Can anyone explain why 14-bit DAC sounds better than 16?

Discussion in 'Audio Hardware' started by JimmyJet, Sep 16, 2015.

Thread Status:
Not open for further replies.
  1. JimmyJet

    JimmyJet Well-Known Member Thread Starter

    Location:
    Gig Harbor, WA
    My subject here is my 29 year old Magnavox FD-2041 which has dual TDA 1540 DAC's that are 14-bit and the unit uses 4x oversampling. It also has the "swing arm" laser. It was made in Belgium in 1986.

    I did some research on the net and all I could come up with is that Phillips didn't like the sound of the 16-bit DAC's of the day and went with dual 14-bit chips and used 4x oversampling to make up the missing bits. Not sure if that is correct, but that is all I could find on the net. Also, the swing arm laser is praised as being superior.

    Over the years, I've had many different Japanese brands because I switched to carousel players. However, 10 years ago, I got back into vinyl big time and haven't played any cd's since. But recently, while cleaning out some storage, I came across the Magnavox FD-2041 and fired it up and it played flawlessly and it sounded so much better than I remember my later model CD players. It has a very warm and organic sound -- "almost" as good as vinyl ;)

    What I'm looking for is an explanation as to "why" these older 14-bit chips sound better than the 16-bit and higher sets. (Also, as a bonus, if someone could explain why the "swing arm" lasers are supposed to better, that would be welcome too!)

    Thanks!
     
  2. Archimago

    Archimago Forum Resident

    Remember that back in the 1980's, *true* 16-bit dynamic range was not within the grasp of the technology at that time, so even if a CD player claims 16-bit chips, it would be measurably worse.

    14-bit chips therefore would really not have sounded any worse providing >80dB dynamic range. And a good 4x oversampling digital filter would sound just fine.

    These days, if you take a 16-bit CD piece of music, did a -12dB reduction to knock off the last 2 bits, then reconstituted the 2 bits by +12dB, then upsampled to 176kHz (4x)... Played back on a modern DAC at 16/176, I'm sure it sounds pretty darn good :).

    As far as I'm aware the TDA 1540 is a mono part, so would have needed 2x for stereo. Possible that 2x mono could sound better than a single 16-bit DAC chip (if this was what they were comparing to). Also, lets not forget that Philips actually wanted the CD to be 14-bit resolution. Thankfully Sony insisted that it be 16-bits. So there could have been some bias there rather than technical rationale...
     
    JimmyJet likes this.
  3. Funky54

    Funky54 Coat Hangers do not sound good

    Everyone gets caught up with DAC's being what can make or break cd audio quality. It does make a large contribution, but a bigger portion of the sound is the synergy of the power supply, transport and DAC combined.
     
    Tim 2, lv70smusic, SamS and 1 other person like this.
  4. Ham Sandwich

    Ham Sandwich Senior Member

    Location:
    Sherwood, OR, USA
    The TDA1540 is a multi-bit DAC. More info here: Evolution of DAC & digital filter
    The linked article explains some of what may be going on with that old chip, and also touches on some of the 14-bit vs. 16-bit performance.
     
  5. Dino

    Dino Forum Resident

    Location:
    Kansas City - USA
    I have no idea why. Given that, I have been somewhat interested in the subject.

    I remember reading about 14bit vs. 16bit DACs in The Absolute Sound magazine back when CD players were pretty new. The reviewers definetly had a preference for the sound of the 14bit. Meridian, the first or one of the first, "Hi End" CD player manufactures had based their first two (at least) players on 14bit machines. They were very well reviewed. At the time it was common knowledge at TAS anyway, that the Phillips/Magnavox 14bit based CD players sounded more musical.

    Flash forward to the last year or so. I read a response to a reader's inquiry in one of the UK audio magazines where one of the staff was talking about how the early 14bit CD players had a special quality to the high frequencies. He likened it to SACD sound. In the context of the response, it appeared that the writer found the highs musical in the way that he found the highs on SACDs musical. (I'm not sure what he meant but it was meant as a complement.)

