Let's test the output of our DACs with RightMark Audio Analyzer.

Discussion in 'Audio Hardware' started by Robert C, May 12, 2016.

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

    darkmass Forum Resident

    I received a PM that has given me additional information concerning the ADI-2 Pro and additional information for one other matter.

    I have the author's permission to post the following extraction from the PM...

    The ripple is part of the used ADC chip, the AK557x family. There is no way around it than using a different chip. In an interview in a German magazine RME stated that they did not like the ripple (like anyone they would have preferred less ripple as found with other ADCs), but still used the chip for the Pro as it was so 'damn good in THD and SNR - and allows to convert to DSD'. Meanwhile the AK557x chips are also used in other, partly very expensive units, for example Merging, and show the exact same ripple there.

    Then this:

    "I was not able to use strictly the ADI-2 Pro to stand alone, via balanced analog, convert DSD to analog at the same time the ADI-2 Pro was converting the analog back to DSD. Nor could I use the ADI-2 Pro to convert PCM to analog at the same time it was converting the analog to DSD."

    The manual includes a note about DSD operation, it is basically independent on AD and DA. But you won't find a software that can record DSD and does playback at the same time as PCM, and the Windows ASIO driver also can not do that. The unit should be in Stereo/2-channel mode, then you can use the DAC with digital to DSD conversion on output 1/2 and digital to PCM conversion at output 3/4, simultaneously. This mode comes into effect when you change the source of output 3/4 in the display's menu to Analog In or Digital In with a PCM signal as source. At least that's what I remember, maybe I have to try that again...but I never need that when I use the unit...

    And allow me to add the final point from the PM, this one concerning the quandary that was initiated by post #75...

    There was a digital loopback active within the Realtek driver/soundchip [for the Fiio DAC], from digital out to record in. The analog Line input was not measured at all.
     
    FledTetcher and Robert C like this.
  2. acdc7369

    acdc7369 Forum Resident

    Location:
    United States
    Thanks for doing that. I also use the Xonar U7...your test proves that it performs better than the published specs indicate. that frequency response is ruler flat
     
  3. back2vinyl

    back2vinyl Forum Resident

    Location:
    London, UK
    Very sorry, darkmass, I've been offline for a few days otherwise I would have long since thanked you for your extremely kind words. Quite uncalled for in my opinion since I learn far more from you than you ever have from me but moving swiftly back on topic:

    First, on the audibility or otherwise of ringing, it was interesting to hear of your experiences with iZotope. Since they don't seem to have answered the question, or not decisively, I wondered whether you'd ever experimented with the Lacinato ABX/Shootouter? It's a little app that provides a way of blind testing yourself on your PC. Although I haven't used it in ages, I keep it on my desktop and I remember having had lots of fun with it in the past, testing for the audibility (or otherwise) of various differences - and often surprising myself. I don't think it's off topic mentioning this here because some might enjoy using it to test for the audibility of differences picked up by RMA.

    I don't mean to get into any subjectivist/objective debate - I only use the ABX/Shootouter for my own entertainment and I'm not trying to foist it on anyone else.

    Now, turning to your post quoting your mystery PM correspondent - how absolutely fascinating! So there we have the explanation for the ripple, and why we find it with the ADI-2 Pro and not so much with other A/D converters. Regardless of the AK557x chip's superior THD and SNR characteristics, I wonder if, more than anything, this is the price one pays for getting an analogue to DSD converter, which I don't think is very common elsewhere (though quite common the other way round). I seem to recall that you had already nailed the ripple primarily to the ADC so you were right and it all fits.

    Having said that, I still maintain that the ripple is purely cosmetic - I don't need the Lacinato ABX/Shootouter to tell me it's inaudible, though who knows, it might reveal the chip's superior THD and SNR characteristics!
     
    FledTetcher likes this.
  4. darkmass

    darkmass Forum Resident

    No, I have not investigated, nor have I even known about, the Lacinato ABX/Shootouter. It sounds like an interesting thing perhaps, but it doesn't arouse my curiosity. I certainly like tools, and like the (often surprising) activities and investigations worthwhile tools can lead to. But one audio philosophy I've long held is, "everything adds". In my view, things such as the Lacinato may lead to parsing of elements that may make no perceivable individual contribution...yet such "imperceptible" elements can still make their particular contribution to an overall, very synergistic, and complex whole. At least that's my story and I'm sticking to it! (Still, I can certainly see occasions where a person needs to sort out which path to head down.)

    The PM correspondent was not someone one I've experienced or known of before. The web has its wonders. But I'm truly grateful for the knowledge they've shared, and have permitted me to share even further. The ADI-2 Pro is now more transparent than it had been.

