digital signal processors). I got to do some very cool stuff then and later in the 80's which played off my audio engineering interests (anyone remember the ATT Studio Systems Digital Console DSP Engine?).

    Based from some of my work at Bell Labs, in 1984 I was hired by Roger Lagadec (merci vielmals Roger!) to work in Studer's "PCM Labor": the main Studer R&D in Switzerland which developed all of their early digital audio technology. Among other cool related projects (converter architectures, sample-rate conversion), one of my main early tasks was part of the system design for the DAC portion of Studer's 2-channed digital recorder.

    You may know that there are people even today who have high regard for audiophile DACs and CD players that use a revered "ancient" 1980s 16-bit DAC chip from Philips, the TDA1541 (ie see http://www.lampizator.eu/lampizator/TDA1541 corner/TDA1541.html). These actually had "real" performance which was more like IIRC 12-13 bits, due to the non-perfect linearity of their conversion process. (aside: you can look ie on the Analog Devices website for good explanations in papers by Bob Adams as to how linearity errors in DAC chips can impact sound quality.)

    Simply stated, in the perfect world a DAC will accept an input code and perfectly create an analog output at the appropriate level of voltage or current. However, things aren't perfect, and sometimes the created output is off just a small amount. This means that there are slight errors in the analog output; "integral" non-linearity errors are errors away from the overall perfect desired output level; "differential" non-linearity errors are errors in the desired perfect incremental sizes from one step to the next in the "staircase" of values which the DAC can put out.

    Well, Studer had done studies on the DAC chips available at the time, and concluded that these Philips parts were the best solution, but they had linearity errors. So what my team was doing was building a system to perform automated measurement on each individual DAC chip before it was used in a product, in order to determine the linearity errors for that particular chip across all of the 16-bit codes; we further created an error-correction system which basically made a table of those measured errors, calculated the inverse (opposite) amount; put a code for those opposite amounts into a table stored in PROMs (memory chips), and then the overall DAC circuit accounted for those "correction factors" associated with every code going into the DAC chip: thereby cancelling out the chip's linearity errors to a very successful degree. IIRC, we corrected the 1541 chips to be typically more like 15 bits of "real" distortion performance.

    Funny enough, Steve's first post in this thread is very similar to one key step we did back then in our testing: high-quality analog recording of transient events (rimshots, castanets, drum hits, and sorry Tom Scott, but I really never again want to hear your "Target" CD, it's burned in my brain...) with lots of reverb; and simultaneously through Studer's best A-to-D converters.... then compare decay signature of the analog version versus digital on playback through our DACs as we tweaked the system to do better linearity correction.

    Twenty-five years later, the converter chips which are at the heart of your DAC boxes generally use different internal architectures, but issues like this still have a lot of impact on perceived performance and sound. And there are some pretty clever system and circuit techniques being used to squeeze the best performance out of today's DAC chips.

    Do hope that was perhaps mildly entertaining for y'all.... :wave:
     
    head_unit and Gardo like this.
  5. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    Very interesting, thank you.

    I have a few players from the early days that sound amazing...
     
  6. GreenDrazi

    GreenDrazi Truth is beauty

    Location:
    Atlanta, GA
    I had one of those early model Philips CD players (1986) with the TDA1541A DAC. It was one of the base models without the crown DAC and became my first foray into modding my stereo system when I upgraded to the crown DAC as well as upgrading the op amps. I was sorry to see the unit die 10 years later and the Toshiba that replaced it was not nearly as good.
     
  7. dartira

    dartira rise and shine like a far out superstar

    I had almost exactly the same experience recently comparing a 96 I/O to a Prism Orpheus.
    The difference was amazing, baffling even. We were very reluctant (and sad) to return the Prism!
     
  8. I love stories like this! :righton:
     
  9. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    The struggles of making digital work. Amusing for you, endless struggle for us in the biz...:)
     
  10. Thanks emmodad - makes me happy that I just paid to have the Studer D730 CD player refurbished after 15 years of lovin'. :love:
     
  11. emmodad

    emmodad Forum Resident

    Location:
    bay area, ca
    amusing? nah, it wasn't amusing (except for some of the partying with my colleagues), it was actually PITA, getting best possible performance out of what Philips thought were "mature" early chips....temperature drift, arggh..... even with today's technologies, it's still endless struggle for others of us also in the biz....


    to this day, two of my all-time favorite players for sound via analog out were the very first Studer product (A725, used selected and corrected TDA 1540, was "really only 14 bit" but sounded oh-so-smooth); and the 1541-based A727 which was sublime over balanced outs. Do regret having sold that one when I moved back to the US... had a chance to hear a well-maintained one a few years ago and was reminded just how good it was for its day of twenty years ago (and today too!)
     
  12. e630940

    e630940 Well-Known Member

    Location:
    Canada
    The Studer B126/226 models were the first players to convince me that the silver disc was not the weak link. In a highly resolving system - these players made vivid low level musical info and micro dynamics that I had no clue existed on a CD. I still have the B126 on a shelf and would love to revive it one day with mods I see posted on the web. Brilliant.
     
  13. Robin L

    Robin L Musical Omnivore

    Location:
    Fresno, California
    I think I first heard the distortion/truncation of reverb patterns in 1981/82, when some of the first Digital Lps appeared. I was living in La Verne, just northwest of Claremont. Claremont had a Rhino Records with plenty of quality used vinyl. I had a fair number of Shaded Dogs and suchlike at the time. Columbia came out with a few Digital recordings of classical music around then, most famously the 1981 Glenn Gould re-recording of Bach's Goldberg Variations. The reverb pattern didn't seem to come to a real rest but evaporated before it should. Compared to the 1955 mono recording, this was a step down. Even worse, DGG's roll-out of Karajan's upteenth Die Zauberflote featured a bonus disc with the 1939 and 1980 recordings of the Overture to Die Zauberflote, both given a full Lp side with the Digital given a heavier cutting, using more of the available surface than the side taken from the 78 source. Guess which one had a plausible Hall Sound and which one didn't?

