DAC-OFF: Soncoz SGD1 vs. Geshelli JNOG

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by purr1n, May 5, 2021.

  1. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    It's just parroting the same BS story over and over. The folks with $15k-$30k to spare for an AP buy it for one reason - to establish authority, usually false authority, instead of using it as a tool for discovery, learning, and real science. We've already seen how Amir fucked up on this early measurements and needed Jude (presumably with AP's assistance) to school him.

    I'm not saying the AP stuff isn't great. It's fantastic actually. The UI, ease-of-use, and speed of the AP stuff is amazing. However, it's like what I've always said, it's the man, not the machine. To be honest, I did not expect the results to be so similar as seen above. It always seems to come to this: same guy and exact same piece of gear = same or very similar result.

    Conclusion: the SGD1 DACs that ASR got were different than the SGD1 I have in my hands right now.
     
    • Like Like x 6
    • Epic Epic x 2
    • List
  2. Roget

    Roget Acquaintance

    Joined:
    Nov 17, 2018
    Likes Received:
    39
    Trophy Points:
    18
    Fair enough Mr Purrin. I can't see the lettering on the DAC chips in the photos. Can you take a better picture? Is this chip ES9038Q2M? Is this cause of bad measurements?
     
  3. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    762
    Trophy Points:
    93
    Location:
    England
    To add to this, I had a demo unit here sent by a friend, whilst I don't have an ap analyser either (i'm actually using an ADI-2 pro ADC as it's to my knowledge the best measuring standalone ADC which also supports ranging input sensitivity), I was unable to get anywhere close to the same measurements as ASR, and saw similar to what is displayed here, with much higher third order harmonics.
    Of course I never expected to get as low as ASR as you'd need a notch filter to achieve that (which I don't currently have but if anyone here would be happy to build/sell me a balanced analog 1khz notch filter shoot me a msg), but the fact that I was still getting better measurements with some other dacs I had on hand had me curious.

    I don't have these measurements saved unfortunately, in honesty I sort of assumed I must have just done something wrong and so didn't save or post them anywhere. But i'll see if I can get that unit back from the friend who loaned it to me at somepoint and re-check it.

    I do however have the LA-QXD1 which I purchased myself to review so i'll see what I can get from that
     
  4. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    762
    Trophy Points:
    93
    Location:
    England
    Also a suggestion (stolen shamelessly from TalbotHound :p): perhaps the 'IMD Hump' fix they applied has resulted in poorer THD performance.
    I believe Gustard mentioned somewhere that this was the reason they'd not implemented an IMD hump fix in some of their products? Because they could get better THD by not fixing it
     
  5. TheTalbotHound

    TheTalbotHound Rando

    Joined:
    Mar 17, 2018
    Likes Received:
    2
    Trophy Points:
    3
    Location:
    Earth
    This might be something to do with the revision that removed the IMD hump. All the previous measurements had been done with a version that had the IMD hump.
     
  6. Baten

    Baten Friend

    Friend
    Joined:
    Mar 18, 2018
    Likes Received:
    1,004
    Trophy Points:
    93
    Location:
    EU
    The two DAC chips are the rectangular chips right above the LPF circuit. There's a third rectangular ESS chip which is a regulator made specifically for these. And AFAIK, the 9038Q2M aren't faked. Good manufacturers however, set thresholds for underperforming chips. 124dB dynamic range is typical, can vary up to or below <=120dB in the crappiest ones. These are mobile grade after all, not PRO. The 9038PRO in turn suffer from thermal problems varying chip to chip.... ESS is a crapshoot lol.
     
  7. TheTalbotHound

    TheTalbotHound Rando

    Joined:
    Mar 17, 2018
    Likes Received:
    2
    Trophy Points:
    3
    Location:
    Earth
    Yeah i have photos from inside mine of all the chips and it is definitely a 9038Q2M.
     
  8. GoldenOne

    GoldenOne Friend

    Friend
    Joined:
    Jul 17, 2020
    Likes Received:
    762
    Trophy Points:
    93
    Location:
    England
    I can't imagine they'd have faked the actual chips themselves. Especially given the other aspects of the design that likely have added significant cost to the BOM like the dual transformers etc.
    It's most likely either QC variation/failure to check.

    Or something that has changed recently, the only thing I can think of being the IMD hump as talbothound mentioned.
    The capacitor values on the output stage did change
     
  9. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Is there a plot of the IMD vs level for the revision that fixed this?

    Even if IMD hump was fixed, such a significant rise in THD will probably do more to wreck the spectrum, e.g IMD from harmonics, more "grass" , et. al.
     
  10. TheTalbotHound

    TheTalbotHound Rando

    Joined:
    Mar 17, 2018
    Likes Received:
    2
    Trophy Points:
    3
    Location:
    Earth
  11. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Too bad with the revision, l7 only posted the IMD sweep and not the other more basic measurements. The rule or guideline is never assume.

    Anyway, more investigation...

    Soncoz SGD1
    No Input Signal, XLR out
    upload_2021-5-8_10-28-11.png

    I don't know where that mysterious ~850Hz signal is coming from. (Note: also seen the AP measurement from above).
     
  12. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Soncoz SGD1
    SMPTE IMD
    -60dbFS, -35dbFS, and -20dbFS. Note: -35dbFS is near the peak of the ESS IMD hump
    upload_2021-5-8_11-2-37.png

    Inconclusive. Too much grass from 20Hz multiples - don't know where the 20Hz harmonics are coming from. 60Hz harmonics would be expected though.
     
    Last edited: May 8, 2021
  13. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Let's check what happens when only a single 60Hz is played back

    Soncoz SGD1
    60Hz -20dbFS
    upload_2021-5-8_11-9-2.png

    Hmm 20Hz mulitple goes away. The strange thing is that 20Hz isn't related to any IME products from the 4:1 60Hz/7000Hz test. Moral of the story: IME sweeps can only tell us so much.
     
