Are all you nuts paid to push HQPlayer or something? Maybe you should use your electricity and CPU cycles for something useful, or at least profitable like mining crypto.
Maybe I was wrong to suggest it's off-topic, but whatever.
Last edited:
How did you know, I used to do both ;-)
Haven't got time to get to dacs anymore, sign of the times with the kids having been home everyday, wife working for the elderly, so no leave etc.
Prototyping will commence somewhere later this year, fingers crossed.
Nowadays, just buying crypto pays off more, and I use a 8 channel oldie dac that is all firewire and pcm.
That software player that you brought to our attention can even do peq filters, inverse phase, adjust distance and convolution, for all the 4 channels I need.
It's oke to use pcm. Lightweight ;-)
Haven't got time to get to dacs anymore, sign of the times with the kids having been home everyday, wife working for the elderly, so no leave etc.
Prototyping will commence somewhere later this year, fingers crossed.
Nowadays, just buying crypto pays off more, and I use a 8 channel oldie dac that is all firewire and pcm.
That software player that you brought to our attention can even do peq filters, inverse phase, adjust distance and convolution, for all the 4 channels I need.
It's oke to use pcm. Lightweight ;-)
I use a laptop with mobile i7 , 2core, 15W processor and I can still comfortably use DSD256 with filters and modulators which are still not -2s, and sound natural and nice. Quite a difference to the dac's in-built crippled half-band leaky filters and pcm modulators.. (4499)
But who am I to stop somebody with that deviant desire to esclusively listen to crãp like that?! So peace, everybody to their own toys.
But who am I to stop somebody with that deviant desire to esclusively listen to crãp like that?! So peace, everybody to their own toys.
You see you can't post something inflamatory and even go out of your way to bypass the forum filter then say 'peace everybody' having just insulted anyone who doesn't use HQplayer. Makes you come across terribly arrogant with a superiority complex.
Fairy nuff. Misread you, sorry. With small kids even low hanging fruit needs binoculars to see
Here's the much sought after close-in noise effect on an analog DAC output. Conditions:
- PCM5142A DAC chip with datasheet dynamic range=112dB SNR=112dB THD=-93dB. This is a poor DAC by any high end audio metric, it is for TVs, auto, AV receivers, other cheap consumer products.
- DAC Clock is a $3 ASTX-H11-20.000MHZ-T (Abracon Corporation) with -130dBc @1kHz offset, no spec below. By all high end audio metrics, a cheap piece of junk.
- The DAC is followed by a 2KHz 8th order filter. Before the filter, distortions in an 80KHz bandwidth is measured -97dB, after the filter is measured -128dB, both THD.
- Single ended output, DAC is @FS (analog: 11.88Vpp = 22dBV)
- SA driven by 10MHz reference from a Datum 2100 TymServe GPSDO with a MTI-270 oscillator OCXO in a PLL. OCXO phase noise is specified as -105dBc @1Hz and -157dBc @100KHz, better when the PLL is closed, no idea how much.
- FFT with 3601 points, RBW=30mHz, span 10Hz, flat window (high amplitude accuracy), 15 averages. Process gain is about 34dB @fs=48KHz.
From the screenshot (vertical 20dB/div, Hor 1Hz/div):
- Noise floor is around -117dB
- No spurs in this BW
- Close-in phase noise creeping from the noise floor at 1Hz around the carrier and extends to -90dB
- SNR = 22dB + 117dB = 139dB. Subtract the process gain and you'll get the DAC SNR = 107dB. Lower than the DAC spec, but this is measured very close (5Hz) away from the carrier, probably in the 1/f phase noise region.
- DAC SNR at lower than 0.1Hz from the carrier is 22dB + 90dB - 34dB = 78dB
Now somebody has to show proof that he can discriminate by hearing -78dBc noise @0.1Hz Note: Masking effect is real.
This is a mediocre DAC chip (to put it mildly) and a cheap DAC clock. Though the Golden Ear that proves, by a rigorous controlled listening test (anecdotal stories are not acceptable), that he can identify this crap DAC, from the best of the best DACs and clocks on the market (self made, modded, or commercial) gets a prize and my apologies for sub estimating the human hearing abilities.
P.S. Why did I bother to show this? It will not change anything, nothing could. I guess I'm an ethernal optimist, though.