    As to your point about the ripple perhaps, perhaps being related to DSD capability in an ADC--I don't know. Maybe at some point someone in this thread will get their hands on a Mytek Brooklyn ADC, (will still need a DAC for the full D/A/D--a Brooklyn DAC could work)--but the chipset also should be made visible.

    I agree that the ripple is cosmetic. RME seemed to have made the same conclusion. However, it's certainly been fun putting the ripple through the wringer!
     
  5. darkmass

    darkmass Forum Resident

    Hmm. The Mytek Brooklyn ADC link I supplied just above in post #103 is not so very good. This is supposed to be better: link.
     
  6. back2vinyl

    back2vinyl Forum Resident

    Location:
    London, UK
    I know what you mean about "everything adds" - the cumulative effect of imperceptible differences. To that extent, I'd say the Lacinato ABX/Shootouter is only half useful. A positive result tells you a lot but if you get a negative result, meaning you can't hear a difference, it doesn't necessarily a difference doesn't exist - it only means you can't hear it. The only reason I mentioned the Shootouter was because we'd been talking about whether ringing was audible and I think that's one question the Shootouoter could answer.

    But it sounds as though I'm pushing the Shootouter and I'm really not - I'm just chatting. And of course the obvious riposte is ,"If you're that desperate to know whether ringing is audible, back2vinyl, then why don't you do the darned test?!" Maybe I even will one day, but for the moment, I'm trying to get a pair of speakers I can listen to and enjoy!
     
  7. darkmass

    darkmass Forum Resident

    I do not at all think you are pushing the Shootouter. It's much more a matter of something you've picked up, tried out, and are putting out there as, potentially, one more tool in the kit. And, like any tool, its benefits and limitations deserve understanding before it can serve a useful purpose...or at least that's my read.

    But we could be drifting off topic a bit, so let me bring us back to RMAA in an obscure way and then add a thought or two about trying to hear ringing.


    I don't know if you followed the link I threw in within the last few sentences of post #97, the link that went here, but it's a link I rather like...and it bears on ringing and on RMAA.

    The purpose of that page is to allow readers to look at results of using various SRC algorithms to convert 96 kHz to 44.1 kHz. Pertinent to the topic of "ringing", one of the results on view for each SRC is "impulse response"--which displays ringing characteristics. There's a help link from that particular page which adds a bit of information for each displayed SRC characteristic, and there's a linked FAQ page. The tie-in to this particular thread is that Alexy Lukin, of iZotope, and of RMAA, joined that SRC comparison project after it started, proposed several new tests, and "The graphs are generated in (slightly specialized) versions of RMAA and [iZotope] Rx progams, which are co-developed by Alexy Lukin."

    Now I would love, love to have those "slightly specialized" versions myself, and more than that I would love to have source test files in each of the common sample rates...and I'd like my slightly specialized version of RMAA to be able to correctly consume whatever sample rate I SRC'd any of the test files to. Now that would be interesting!

    But the FAQ page does have a link to 96 kHz test files (http://src.infinitewave.ca/TestSignals.zip), and that's not unuseful. Specifically, I've taken the "Pulses_24.wav" file, SRC'd it to 44.1k and 192k, using different iZotope "Prering" settings, then zoomed way in to look at results. In the source test file each pulse is one sample wide, but after an SRC the zoomed in pulses show linear/minimum/intermediate phase ringing effects, depending on how I set the iZotope Prering option. Also, with adroitness, it might be possible to run the initial Pulses_24 test file through some D/A/D configuration and zoom in to look at individual pulse results.

    Back to ringing audibility... As I see it, a significant complication with aurally differentiating ringing is the difficulty of having only a single ringing type feed the audio. For example, if you SRC a file of clean pulses to a file of minimum phase pulses, then listen to the results through a DAC, you are listening to "DAC phase" + "minimum phase". Of course if you SRC'd to linear phase pulses, through that same DAC you are listening to "DAC phase" + "linear phase". If a person has a DAC with various filter options it probably makes the most sense to try one DAC filter option or another while listening to the straight, un-SRC'd, 96k "pulses" test file through the DAC. In the case of your Lynx Hilo, which indeed has no filter options, then an SRC (assuming filter options) may be the best way to go, even if the audible filter phasing was really a compound phasing. Try to isolate variables as much as possible.

    I'm not "riposting" you, of course. I'm just setting down my thoughts concerning how a ringing evaluation might be handled. Rather than a "pulse" test file, music with well-known characteristics could serve as a source--though one concern is that any digital music recording (hmm, probably not good DSD) would additionally have its own filter phasing (probably linear phase) embedded.