    A few months after that, I read something similar in another magazine by a writer that was a member of the staff - an American magazine this time.

    I was surprised that after all of these decades, some found the 14bit DACs still competitive and in ways preferable to modern DACs.

    Now I need to read Ham Sandwich's link.
     
    kevinsinnott and JimmyJet like this.
  6. JimmyJet

    JimmyJet Well-Known Member Thread Starter

    Location:
    Gig Harbor, WA
    Hey, thanks guys for all of the informative responses -- very nice link Ham Sandwich; and yes, Archimago, the specs for my FD-2041 list "dual" TDA 1540's, so I see that as a plus having two mono sets ;). Boy, I'm glad I hung onto this having no idea back in 1986 when I purchased it it would be a sleeper... And, I'm also feeling very old: in 1986, I was 26 and remember telling my fiance, "when CD players get down below $150, I'll spring for one." (We were struggling college students). Then one of the discount stores at the time, called Evans, advertised in the Sunday paper circulars of the Washington Post, the Magnavox FD-2041 as their featured sale item for that week at only $139! And of course, they printed those magic words, "in limited quantities" -- I remember hurrying up my fiance to get ready as I couldn't get out the door fast enough and put my pedal to the metal... Then, with my new baby in arms, we had to go out and buy our first cd's; she picked one and I picked one, that blew our budget as back then, cd's were $16.99 (at least the titles we wanted...) -- good times: broke but life was simpler and we were happier.

    Fast forward almost 30 years and now listening in comparison to my vinyl with a MC cart, I would say the Mvox 2041 sounds just as warm, but what I notice is the dynamics seem more compressed - you know, a little boxy... but damn good for digital and I'm enjoying listening to my catalog of cd's that I haven't replaced on vinyl or mp3's. Yes, I know that if you buy an audiophile cd player these days, 2k and up, outboard DAC's , etc. they will sound better than my 14-bit player, but for now, I'm okay with the poor man's version of an audiophile cd player ;)

    PS: Last night I came across an article that said a company called California Audio Labs produced an audiophile cd player costing $2k back in 1986, and under the covers was a Mvox 2041: they threw in some tubes in the output stage and dressed it up -- the name of the article was "Mutton In Wolf's Clothing" LOL! Check it out:

    http://www.stereophile.com/asweseeit/mutton_in_wolfs_clothing/#RkCepIuIckCIws52.97

    Cheers!
     
    kevinsinnott and Dino like this.
  7. head_unit

    head_unit Senior Member

    Location:
    Los Angeles CA USA
    Once upon a time, a magazine (Audio?) tested a similar Magnavox CDB-650. It was the only test I ever saw where it could reproduce a "staircase test" of the bits getting bigger very accurately. Later tests, especially 1-bit, did not look anywhere near as clean.

    I think partly the 14-bit DAC they used had a very good 14 bits compared to lesser 16-bit parts, which maybe couldn't even perform at 14 bits resolution. THEN, the analog stages make a huge difference to the sound, maybe more than the DAC stage. AND, the power supplies make a huge difference, because if they are poor, then for example a servo struggling to read a scratched disc can "pump" the power supplies and contaminate the analog audio if the power supplies are not robust.

    I worked at one of the largest sound companies, and one day after work we compared by -650 to a Phase Linear. Unmistakably better. Especially a Telarc track, I think Don Dorsey's "Ascent" which has an explosion, the reverb tail from the explosion was noticeably better; the Phase Linear sounded corrupted.
     
    JimmyJet and Dino like this.
  8. KT88

    KT88 Senior Member

    I would have to say that it doesn't. I don't think it is that important whether it was a pair of 14 bit mono or a 16 bit stereo DAC chip back then. They were all pretty lame by today's standards. Very few DAC chips were even available and what made a bigger difference between players was build quality, power supplies, and analog output stages. the Cal Audio Labs players had nice output stages. It is essentially like tagging a preamp along on the output of the DAC, so the better the analog performance, the better the sound. Isolating of power supplies is also important, as mentioned earlier here.
    -Bill
     
    Tim 2 likes this.
  9. lv70smusic

    lv70smusic Senior Member

    Location:
    San Francisco, CA
    I agree 100%. I'm amazed by the number of posts I see where people act as if the only thing that affects sound quality in a digital device is the DAC.
     