  14. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Let's take a look at the filters with 44.1kHz and using USB input. Note the below is not a exact test of FR. We are simply trying to ascertain the general filter characteristics on FR.

    Geshelli JNOG
    Supplied filter (only one not changeable)
    White Noise
    32 FFT average
    upload_2021-5-8_12-18-45.png

    Soncoz SGD1
    APOD, BRIC, FAM, FAL Filters
    White Noise
    32 FFT average
    upload_2021-5-8_11-27-56.png

    Soncoz SGD1
    HYER, SLL, SLM Filters
    White Noise
    32 FFT average
    upload_2021-5-8_11-30-52.png

    None of the Soncoz filters seem to suppress ultrasonics very well. The JNOG does a fine job.
     
    Last edited: May 8, 2021
  15. crenca

    crenca Friend

    Friend
    Joined:
    May 26, 2017
    Likes Received:
    3,096
    Trophy Points:
    113
    Location:
    Southern New Mexico
    Is it safe to assume that all these filters are built in to this ESS ES038Q2M chip - even if they have been (re)named by Soncoz?
     
  16. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    ES9038Q2M datasheet lists these:

    Apodizing, hybrid, brick wall, linear phase fast, linear phase slow, minimum phase fast, minimum slow. I need to check the above results again since mine do look kind of funky and unexpected. Or maybe I won't bother - too time consuming.
     
  17. crenca

    crenca Friend

    Friend
    Joined:
    May 26, 2017
    Likes Received:
    3,096
    Trophy Points:
    113
    Location:
    Southern New Mexico
    I would not bother as the chances they programed there own are pretty low IMO. Its just more cookie cutter stuff...
     
  18. Raimei Templar

    Raimei Templar Friend

    Friend
    Joined:
    Dec 8, 2019
    Likes Received:
    611
    Trophy Points:
    93
    Location:
    Osaka

    There is either something funky happening with that particular unit or Soncoz screwed something up as those filters are not supposed to have that kind of response. ESS filters are generally sharper than their AKM equivalents. Jnog uses the short delay sharp linear phase I believe, the ess fast linear phase should show earlier and slightly greater ultrasonic suppression.

    Might partially explain why it doesn't sound particularly good if it is a botched implementation . That being said beyond figuring out rather this unit is defective which would require a 2nd unit to confirm there is not a huge value in doing a load of measurements again.

    I would not be surprised if the implementation is off as ESS documentation is notoriously bad and it takes a lot of effort to get those chips sound right.
     
  19. Toutoui

    Toutoui Rando

    Joined:
    May 12, 2021
    Likes Received:
    5
    Trophy Points:
    3
    Location:
    Bois-Colombes, France
    Hi,

    Just wanted to share my experience with the the new JNOG… Listening impressions is all I can share, as I have no technical notions.

    I have owned three DACs in my life :
    • a french Atoll DAC100 (entry level with Burr Brown) : I couldn’t listen to this without lowering the volume after a few seconds… Sound was like ’sticking’ to the air, giving an unpleasant ‘saturated’ effect, as if the DAC didn’t know how to move from a sound to another :-(
    • my actual Metrum Acoustic Musette : quite the opposite ! Natural flow, much more details, natural staging (not « in the face »). With a good power cable, that Musette remain with me for five years, giving constant satisfaction :)
    Wy did I order the JNOG ? Mainly by curiosity, as I see so many discussions about R2R vs Sigma-Delta DACs. How a good D/S DAC sounds ? That what I wanted to know… It’s probably unfair to compare both DACs, as one is more than 3x the price of the other, but I can’t resist the temptation :)

    That second version of the J2 (let’s call it J2 v2 ?) il running since 50 hours, linked to my old iMac.

    Right out from the box, my first impression was : the VOICES ! How good they are, how immaculate, how lisible ! Across favorite songs I could suddenly understand many words that didn’t reach my mind before… WHAOOO ! But very soon after that, I was struck by how thin the treble was… Flutes, cymbals, violins, trumpets sounded so slim… no texture at all ! And goodness, where the lows had gone ? Was it possible that only the ‘mids’ sounded so good ?!

    I plugged an Intona Hi-speed Isolator between Mac and J2 v2 and let it run a few hours before coming back. Everything appeared more balanced yet, but still that thinness in the highs, and something a bit blurred in the lows. Next to the Musette, J2 seems to ‘fight’ with the lows. But anyway, it’a easy to hear how refined the J2 v2 sounds : it’s all about separation, delineation, making every musical line sound so clear, so intelligible… Musette is a raw force next to it. But maybe Musette’s ‘full-bodied’ sound makes the listening experience more… natural ? More direct ? More communicative ?

    As I write these lines, J2 runs since 50 hours and I must say I enjoy it more and more… Is it possible some components inside need to warm up ? Texture seems to come, lows are much better… Or maybe my ears just get used to it ?

    Note that J2 is fed with its provided little supply… Would a 12v LPS enhance the experience ?

    Hope these impressions can help ;-)
     
  20. purr1n

    purr1n On vacation

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    79,362
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    It was my experience that the JNOG sounded better after a few hours of being left on. I generally leave it on now. The JNOG sounds more tonally lean with mushier transients and more digititus when cold.

    With some DACs, mostly DS DACs, it makes no difference. In a few others, leaving it on longer made it sound worse.

    Maybe. Maybe not. I think Geshelli uses good opamps with 100db+ PSRR, so power supply may not make a huge difference. I'm sure the design of the PS is also well done. Otherwise the JNOG would not measure so well.
     

Share This Page