    All that said, speakers are rather vital to this pursuit of ours. I hope you can find speakers that suit your needs and desires.
     
    Last edited: Jan 8, 2018
  8. back2vinyl

    back2vinyl Forum Resident

    Location:
    London, UK
    I must confess that, no, I didn't initially follow that link you kindly provided. But having now done so, I discover that I've stumbled on that website before. I remember exploring their test results and being pleasantly surprised to find that Adobe Audition CS6 performed particularly well, which gave me the excuse I needed (apart from cost, of course) to go on persevering with it and not to switch to the subscription-based CC model. In fact, Audacity in best quality has nothing to be ashamed of as far as I can see and it's on a later version now (2.2.1) than it was when they tested it.

    That aside, it's interesting that you found a way of using their 96 kHz test files to do some measurements of the different settings on the iZotope prering sider. I would think that since you've been able to do some actual measurements, you probably don't need any Shootouter to tell you which looks best, regardless of audibility. And as you point out, there'd be a big question mark in any case over how you'd set up a Shootouter test to yield any useful results. But thank you for pointing a way forward if I ever do get around to this. Probably a bit unlikely, to be honest - I'd be the first to admit that I'm struggling to keep up with you at this point!
     
  9. darkmass

    darkmass Forum Resident

    It's certainly time I checked out of this thread. At least for now, at least till something of interest appears.

    Adobe Audition truly does exceptionally well with respect to other SRC implementations. It does make me wonder, curious as I am, if there might be areas of compromise for Audition outside the region of the infinitewave tests. There would be nothing wrong with that of course. If "there are always trade-offs" is a reasonable check on desires versus reality, some trade-offs are better than others...and I believe that Adobe would do a very careful job of keeping its ducks in a row.

    As a side note on "subscription-based" software, I ceased my full Adobe Photoshop upgrades when Adobe went to subscription for its best software products. I have no desire to rent software.

    Back to vaguely what this thread is about...

    When I "zoomed in" to iZotope SRC results, I did not do any measurement. I simply looked at the shapes that resulted from various values of iZotope SRC's "prering" setting. Since your response, I went back for another look, maybe a bit more careful look, and put some things together so you'll have an idea of what I saw.


    The source, followed by three filter phasing results:
    [​IMG]


    The top example in the set is the first impulse from the infinitewave "Pulses_24.wav" file. As I mentioned, it is a 96k/24 file. The dots in the image each represent a sample point. Since Sound Forge Pro connects sample readings with straight lines, the pulse looks like a cone, however the actual pulse is only a single, point wide, sample.

    The three examples below the extraction from the Pulses_24 file are the results of using iZotope SRC to resample the Pulses_24 file to 192k/24. In order, from upper to bottom, the examples are with iZotope "prering" at 100% (full "linear phase"), 0% (full "minimum phase"), and 25% (the way I have been normally down sampling with iZotope). Note that each of the four windows represents the same overall time span. The sample points in the 96k/24 window are twice as far apart as the sample points in the three 192k/24 windows. With a 192 kHz sample rate, individual samples are only 5.2 microseconds apart! (The "1/4 minimum phase" is a misnomer, but I'm going to stick with it. It's more properly "25% linear phase and 75% minimum phase. Blame iZotope's 0% to 100% "prering" slider. I guess I tend to think of it as 25% of the way up from minimum phase.)

    Visually, there is little to differentiate the full minimum phase from the "1/4 minimum phase". Following the pulse, things are close enough to identical. Preceding the pulse is where the differences are, and they are subtle indeed.

    So when I ran my listening tests to determine my personal optimum "prering" setting in iZotope SRC did I actually hear something that might withstand a full "blind" test? I'm convinced I didn't hear anything of the sort. But I also know I don't care about that in the slightest. My full sensory input--moderated by my personal history as well as assorted capricious whims and stray thoughts--permitted me to believe I liked something. And that's good enough for me. :)


    Back2vinyl, this has been one more in our series of very enjoyable conversations. Thank you for that. I look forward to the next.
     
  10. back2vinyl

    back2vinyl Forum Resident

    Location:
    London, UK
    Haha, good enough for me too! Excellent charts - in all of this, I think I've actually managed to garner some slight understanding of ringing and the difference between linear phase and minimum phase. I presume linear phase is exactly what I'm getting from my old Hilo and since I have a bit of a soft spot for the thing, and since my OCD leads to a desire for symmetry, I think my own capricious whims would allow me to convince myself that I was in a good place with no need to change! Thank you again for the painstaking research and explanations which are such an asset to our hobby and as always, I very much look forward to your next contribution.
     
    darkmass likes this.
Thread Status:
Not open for further replies.

Share This Page

molar-endocrine