Topping D70 DAC Review and Measurements - Stream of Consciousness

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by purr1n, May 17, 2020.

  1. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    TBH, I haven't heard big differences between 4490 and 4493 DACs. I actually thought the Modi 3 was using the 4493 because of how good it sounded! Jason had to correct me.

    I've polled a few people who have played with these chipses. The word is that the 4497 is a bit "different" from the others and the AKM4499 is quite good, but at a crazy price. Could this by why I found the D70 to be less desirable, i.e. leaner, brighter, sigma-delta-ier? Or could the D70 could underperform because Topping only knows how to make cookie-cutter designs? The AKM4499 is $83.70 at Digikey. That's a fricking ripoff for a "quite good" sigma-delta part.

    The lower-end Focusrite Scarlett stuff is dull garbage. So much in fact that it's put a shit-stain on their brand in my eyes where I would be afraid to try anything else from them.

    Two things:

    That only tells me that their engineer only listened to the filters. There's not much, but still some wiggle room for designing around the AKM Velvet Sound DACs: power supply, output stage (even if a chip), and digital receiver. It's my understanding that there is actual room to improve ("secrets") around the cookie-cutter designs around the AKM DACs. Both RME and Schiit have implied this. I'll believe that Topping engineers listen to stuff when they talk about the different USB receivers or opamps that they've tried, or when they don't make something that under-performs for the price. Also, when an engineer actually listens to gear, they don't make DACs with a Cirrus chip one minute, ESS another, AKM next, and then shortly after, the TOTL SINAD AKM4499. It takes time to develop products by ear (and measurements).

    I don't feel bad for the Topping engineer. If you want to live by measurements (recall that one-third of their user manual is AP555 measurements), then die by them too. Anything else would be hypocrisy. If Topping backs themselves into the measures good (with a few steady state sine wave) but sounds meh for the price, that's their problem.
     
    Last edited: May 19, 2020
    Sonorus, Xecuter, Ksaurav402 and 9 others like this.
  2. Psalmanazar

    Psalmanazar Most improved member; A+

    Slaytanic Cliff Clavin
    Joined:
    Sep 27, 2015
    Likes Received:
    3,952
    Dislikes Received:
    121
    Trophy Points:
    113
    Focusrite has been muntzed from cost no object 80s and early 90s Neve circuits since they started ripping out the transformers. Every input used to have Lundahls and every out Carnhills. The current stuff gets less offensive as it goes up in price but everything they make now makes everything sounds like it’s playing inside a pillow rather than pillows in the drums. The production move to China was the end. Everything made afterwards is Chifi and sounds like a totally different product and brand from the U.K. made equivalents. The current ISAs are still bad. The 90s UK Focusrite gear was great. You can just tell from the fit and finish and casework being generic Chinese crap rather than killer steel or aluminum that can function as a medieval shield in jiffy.

    Here is their current Red 4 Pre. It looks like the similarly lame, made in China UAD Apollo:
    https://pro.focusrite.com/category/audio-interfaces/item/red-4pre

    1990s made in the U.K., full anodized all around, heavy aluminum chassis. it’s a tank.
    https://reverb.com/item/33745544-focusrite-red-1-4-channels-of-isa-style-mic-pre-s
     
    dark_energy likes this.
  3. yotacowboy

    yotacowboy McRibs Kind of Guy

    Friend
    Joined:
    Feb 23, 2016
    Likes Received:
    3,294
    Dislikes Received:
    11
    Trophy Points:
    93
    Location:
    NOVA
    Home Page:
    JFC. I hope folks realize that even with 500-1000 chip order pricing, that means that single chip accounts for $250 to $350 of the retail price, if the manufacturer keeps with normal markups. So, when you see a $900 AK4499 DAC you gotta KNOW every ounce of shit around that chip is value-engineered to be cheap cheap cheap.
     
    luckybaer, fraggler, alucart and 6 others like this.
  4. FlySweep

    FlySweep Friend

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    1,233
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    Portland, OR
    I want to hear more about the Motu AVB Ultralite MK4, @purr1n . Was dead set on a Bifrost 2 for a while but the Motu stuff's been on mind mind as perhaps a meaningful alternative.
     
  5. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    The below are via AES input, SE output. AverLAB.

    Topping D70
    1kHz 0dbFS
    upload_2020-5-22_9-25-40.png

    Topping D70
    1kHz -30dbFS
    upload_2020-5-22_9-28-5.png

    Topping D70
    1kHz -60dbFS
    upload_2020-5-22_9-29-47.png

    Topping D70
    1kHz -90dbFS
    upload_2020-5-22_9-30-52.png
     
  6. Vtory

    Vtory Illogical Spock

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    4,822
    Dislikes Received:
    37
    Trophy Points:
    113
    Location:
    Greater Philadelphia Area, PA
    For lazy people, I copied and pasted from Matrix X-Sabre Pro thread. 0dbFS and -30dbFS respectively.