  10. JimmyJet

    JimmyJet Well-Known Member Thread Starter

    Location:
    Gig Harbor, WA
    Thanks for the additional comments - learning a lot here. Can anyone answer the 2nd part of my question as to why the "swing arm" lasers were considered to be superior to, well, those that didn't "swing" ;)
     
  11. JimmyJet

    JimmyJet Well-Known Member Thread Starter

    Location:
    Gig Harbor, WA
    Very cool! Got me interested in that "staircase test"... I'll poke around on the net to find out more -- thanks for that info!
     
  12. dmckean

    dmckean Forum Resident

    Location:
    San Diego, CA, USA
    I think they're liked because they tracked well, were reliable and repairable.
     
  13. Black Elk

    Black Elk Music Lover

    Location:
    Bay Area, U.S.A.
    Firstly, the early Philips/Magnavox/Marantz players were NOT 14-bit, they used noise-shaped, 4x oversampled 14-bit, which means they were 16-bit! There was no truncation of the two LSBs. You can read about the process here:

    https://books.google.com/books?id=C...on art of digital audio noise-shaping&f=false

    The Philips system is shown in Fig. 4-53 on Pg. 262.

    Philips did not opt for 14-bit converters because they preferred them, they had tooled up for 14-bit, only for Sony to insist on 16-bit in the Red Book, so their engineers were forced into going the noise-shaped, oversampled route in order to get 16-bit performance. (The shift from 11.5 cms to 12 cms for the disc also came about late thanks to Sony.)

    Why were they liked back in the day? A number of factors. Firstly, the 16-bit converters of the time were not very accurate, and had poor linearity. Secondly, Sony's engineers seemed to have a preference for classical 'brickwall' reconstruction filters which introduce appreciable phase distortion (let's ignore any ripple in the passband response). Oversampling relaxes the reconstruction filter requirements, and allows more gentle (linear phase) filtering. On top of that, they used different optical pick-ups, different digital filters, different analog stages, different clocks, different power supplies, etc., all of which (cumulatively) could result in different/more pleasing sound.

    Why are they still liked today? Fetishism! :) They are completely outperformed (in all metrics) by sigma-delta designs.

    With regard to the optical pick-ups, it all comes down to one beam vs. three beam designs. If you opted for three beam, you were forced to use a radial motion, whereas you could use an arc with one beam. I'm sure the designers at the time felt that there were pros and cons to both principles, but, in practice, it does not seem like one system out-performed the other, at least according to these links:

    http://home.mira.net/~gnb/mac-cdis/cd5.html

    http://www.repairfaq.org/sam/cdfaq.htm#cdcdopop

    http://www.repairfaq.org/sam/cdfaq.htm#cdopdeck

    Of course, the early Philips CDM optical blocks were all robust metal designs, which may have had more to do with their performance than the way the pick-up moved. For more info, see:

    http://www.dutchaudioclassics.nl/?strPage=Info&strBrand=Various&strType=PhilipsCDM

    Compare them to the flimsy plastic designs of today (which may well be just as accurate, or even more so, but don't give you that 'bulletproof' reassurance).
     
    Tim 2, Mal and Dino like this.
  14. head_unit

    head_unit Senior Member

    Location:
    Los Angeles CA USA
    That sounds like a piece I read which was saying our digital TV was overkill, that we could have used much less bandwidth if we had a system predicated on using oversampling as part of the reconstruction.

    And the more gradual filtering (plus, I think, the quality of the circuits and parts Philips/Magnavox used) probably helps.