- PCM5142A DAC chip with datasheet dynamic range=112dB SNR=112dB THD=-93dB. This is a poor DAC by any high end audio metric, it is for TVs, auto, AV receivers, other cheap consumer products.
- DAC Clock is a $3 ASTX-H11-20.000MHZ-T (Abracon Corporation) with -130dBc @1kHz offset, no spec below. By all high end audio metrics, a cheap piece of junk.
- The DAC is followed by a 2KHz 8th order filter. Before the filter, distortions in an 80KHz bandwidth is measured -97dB, after the filter is measured -128dB, both THD.
- Single ended output, DAC is @FS (analog: 11.88Vpp = 22dBV)
- SA driven by 10MHz reference from a Datum 2100 TymServe GPSDO with a MTI-270 oscillator OCXO in a PLL. OCXO phase noise is specified as -105dBc @1Hz and -157dBc @100KHz, better when the PLL is closed, no idea how much.
- FFT with 3601 points, RBW=30mHz, span 10Hz, flat window (high amplitude accuracy), 15 averages. Process gain is about 34dB @fs=48KHz.
From the screenshot (vertical 20dB/div, Hor 1Hz/div):
- Noise floor is around -117dB
- No spurs in this BW
- Close-in phase noise creeping from the noise floor at 1Hz around the carrier and extends to -90dB
- SNR = 22dB + 117dB = 139dB. Subtract the process gain and you'll get the DAC SNR = 107dB. Lower than the DAC spec, but this is measured very close (5Hz) away from the carrier, probably in the 1/f phase noise region.
- DAC SNR at lower than 0.1Hz from the carrier is 22dB + 90dB - 34dB = 78dB
Now somebody has to show proof that he can discriminate by hearing -78dBc noise @0.1Hz Note: Masking effect is real.
This is a mediocre DAC chip (to put it mildly) and a cheap DAC clock. Though the Golden Ear that proves, by a rigorous controlled listening test (anecdotal stories are not acceptable), that he can identify this crap DAC, from the best of the best DACs and clocks on the market (self made, modded, or commercial) gets a prize and my apologies for sub estimating the human hearing abilities.
P.S. Why did I bother to show this? It will not change anything, nothing could. I guess I'm an ethernal optimist, though.
Attachments
Last edited:
I think we should just agree, Andrea's is chasing a unicorn, he's interested in achieving theoretical perfection of one single performance metric. And that's fine. He's not looking to rigorously define a standard for audibility thresholds of said metric.
He's just looking to build a car with the best top speed, because that's where he believes nirvana resides. Once he's done that I have no doubt he'll do similar for the handling, fuel economy and other things he believes will make it better.
We can all agree it's a fast car, we don't have to agree it's the best driving car.
Step back, take a breath. Count to ten.
He's just looking to build a car with the best top speed, because that's where he believes nirvana resides. Once he's done that I have no doubt he'll do similar for the handling, fuel economy and other things he believes will make it better.
We can all agree it's a fast car, we don't have to agree it's the best driving car.
Step back, take a breath. Count to ten.
A quick pass with a BW of 100Hz (Hor 10Hz/div) confirms the above hypothesis, see attached screenshot: process gain is now 24dB @fs=48KHz and therefore the SNR is 22dB + 113dB - 24dB = 111dB which is right at the PCM5142A datasheet SNR of 112dB. Conclusion: the crap Abracon oscillator is actually not bad at all for this cheap DAC. It's phase noise corner frequency is likely buried in the noise floor.
Attachments
Well, I have to take at least partial responsibility for the HQPlayer following
here as I first recommended it to Mark quite a few years ago.
Ironically, (from memory), Mark, being the single minded person he is, initially argued tooth
and nail with me that HQP would not improve the sound over a good H/W
digital filter or upsampler.
So much for expectational bias as some here call it.
Just out of interest, have any of you HQP critics have taken the time to sit
your bum down and listen to it using a semi decent replay system?
TCD
Certain people keep saying they want proof, but they never say what they would be willing to accept as proof. Say, if I organized a group of several people and blind tested, double blind tested them if necessary, and most or all can hear a difference 9 out of 10 times or better, would that satisfy you?
I don't think so. First of all I don't think you would trust me to do it. If not, then who would you trust that lives near Auburn? If nobody, then your claim of wanting proof is BS as far as I'm concerned. You want hundreds of people tested by uninvolved university researchers? I'm sure not paying for that, so again, it not a serious offer to look at proof.