    [​IMG]

    [​IMG]

    I'm assuming D70 also measured with 8192 window size.
     
    crenca likes this.
  7. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    A HA, EXPOSED!

    FYI, the 0dbFS or what "SINAD" often refers to is actually not very good in regards to the AKM4497 which is a -116db part, especially keeping in mind that there are x2 of them in the D70, where in theory it should be -122db.

    D70
    [​IMG]

    Modi 3 for comparison
    upload_2020-5-22_10-5-9.png

    Modi 3 and D70 overlaid
    GRN = Modi 3
    RED = D70
    upload_2020-5-22_10-5-42.png
    The Modi 3 results are slightly better, with the only difference a swap between the 5th and 7th harmonics.

    What's disturbing is that the Modi 3 uses the AKM4490, which is only a -112db part! Schiit is actually getting superior results with a far inferior chip (the 4490 is two levels down from the 4497), and only a single chip!


    Never mind my subjective impressions where I felt the D70 sounded cooler, more "digital", and less resolving than the Modi 3. Scientifically speaking, if you believe in the idea of 0dbFS SINAD being the major indicator of sound quality, you are much better off with Modi 3 at $99. It's not that the D70 is bad; however, bottom line is that that Schiit saves you from overpriced Chinese designed and Chinese made gear.
     
    Last edited: May 22, 2020
    edd, Philimon, FlySweep and 12 others like this.
  8. RedFuneral

    RedFuneral Almost "Made"

    Joined:
    Jun 8, 2016
    Likes Received:
    143
    Dislikes Received:
    2
    Trophy Points:
    43
    Location:
    CT, USA
    Sample size of one on both counts, but I doubt either brand walks too far from ready-made circuits(JDS OL: 4490, SMSL M300 MKII: 4497) I found the 4490 to suffer from brightness & glare. The SMSL variation of Topping's 4497 DAC is muted & veiled. Lacking in treble volume although there's still a slight upper-end annoyance in there which the slow filters mitigate. I find both DACs to carry a bit too much weight across the band which becomes incoherent topped with the faster treble. Incoherent really is the best word for these things: I hear them like there's a vaseline filter over the tonal properties but they're normal delta-sigma in the transients/speed. They can't decide if they want to be soft, exciting, or accurate so they try to do all three. AKM is trying to warm up the bass for the 'musical' crowd while keeping the treble high for the 'high-res mosquito fart' crowd.

    I'm wondering if my dislike for the Heresy was because I was being given a straight shot of Velvet Sound, I've been leaning more toward amps with a homogenizing property since taking in these AKM DACs, trying to find a more uniform sound, where I'll usually prefer the thinner faster sound as long as there's no fuckery afoot.
     
    FlySweep likes this.
  9. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    The below are via AES input, BAL output. AverLAB.

    Topping D70 BAL output
    1kHz 0dbFS
    upload_2020-5-22_10-54-4.png

    Comparison with SE output below. It looks like we are seeing more grass above compared to the SE output, but we really are not. It's just that the noise floor is lower. The measurements are more or less the same.

    Topping D70 output
    1kHz 0dbFS
    RED = BAL out
    GRN = SE out
    upload_2020-5-22_10-58-15.png

    Note that none of my measurements have the AC mains spikes that ASR has. This isn't because I'm better at this or have better gear. It's just the nature of the beast: measurements are hard! Believe it or not, I can make AC mains spikes appear by simply plugging into a different spot on the same power strip! Or sometimes it's the time of the day. The reason I'm mentioning this is that I've seen ASR go off on AC mains noise depending upon his mood or whether he likes or dislikes the manufacturer of the product. I'm simply saying AC mains noise, often from mini ground loops, is in many instances difficult to prevent or isolate.
     
    Last edited: May 22, 2020
  10. Roget

    Roget Rando

    Joined:
    Nov 17, 2018
    Likes Received:
    15
    Dislikes Received:
    31
    Trophy Points:
    3
    I studied the graphs. You are better at not getting AC noise. Can you share what you do?

    @purrin: Your conclusions are different from Amir's? Can you explain why?

    Conclusions
    The Topping D70 came dangerously close to unseating every other DAC. Its full output distortion is a few dB worse than the best so it didn't get there. But in all other respects, the D70 produces exceptional measured performance. Its crown jewel is exceptionally quiet noise floor that creates one of the best dynamic ranges I have measured. Use of AKM chip takes advantage of that to produces exceptionally clean performance where your volume control normally would be.
     