    As for sigma-delta, both those and top-shelf multi bits have greatly improved in terms of signal-to-noise and linearity…sometimes. I wonder, at the less-than-high end, if we are really progressing or if stuff just gets cheaper. I had a coworker testing FM tuners at a large audio company, and due to cost each one was not-quite-quite as good as the previous. I don't ever see any tests of low-cost devices like iPhones, or even receivers (which Stereophiles' John Atkinson told me "we're STEREOphile" and receivers aren't stereo), and I question whether the average accessible performance is really improving.

    I'd also observe that sigma-delta systems can be less time-stable for steady state tones. There was an interesting SACD vs. 24/96 demo of this at one of the surround sound conferences in L.A. (One correspondent's observation about that, and the much higher ultrasonic noise vs multi bit was "If SACD measures so bad, why does it sound so good?" (and for "SACD" maybe we could substitute "LP" heh heh)).

    So we should always remember that the "metrics" we measure are probably only tiny windows into what we hear-which differs for each person-and windows whose relevance we don't really know.
     
  15. Funky54

    Funky54 Coat Hangers do not sound good

    A while back I posted that a group got together where I work and we just played around with different speakers, tables, subs, DAC's, players, amps, preamps.... Arcom, Classe, Bryston, Sonus Faber, Monitor Audio, Wilson Audio, B&W, magnapanes, Rega, Thorens, Audio Research.... Tube, Solid State, treatments you name it. One thing that was obvious, the physical CD sounded better than the digital consistently. We used Tidal and HDTracks and found versions of all kinds of music that we could duplicate. It really didn't matter if it was thumb drive, Sonus, or streaming through high end DAC units. The physical CD at 16 bit sounded better to all of us with each example. The owner of my company previously had shared that with me when he was comparing. In the room were a number of guys who have been in the industry for a long time. A studio engineer that worked with some of the biggest names in the industry, and one super techy high end programmer/ engineer who only loves computers and could care less about audio or this hobby. He has made fun of us audio nerds for decades. He heard it too.

    Go figure that with math and calculators. The noise floor in my opinion was a little higher with CD but so was the detail and openness. All said and done though.. I stil prefer vinyl.
     
    The Beave and JimmyJet like this.
  16. JimmyJet

    JimmyJet Well-Known Member Thread Starter

    Location:
    Gig Harbor, WA
    Yes, you are correct, I actually knew that they were making up for missing two bits to take it from 14 to 16 with the oversampling - I guess I didn't word my subject title correctly as I was questioning, why to my ears, the physical 14-bit TDA 1540's along with the... 4x oversampling and whole process Phillips used to bring it up to an effective 16-bits, sounded better than my Sony "true" 16-bit chip sets. But thanks for pointing that out as it's good to know - I actually read about that
    noise-shaping algorithm that got 16-bit resolution out of the physical 14-bit DACs in a magazine in the mid 80's called Digital Audio and CD Review. Below is the datasheet for the 1540:
    http://www.datasheetlib.com/datasheet/314229/tda1540_signetics.html

    Cheers!
     
  17. crooner

    crooner Tube Marantzed

    The secret is the 4X oversampling digital filter. No one else had this back when CD was introduced. It pushes the aliasing artifacts way beyond the audio band where a more benign analog filter can be used.

    Competing "16 bit" systems used a single DAC multiplexed or shared between channels, with a resulting delay. The lack of oversampling meant that a sharp "brickwall" filter had to be used to get rid of quantization noise. The most common one being the Chebyshev, which introduced nasty phase artifacts in the audible range. The Sony CDP-101 used this filter.

    That's the reason why early Philips/Magnavox/Marantz players sound so good.

    I am a huge fan of the Philips CD100/Magnavox FD1000/Marantz CD-63 (original).

    Today UPS delivered my third FD1000. A super early unit made in March 1983 in the original box with manual, accessories and sampler disc.
     
    Last edited: Sep 19, 2015
Thread Status:
Not open for further replies.

Share This Page

molar-endocrine