Tír Na HiFi

sbgk wrote:Thanks John, informative stuff.

how does the device know it's dealing with a poor usb signal ?

It is just a USB hub chip that is well implemented. What a USB hub does is allow a USB in connection, decode it & regenerate a new USB signal that is transmitted out of a number of USB ports. This particular USB hub chip has 2 output USB ports but only uses one of them.

It is this regeneration of the USB signal happening on the chip that cleans up the signal - as long as the PS & clock are properly implemented. I've done some experiments & even without the careful impedance matching that JS talks about, there is a worthwhile increase in SQ.

Maybe better impedance control would bring more improvement - I don't know but I suspect that the improved SQ is more about the other factors I mentioned. In fact what we might be seeing is the same thing as Nige has done - but he did it at the PC end i.e changed the USB clock & improved the PS feeding the USB port - whereas this is being done at the receiving end or close to it

We won't know this until we try it on some of Nige's builds that have a treated USB port

We are all suffering from the design decisions made by designers/engineers who thought bits are bits.

Yes, it's all a consequence of using equipment designed for digital processing where noise below a certain level is not of consequence i.e if it doesn't flip bits then it will work in the digital domain.

As we've always said here, general purpose computers are not optimal for audio - it takes trojan efforts to get anyway close to this.

We are also faced with a number of challenges in all of this:
- we can hear the SQ improvements but so far nobody has been able to measure them
- I suspect that low level noise floor fluctuations are at the bottom :) of this & this will prove particularly difficult to measure especially as the tool of choice for audio measurements is FFT
- Another challenge is that these SQ improvements are not usual stuff that we dealt with in the analogue domain - amplitude/frequency/phase changes - they seem to be more insidious than this - SQ improvements that affect the whole presentation of the audio scene. BTW, I don't want to mention the dreaded blind A/B test , but I will - these sort of differences are not easy to identify for a number of good psychoacoustic reasons.
- finally, we are up against the fact that digital audio works reasonably OK, even with this noise issue so there are a lot of bits are bits types that think they have reached "perfect sound"

Maybe we should start another thread on this rather than pollute this one?

Ah, I see Nige has posted on this also - maybe the mods could hive off these posts onto another thread - what should it be called "Uptone Audio Regen Device"?

nige2000 wrote: its rather discouraging that so much has to be modified on the pc side to provide a clean signal
and yet clear improvements still seem to come with further experimentation

Yes, I agree Nige - I believe it's because we are using computers which are only concerned about getting the digital domain correct i.e. no bit errors. As a result there is a much larger leeway in these designs as far as the specification of noise is concerned - it's the blessing & the curse of digital audio

ive never heard anything yet immune to upstream power/ clocking influences (sure some devices more affected than others)

with ians fifo reclocker there was still clearly differences in sq with different software, pc mods, usb cable and type of usb to i2s device along with latency settings

the regen isnt really anything new id doubt it would out perform a cleanly powered usb card

Yes, that will be a good test - use one on the end of a cleanly powered & reclocked USB port & see if it makes a difference

i wonder if the 1's are mean mistake'd for 0's because of an unclean signal

I think we might have very noticeable sonic glitches if that were the case, depending on where the bit was flipped, of course

would be nice if we were able to design a very simple pc/player from scratch with what we learnt over the years

Ah, yes - it's like the old Irish joke - when a passerby is asked "How do I get to this place" - he answers "Well I wouldn't start from here"

We are starting from a place that is way off the beaten track of where we want to get :)

jkeny wrote:

nige2000 wrote: its rather discouraging that so much has to be modified on the pc side to provide a clean signal
and yet clear improvements still seem to come with further experimentation

Yes, I agree Nige - I believe it's because we are using computers which are only concerned about getting the digital domain correct i.e. no bit errors. As a result there is a much larger leeway in these designs as far as the specification of noise is concerned - it's the blessing & the curse of digital audio

ive never heard anything yet immune to upstream power/ clocking influences (sure some devices more affected than others)

with ians fifo reclocker there was still clearly differences in sq with different software, pc mods, usb cable and type of usb to i2s device along with latency settings

the regen isnt really anything new id doubt it would out perform a cleanly powered usb card

Yes, that will be a good test - use one on the end of a cleanly powered & reclocked USB port & see if it makes a difference

i wonder if the 1's are mean mistake'd for 0's because of an unclean signal

I think we might have very noticeable sonic glitches if that were the case, depending on where the bit was flipped, of course

would be nice if we were able to design a very simple pc/player from scratch with what we learnt over the years

Ah, yes - it's like the old Irish joke - when a passerby is asked "How do I get to this place" - he answers "Well I wouldn't start from here"

We are starting from a place that is way off the beaten track of where we want to get :)

lol
yea definitely agree
after lots of experiments its very evident pc wasnt a good place to start :)
its astounding how good it sounds even with some gremlins not yet evicted

nige2000 wrote: lol
yea definitely agree
after lots of experiments its very evident pc wasnt a good place to start :)
its astounding how good it sounds even with some gremlins not yet evicted

Yes, this is the reason for continuing along the route - it does bring SQ returns even though they are hard won.
To be honest, Nige, you are probably one of the very few who has taken it as far as you have.
Some day, hopefully all the battery supplies, crocodile clips, relays, stack of dead motherboards will be binned & replaced by a small, well designed single purpose yoke.

BTW, it seems the Beaglebone Black can now be run from 2 external audio clocks which is the optimal way to reclock the I2S signals coming off the board. This is probably worth revisiting now & seeing what hardware (PS mods, clock mods, etc) can be done on the board. Probably easier than PC mods & certainly cheaper.

Yea
Hope to look into BBB again twisted pair and ackos reclock design seem to have a lot of jewelry components that probably aren't necessary in the absence of noisy ps

I have suspicion that Linux might be a limitation always seemed looser and less detailed than the windows man combo
This is mostly due to testing daphile vs man on my audio PC
So we probably need mqn for Linux

Which reminds me when I first tested Ian's fifo relocker on the RPI I was amazed how much it made it sound like mqn
Just not as good
Regardless the BBB reclocked/synronously clocked is a very worthy project

Some interesting article on impedance matching vs length of the cable.
http://www.positive-feedback.com/Issue14/spdif.htm

I have an issue with the latest MQn "ldn v" versions.
Without AO 1.30, the popup "mqnplay.exe stopped working" appears but music plays OK. If I close the popup, music stops.
With AO 1.30, music plays for a couple of seconds and stops.
Is anyone having the same behavior? Any workaround?
Tks.

adolfo.a.aguiar wrote:I have an issue with the latest MQn "ldn v" versions.
Without AO 1.30, the popup "mqnplay.exe stopped working" appears but music plays OK. If I close the popup, music stops.
With AO 1.30, music plays for a couple of seconds and stops.
Is anyone having the same behavior? Any workaround?
Tks.

Hi adolfo

I also use AO 1.30 and do not have this behaviour.

I did have issues with v11 and v12, but no longer with v13, v14, v15 and not before v11.
Which versions did you try?

The last one I tried was v12.