1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Hugo M Scaler by Chord Electronics - The Official Thread

Discussion in 'High-end Audio Forum' started by ChordElectronics, Jul 25, 2018.
605 606 607 608 609 610 611 612 613 614
616 617 618 619 620 621 622 623 624 625
  1. delirium
    Loudspeakers are pretty safe....but headphones.....be careful.
    Last edited: Nov 10, 2019
  2. miketlse
    @mlxx has made some interesting posts (like the one you quote) on the subject, and they piqued my interest to try and understand the mathematical/coding challenge behind these upscaling discussions. I have posted briefly in an attempt to improve my understanding based on one of the links that was provided.

    As mlxx points out 'With a 44.1kHz song that is 4 minutes long, it would have 10,584,000‬ samples and we would need that many overlapping sinc functions.
    They would need to be long enough to cover the entire song and we would need to wait for all the samples before we reconstructed.
    This isn't practical, so we have to truncate the sinc function and do the reconstruction over smaller time periods with a smaller amount of samples.
    How you do this and the precision of the calculations affects the reconstruction and the sound quality (this is where WTA plays a part).'

    Rob has posted that it took him 10 years to perfect the WTA algorithm, which he has then used in his dac designs for the subsequent 20 years. Rob has also posted that it is possible for anyone to code an upscaler, but the difficult part is to develop an algorithm equivalent to truncating the sinc function and performing the reconstruction over smaller time periods with a smaller amount of samples - plus keep all the data in the samples aligned and in sync.

    As I mentioned, my interest was triggered to explore the upscaling topic this winter, but I am keeping in mind a quote by Einstein 'if someone gave me an hour to produce a solution, I would spend 55 minutes understanding the problem, then 5 minutes on the solution' - the 55 minutes took Rob 10 years, so I am under no illusions that I am only a few microseconds into that 55 minute period.

    To summarise, yes you can code your own upscaler, but do not expect it to be an overnight task to match the performance of the MScaler.
  3. dmance

    Jussi (HQPlayer)comes very close to the implementation quality needed to be WTA 1 comparable. I've done many hours of A/B using his latest ( just released) poly-sinc-long-lp filter. Not only is it a perfect match sonic signature, it's got way more bit resolution than 16-bits.
    Progisus likes this.
  4. nomad777
    lol... make your own mscaler ... I guess I could also make a Ferrari at least it would look like one...

    I'm listening to the HMS as I type this. Having owned both the blu MK2 and the HMS I can say the HMS is superior. More detail , smoother, timing. Some claim they don't hear any difference. Again don't know how that's possible again I'll point to their setup.

    See Audio bacons review and his thought and HMS vs Blu2

    Last edited: Nov 10, 2019
    Andrew DiMarcangelo likes this.
  5. rkt31
    Rob's FPGA's uses all it's processing power in real time while hq player can run on a decent i5 or i7 processor in real time. If hq player is giving more bit resolution, does it mean FPGA used in HMS is less powerful than i5 or i7 processor or Rob's algorithm is using more processing power and still giving less bit resolution ? Imho if it was so easy, rob wouldn't have required a FPGA.
    Last edited: Nov 10, 2019
  6. miketlse
    Always remember that Rob uses the capability of the FPGA to perform parallel processing, and keep all the data streams in sync.
    A simple i5 or i7 performs serial processing, so Rob has pointed out that this makes it difficult to keep all the data streams in sync, and so potentially introduces errors. One way to negate this is for the i5 or i7 to delegate all that processing to a graphics card that can perform parallel processing (eg CUDA enabled).
    So just comparing the FPGA chip to i5 or i7 processors is too simplistic an approach, because this does not consider how the data streams are handled.
  7. dmance
    PCM to PCM upsampling is far less burdensome to compute than PCM to DSD modulation.

    HQPlayer PCM to PCM is performant on a Core i3. And yes, a modern CPU runs rings around an FPGA. Some digital experts know FPGAs some know x64 instruction sets. Also, CPUs have access to extended precision 80-bit floating point vs FPGA 64-bit precision.

    I'm so convinced of the benefits of software upsampling dominating that ive staked my commercial reputation on it.
  8. miketlse
    Well at least that clarifies that in my mind, that there is no potential for bias of any kind in your posts.
    zenlisten and kelly200269 like this.
  9. dmance
    Mike - everyone has an agenda but in this world, even on this forum, there is such a thing as altruism - an honest desire to help fellow music lovers better understand their systems and the ways to obtain the best from it.
    zenlisten and Ards like this.
  10. Drewligarchy
    Anyone have any idea why this battery wouldn't work with HMS? Any chance it could damage it based on how it's described?


    Also, what is the ideal voltage for HMS? I "think" I saw Rob say in his preso (or elsewhere is a post) it was 9-15. Are there any differences, or as long as you are in this range you are good? This is variable but would have to be set on 12 or 16?

    I could order a PowerPilot from Amazon, but B&H is physically close to me so I could easily buy pop in and buy it. That said, I think Rob uses the PowerPilot - so if there is any chance it will screw up the HMS - I'll just order from there.
    Last edited: Nov 10, 2019
  11. miketlse
    Yes, but posters have been quite happy to claim 'well he would say that' about Triode and cables, and Rob and his dac designs, and now they will be able to suggest the same about your posts about software upsampling.
    That is a fact of life, once anyone initiates a commercial interest in a product, and there will always be questions about potential bias.
    You may well have 'an honest desire to help fellow music lovers better understand their systems and the ways to obtain the best from it', but all your potential customers on head-fi and elsewhere, will be the ultimate judges.
  12. miketlse
    Rob has posted to set PowerPilot to maximum of 12V for the HMS, because there is an input protection diode that will fail at less than 16V, and Chord will use it as the criteria when assessing warranty claims.
    Last edited: Nov 12, 2019
  13. Drewligarchy
    Thanks. Just want to make sure I understand correctly. Based on the posts Rob says to set to 12V. Therefore, did you mean the input protection diode would fail at more than 16V (Rob said in one post to set to 12, and in another that 15 is the max input rating).

    In other words, 12V is safe?
    Last edited: Nov 10, 2019
  14. nomad777
  15. miketlse
    12V is fine, but the MScaler will handle 15V as well.
    I think there is a post somewhere where the failure voltage is mentioned in more precision, and it is 15.x V which gives enough tolerance margin for supplies rated at 15V.
    Last edited: Nov 10, 2019
605 606 607 608 609 610 611 612 613 614
616 617 618 619 620 621 622 623 624 625

Share This Page