I don't think so. First of all I don't think you would trust me to do it. If not, then who would you trust that lives near Auburn? If nobody, then your claim of wanting proof is BS as far as I'm concerned. You want hundreds of people tested by uninvolved university researchers? I'm sure not paying for that, so again, it not a serious offer to look at proof.
Finally the measurement that everyone was waiting for has arrived.
Then the dacs are all the same!
This I really did not imagine.
Many thanks, now I throw out my devices and stop all new designs.
I don't even need to buy the DAC, I already have a nice 47 Euro ALLO Piano kindly offered by the manufacturer.
Of course, I also throw out the ALLO Kali, the FIFO buffer does not matter because there is no difference between its NDK oscillators and that of the RPI source with its nice PLL.
Nor there is a need to isolate the time domains between the source and the DAC, timing errors don't matter.
Recommended for everyone.
Maybe someone should change the title of this thread from "Best DAC ever" to Best DAC ever: they are all the same"
Congratulations, you have measured the diameter of a microbe with the tape measure.
BTW, I didn't say anything about HQ Player recently in this thread. Chris brought it up. I only said DSD, which more and more software can do a reasonable job of. For example, the DSD conversion in Roon is reasonably good. In addition, eventually someone is going to figure out a good way to do it hardware.
Think again. If you disclose the test protocol, the sample size ("several people" is not good enough), equipment, preparation, methodology, the whole shebang, then show the data, its interpretation, and the correctly drawn conclusions, then you are certainly going to be taken seriously, at least at the level to be asked serious questions, instead of bashing you for anecdotal bedtime stories. You may even persuade some to look closer at your report and identify the underlying phenomena, I know I would.
But I don't believe this is something you will ever try, it doesn't pay off, it is much easier (and cheaper) to attempt getting a reputation by massacring otherwise perfectly fine DAC boards, and creating story lines.
Last edited:
Mastering Engineers the world over (Bob Katz etcc) do blind listening tests to identify / choose / upgrade DAC's and ADC's a lot better than this.
WRT your specific clocking example, search Bob Katz and Grimm CC1 blind test. The jitter levels would be significantly lower than what you have shown
and specifically close in due to Protools PLL rejection CF.
This has been brought up and ignored here before.
TCD
"This method does have two drawbacks. First, if windowing is used during the FFT processing, the spectral resolution becomes blurred by the impulse response of the window. Second, for most reasonable FFT sizes, the spectral resolution is quite limited. For example, if an encode rate of 61.44 MSPS is used and a 64K FFT is performed, each FFT bin represents a bandwidth of about 938 Hz. It is reasonable to expect that clock noise within
several FFT bins will be lost to spectral blurring resulting in the loss of information several kHz on either side of the fundamental where much of the phase noise exists.
Even in the case where synchronous FFTs are performed and windows are not used, the limitation of at least one FFT bin is still imposed, representing about 1 kHz. From a close-in phase noise point of view, much of the energy is usually contained in the first few kilohertz around the clock source. Therefore, by using the FFT method for estimation of jitter, much of the clock noise is lost in the method."
"For this section it is assumed that the noise limitations of the sampling process are completely in the wideband noise of the clock."
The microbe and the tape measure.
So from that article
So two tenths of a gnatfart difference. 60% correct on 80 trials.
On HQplayer. If you change the impulse response with a different filter then it's not a huge leap to say the output of the DAC will change. DACs for years have had a choice of flavourings for those who reel in horror from a sinc pulse. So are you just putting in your own preferred filtering and noise shaping response and claiming different is better? Honest question as mathematically surely only one filter can be right to reconstruct the event?
Edit: Bob Katz last comment on that CC1 article
Last edited:
I disagree completely about practical proof. If I can find one person who can tell two clocks apart blind, all other variables held the same, and that person can do it every time, then that's enough to prove it is humanly possible.
On the other hand if you want statistics on what fraction of population can do it, after being trained on the apparatus of course, then you need at least a few hundred if not more test subjects. Otherwise the sample size is too small for proper statistical human perceptual testing.
My only interest would be in proving it is humanly possible .
On the other hand if you want statistics on what fraction of population can do it, after being trained on the apparatus of course, then you need at least a few hundred if not more test subjects. Otherwise the sample size is too small for proper statistical human perceptual testing.
My only interest would be in proving it is humanly possible .