1. WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

UpTone LPS-1 Linear Power Supply Review and Measurements

Discussion in 'Digital Audio' started by amirm, Aug 16, 2017.

  1. svart-hvitt

    svart-hvitt Active Member

    Joined:
    Aug 31, 2017
    Messages:
    206
    Likes Received:
    38
    [​IMG]
     
  2. Sal1950

    Sal1950 Major Contributor The Chicago Crusher

    Joined:
    Mar 1, 2016
    Messages:
    2,763
    Likes Received:
    384
    Location:
    Central Fl
    It's always the same fairy-tale stories. Superdad and the rest of the subjective clan will make claims of having thousands of customers that will testify to the improved SQ of their widgets and devices. So does David Copperfield, "that 747 disappeared right off the runway, I seen it with my own two eyes". But when it comes to ever supplying any supporting evidence they scatter like some insects do running from the light. They've been invited over and over at CA to participate in some bias controlled listening tests but no way Jose, that ain't never gonna happen.
     
  3. Jinjuku

    Jinjuku Senior Member

    Joined:
    Feb 28, 2016
    Messages:
    427
    Likes Received:
    105
    Does the randomness of coin flipping effect the eventual outcome? You let me know how many times you have a flip a coin to get something other than a ~ 50/50 difference.
     
  4. danadam

    danadam Member

    Joined:
    Jan 20, 2017
    Messages:
    16
    Likes Received:
    2
    As far as I can tell the discussion is about 60/40 not 50/50. You said that with a coin flip you could do as good or better than 60/40. I'm not a statistician but it seems to me that this may be true with 10 trials, but with let's say 40 trials the probability of getting 24 or more successes is only ~13%. So it very much depends on the number of trials.
     
  5. amirm

    amirm Founder/Admin CFO (Chief Fun Officer) Staff Member

    Joined:
    Feb 13, 2016
    Messages:
    8,622
    Likes Received:
    1,211
    Location:
    Seattle Area
    First welcome to the forum. :) Second, thanks for the reasonable way you are asking your question.

    "Noise" is the easiest thing to measure with instruments. And that is what I have been showing. That other than a single, poorly designed DAC (Schiit Modi 2), all other DACs are so immune to noise from the PC that no extra cleaning of either USB bus or power supply makes any difference. The measurements have been so accurate as to show that the external power supply that these devices can generate noise that was not there at the start.

    All of this points to no value unfortunately in this class of products even if your ears were as sensitive as our instruments.
     
    tomelex likes this.
  6. Wombat

    Wombat Member

    Joined:
    Nov 5, 2017
    Messages:
    84
    Likes Received:
    25
    Location:
    Australia

    You don't understand statistics and its application. Your question reveals this.
     
  7. bennetng

    bennetng Member

    Joined:
    Nov 15, 2017
    Messages:
    21
    Likes Received:
    2
    I attached a simple html app to simulate brute forced ABX. The questions and answers are randomly generated and every time browser is refreshed a new set of results will come out. Here is a sample result:

    6/10 or above in 1000 tests: 428
    7/10 or above in 1000 tests: 187
    8/10 or above in 1000 tests: 51
    9/10 or above in 1000 tests: 9
    10/10 in 1000 tests: 1

    Which means in this particular simulation, a wild guess without listening can yield 428 positive results in 1000 tests with 10 trials if 6/10 or above is considered as "pass". Also, it is not uncommon to see zero in 10/10 since the chance is very low.
     

    Attached Files:

    • abx.zip
      File size:
      548 bytes
      Views:
      5
    Blumlein 88 likes this.
  8. extracampine

    extracampine Member

    Joined:
    Nov 13, 2017
    Messages:
    8
    Likes Received:
    0
    I think this is called a tangent :) And thanks Amir for confirming your thoughts on the matter. Perhaps if that PC that was being removed from the listening room had a fan based cooling system which was audible, THEN there would be some benefit to the MicroRendu :)

    I find it fascinating that there can exist such a chasm between the views of those on some of the other audio forums, compared with those on more scientifically oriented forums such as this one. I broached the subject on the Roon forums (great software BTW) and the reply I received (from a user I didn't know) was somewhat adversarial.

    A common approach from the "scientific" camp in attempting to bridge this gap, as far as I can tell, is to suggest that the audiophile engage in double-blind tests of the equipment. I agree that this would make sense. I think it is done quite a lot, though I imagine that there are those (maybe more linked with manufacturers or sellers) that might not be keen on this option?

    The other issue though is to ask the scientific camp if they have "tried" the item in question and attempted to discern any difference with their own ears rather than with instruments. Of course this opens up the area of psychoacoustics, but this can be remedied with the double blind method. In addition, it tends to be stated that a sufficiently "resolving" (which equates to a large extent with cost) system is required to hear the differences. So we would require someone with a scientific approach, expensive hifi equipment, reasonable hearing and of course an open mind. I wonder if they exist :D
     
    Last edited: Nov 15, 2017
  9. Blumlein 88

    Blumlein 88 Major Contributor

    Joined:
    Feb 23, 2016
    Messages:
    2,071
    Likes Received:
    537
    Well suppose that person did exist. As a starting point. So what happens when their path of investigation leads them one way rather than another. The result is then used to accuse. Not so good hearing after all, gear that isn't resolving or becoming closed minded. Neat circle of logic.
     