    For the record, my gear at the time consisted of a recent Dual 606 turntable with a low-cost Grado cartridge, a NAD 3020 and a pair of no-name 2-ways from a Pasadena Hi-End shop, 8" bottoms, soft dome 1" tweeters, smallish sealed enclosure, crossover consisting of a single cap. Pretty nice speakers as I recall.

    In any case, it was impossible not to hear the difference between good analog and early Digital. My first experience with CD was with the first CD mastering of Kind Of Blue and that did a worse job with ambient retrieval than the early Digital LPs. The generally awful sound of the Sony 501 that was my first a/d convertor didn't have to bother me for long as the Technics DA-10 that was my first DAT machine smoothed thinks out a little. Later, a t.c. electronics m-2000 struck me as better at ambient retrieval than some Apogee converters I rented. Starting around 2004, with my first exposure to SACD, things got noticeably better. I did a lot of recording between 1989 and 1999, there was real improvement in digital recording gear during those years. But right now I have a lot more faith in Digital recording gear than when I made a living using it.
     
  14. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    Some people still don't (can't, or won't) hear it. I hear it, and I work around it in mastering..
     
  15. Espen R

    Espen R Forum Resident

    Location:
    Norway
    A question about the A/D converter that Dennis Drake used to master the Mercury Living Presence catalog for CD in the early 90's:
    In the booklet is says: ...no equalization, filtering, compression or limiting was used. The actual analog-to-digital conversion device used, 128 times oversampling with properietary noise-shaping techniques.

    What kind of A/D converter was used? A Wadia...?
    The funny thing is that I'm very sure that the same A/D converter was used when Dennis Drake mastered Ella Fitzgerald "Clap hands" at Polygram in-house mastering studios in 1989. They share "the same sound". That sound is not 100% tonality correct, Imo. It has a very analoglike presentation, especially in the highs, tape hiss sounds very credibe. But the sound is a bit lightweight in the bass area. I doubt this is due to Mr Dennis Drake mastering. In my ears it has a "sound" to it, like all Super Bit Mapping CD's has a specific "sound" (but not the same).

    But all in all, I think this was a very good A/D from the late 80's, it sounded "analoglike", and that was the most important thing to me.
     
  16. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    Apogee, not Wadia. Slightly light on the upper bass, small midband suckout. Nice top...
     
    LeeS likes this.
  17. Espen R

    Espen R Forum Resident

    Location:
    Norway
    Ahh...of course, Apogee. Forgot that brand. Yes, your sound describsion of that AD mirrors mine.
     
    Steve Hoffman likes this.
  18. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    Good. We are in sync. Funny how certain A/D converters have a sonic signature. Most brains will tell you that's impossible, but it is not. It is reality.
     
  19. Gardo

    Gardo Forum Resident

    Location:
    Virginia
    Steve, and Barry too, what's the first big step in better digital audio: increasing bit depth from 16 to 24 bit, or increasing sampling rates to 88khz or above? For example, all other things being equal, are we likely to hear better sound from 44.1khz/24 bit recordings or 96khz/16 bit recordings? (I hope the question makes sense.)
     
  20. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    First big step? Eh, can't comment. Worm cans need to stay closed.
     
    OcdMan and MrRom92 like this.
  21. James_S888

    James_S888 Forum Resident

    CD is dead.
    The future is downloads. Pick your resolution.
    Decoding algorithms, pick your resolution, will also become increasingly a part of computer software. iTunes, whatever.
    It's all code folks.
     
    rxcory likes this.
  22. James_S888

    James_S888 Forum Resident

    And I hate to say it, in about 10 or 15 years, once most of all those tapes that were transferred at 44.1/16 to CD go back into various studios and go through at 192/24, LP will likely also be pretty much dead.
    Just my opinion....
    Except for those cases where the master tapes no longer exist or were incinerated... Chess masters for example...
    In these cases, the LP is so far ahead of the 44.1/16 transfers, the LP will live on :)
     
  23. Steve Hoffman

    Steve Hoffman Your Host Your Host Thread Starter

    I had a bump request for this. Here you go..
     
  24. quicksrt

    quicksrt Forum Resident

    Location:
    City of Angels
    But every way this topic is discussed seems to be wormy cans no? One brings up 24/96 and 24/192, or even 24/48 and you got fans of tin foil cookies inside clear plastic coatings screaming it's a fraud or there is a conspiracy! People blew their life savings on those 16/44 discs, before time (and technology) marched on and up.

    So I'll just say that pre-amps are another thing that matters to many when slicing and dicing to ones and zeros.
     
  25. head_unit

    head_unit Forum Resident

    Location:
    Los Angeles CA USA
    Yeah! We used to use Don Dorsey's "Ascent" on Telard to compare CD players, and that was exactly what we heard differently, the reverb tail.

    My big question is, how did you feel the analog tape compared to a direct feed into your speakers from the reverb setup?

    And have you ever messed with DSD recording? I've only heard comments from two folks who did true A/B comparisons:
    - John Eargle told me that he had compared both and while he had nothing against DSD he felt 94/24 was quite transparent (I didn't think to follow up asking just how transparent!) and, given that, the usability of multibit trumped the what I would call the pain-in-the-buttness of working with DSD at the time.
    - One of the Telarc guys whose name I don't remember offhand said they had set up complete 96/24 and DSD chains, and they felt the DSD chain had the edge...though neither was quite up to the live mic feed.
     

Share This Page