  11. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    Topping D70, 22kHz bandwidth
    THDN / SINAD from -100dbFS to 0dbFS with 1kHz steady state sine wave
    SE out
    upload_2020-5-22_11-35-40.png

    Note that distortion rises a slight bit as the signal gets hotter. Some DACs do this, others don't.
     
    Last edited: May 22, 2020
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    Topping D70, 44kHz bandwidth
    THDN / SINAD from 20-20kHz at 0dbFS with 1kHz steady state sine wave
    SE out
    upload_2020-5-22_11-51-0.png

    Topping D70, 44kHz bandwidth
    THDN / SINAD from 20-20kHz at -30dbFS with 1kHz steady state sine wave
    SE out
    upload_2020-5-22_11-52-57.png

    Topping D70, 44kHz bandwidth
    THDN / SINAD from 20-20kHz at -60dbFS with 1kHz steady state sine wave
    SE out
    upload_2020-5-22_11-55-41.png
     
    Last edited: May 22, 2020
  13. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,814
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Well, the digital filters seem to be just the default AKM ones, but I wonder why Topping doesn't focus on digital filters more if measurements are really everything. Also why ASR doesn't focus more on digital filter response. How much passband ripple is there, what's the phase response of the filter, etc. Why not use a super long filter with super sharp rolloff?
     
  14. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    5,808
    Dislikes Received:
    24
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    I consider myself one of the least knowledgeable on the forum as far as noise-shaping, the neural processes involved in auditory perception, filters, room acoustics, what difference diaphragm material makes, and basic engineering stuff etc go (my brain just freezes when you or @ultrabike talk about things sometimes, lol) but if I've learned anything it's that it is remarkably easy to talk a big game about being measurements-focused but another thing entirely to actually walk that talk.

    They're fronting, is what I mean to say.
     
    The Life likes this.
  15. RedFuneral

    RedFuneral Almost "Made"

    Joined:
    Jun 8, 2016
    Likes Received:
    143
    Dislikes Received:
    2
    Trophy Points:
    43
    Location:
    CT, USA
    My main complaint of ASR is the lack of impulse testing with DACs. I don't think they do phase response on speakers either. They overlook the entire time domain chasing perfect frequency/distortion charts. ASR can't talk about filters without also mentioning the purpose behind the various filters and calling attention to this omission.
     
    Azteca, Josh83, Serious and 1 other person like this.
  16. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    58,899
    Dislikes Received:
    101
    Trophy Points:
    113
    Location:
    Antarctica
    With the Topping D70, the best filter in terms of frequency domain is Mode 3, which I believe is the default??? That should be all that matters if your objective world only lives in the frequency domain.

    We wouldn't want to confusion people with ever more complex measurements. Hard to line up in an Excel bar graph like FPS for graphics cards or benchmarks for CPU. People like easy answers.
     
    Grattle, earnmyturns and RedFuneral like this.
  17. crenca

    crenca Friend

    Friend
    Joined:
    May 26, 2017
    Likes Received:
    562
    Dislikes Received:
    5
    Trophy Points:
    93
    Location:
    Southern New Mexico
    This. Yet, thinking about this from a positive point of view: in a world where most audiophiles believe that the "stair step" analogy accurately describes waveform reproduction from digital sampling, how to begin to move past very simplistic descriptions and bar graphs?

    @Serious, for ASR (or anyone else) to go down this road they would have to an honest grasp of what filter does and the inter-relationship between filter design and "ringing", IMD, phase, and how all that relates to what you hear. I don't get the sense that they are on that level...
     
    Azteca and RedFuneral like this.
  18. Superexchanger

    Superexchanger Friend

    Friend
    Joined:
    Jul 11, 2019
    Likes Received:
    999
    Dislikes Received:
    1
    Trophy Points:
    93
    Location:
    Flavortown, USA
    SINAD O'Connor belting out "NOTHING COMPARES 2 U"

    Just starting my Memorial day weekend drinking a little early, folk.s
     
  19. bixby

    bixby Friend

    Friend
    Joined:
    Sep 26, 2015
    Likes Received:
    2,409
    Dislikes Received:
    7
    Trophy Points:
    113
    It would be interesting to hear how the Modi 3 would sound without the slow, rolled Velvet filter that most seem to select when using the 4490 in their design.

    I use one of the 5 built in filters called "traditional" on my 4490 based JDS and like it quite a bit more than the velvetness of slow and rolled.
     
    crenca likes this.
  20. crenca

    crenca Friend

    Friend
    Joined:
    May 26, 2017
    Likes Received:
    562
    Dislikes Received:
    5
    Trophy Points:
    93
    Location:
    Southern New Mexico
    I have spent little time with the 4490 card that I ordered in with this Asgard (most of it on Youtube :p), but I want to say that it is a standard linear phase sharp roll off, which I think is the default...
     

Share This Page