  10. extracampine

    extracampine Member

    Joined:
    Nov 13, 2017
    Messages:
    8
    Likes Received:
    0
    Well, of course some might accuse. But that should by no means deter from this method. There will always be those that eschew a particular method or approach in the search for truth and understanding, but that shouldn't hinder the search.
     
  11. BE718

    BE718 Major Contributor forum experimenter

    Joined:
    Mar 1, 2016
    Messages:
    1,353
    Likes Received:
    260
    Well as an example I bought a Regen and couldnt hear a difference with any DAC I tried it with. I then measured it and found it caused increased noise on the DAC output. 8 kHz usb packet noise. This was caused by a ground lift resistor they had "subjectively" decided to insert in the product. This noise was confirmed by Amir and others.

    I think its fair to say there is a bit of history here and Uptone have been engaged with for a few years. Nothing compelling has come out of it.

    Unfortunately the truth is that there is a lot of hifi foo out there and people fall victim to psuedo science marketing all too easily, hence the chasm between this and some other forums. As soon as you put people under controls the magic trinkets are suddenly not so magic anymore.
     
    tomelex likes this.
  12. Jakob1863

    Jakob1863 Active Member

    Joined:
    Jul 21, 2016
    Messages:
    279
    Likes Received:
    60
    Location:
    Germany
    Which is a different situation. Doing a 10 trial ABX and repeating it 1000 times isn´t the same as to ask what number of trials it needs to get a significant result when delivering 60% correct answers.
    In the case of a 10 trial ABX the probability to get 6 correct answers by random guessing is p = 0.377 , so the expected value to get this result when doing the same test 1000 times is 377 tests with this result.

    But otoh if you do a 80 trial ABX then you´ll have a significant result on SL=0.05 level if the participant got 60% correct answers.
    P(48 l 80) = 0.046 .

    Btw, doing an ABX without training of the participant isn´t a good idea as it is known that the internal mental processes are more involving compared to A-B tests.
    Doing such tests without positive controls isn´t a good idea.
    Doing an 10 trial ABX isn´t a good idea unless the detection ability of the participant isn´t really high (means > 90% ) otherwise the probability of committing an error of the second kind (means to not reject the null hypothesis although it is false) will be way to large.
     
    Soniclife likes this.
  13. bennetng

    bennetng Member

    Joined:
    Nov 15, 2017
    Messages:
    21
    Likes Received:
    2
    I don't know why you interperted my simulation as a human being doing 1000 ABX tests. It would take at least several months if I am going to do that, unless I don't listen at all but guess. A more reasonable assumption is 1000 different person doing the same ABX test.

    I just wonder are you too sensitive about what I said because of your previous history in debating with other people in this or other forums? I didn't ever mentioned p-value in my post. If you want to find someone to start such a debate, find someone else, not me.
     
  14. Jakob1863

    Jakob1863 Active Member

    Joined:
    Jul 21, 2016
    Messages:
    279
    Likes Received:
    60
    Location:
    Germany
    I simply cited your post and pointed out that it described a different situation and added some general advice that should be taken into consideration when doing controlled listening experiments _and_ being interested in getting _correct_ results.

    Thought it could be of interest to some readers..... :)
     
    Last edited: Nov 15, 2017
  15. Jinjuku

    Jinjuku Senior Member

    Joined:
    Feb 28, 2016
    Messages:
    427
    Likes Received:
    105
    Not sure who you are thinking of that doesn't understand. Your array and my understanding line up. Someone even wins the lottery now and again but they were still 100% guessing what the #'s were going to be.

    My point is you need to flip an awful lot of coins to get what seem to be non-randomized results. while doing 10 coin flips 1000 times, by distribution, should net you at least one sequence where it all comes up heads while not impossible, highly improbable.

    Just to put this into context I've been offering up to $10,000 to someones $2000 for an 18 out of 20 correct choice. Loser pays travel expenses. I understand they are flipping coins. If I could get 1000 audiophiles to do this I'd be well off indeed.
     
    Last edited: Nov 15, 2017
    bennetng likes this.
  16. Soniclife

    Soniclife Member

    Joined:
    Apr 13, 2017
    Messages:
    84
    Likes Received:
    13
    Location:
    UK
    Sorry, my question wasn't clear looking back. I was asking how many tests did the subject do with the cables, that ended with 60% right?
     
  17. amirm

    amirm Founder/Admin CFO (Chief Fun Officer) Staff Member

    Joined:
    Feb 13, 2016
    Messages:
    8,622
    Likes Received:
    1,211
    Location:
    Seattle Area
    Absolutely. That functionality of being able to place a PC in a remote location is absolutely useful. That should be the main reason people buy these streamers.
     
  18. jtwrace

    jtwrace Active Member

    Joined:
    May 31, 2017
    Messages:
    113
    Likes Received:
    18
    Location:
    Charlotte, NC
    If anyone is looking for an LPS-1 I have one for sale. Amazing product but I have no use for it. :D
     
    tomelex likes this.

Share This Page