M2TECH Hiface USB->SPDIF 24/192Khz asynch

[labjr]

Originally Posted by jkeny

Finished most of the potential mods to this unit - here's my experience/impressions:
- The clean 3.3V supply to the clock is a must do - it clears up the sound substantially
- After I did this I put a scope on the existing 3.3V on-board - pic below - looked like 40mV peaks @ 80KHz?
- I took off the 3.3V switching down-converter & supplied 3.3V from battery to the rest of the board - another jump in clarity - not surprising as this supplies the PS to the CPLD IO banks & the clock is Muxed & buffered there.
- The only switching converter left on board is 1.8V & removing it may not have too much benefit as it supplies the core logic to the CPLD (apart from removing whatever hash this puts on the ground plane which may be no bad thing)

This is the best audio I've had in my system to date - I reckon it would compete well against the best. The weakest link now, I reckon, is the fact that there's no way to galvanically isolate the USB 2.0 signal so the quality of the sound depends on the computer (& PS) feeding it.

Time to shift it to my main listening area and out of the work-room & do a bit of listening before I tackle the Musiland or maybe the Squeezebox receiver

Wow, Looks interesting ! I'm wondering how does anyone else galvanically isolate the USB signal?

[analog_sa]

Originally Posted by labjr

Wow, Looks interesting ! I'm wondering how does anyone else galvanically isolate the USB signal?

Most do not. I am tempted to try an ADuM4160. My current solution is to simply isolate the output of the dac chip through a transformer which means that all the PC ground garbage gets through to the USB chip, oversampler and dac - not ideal.

[analog_sa]

Hi Jkeny

Thanks for the great research. I've been looking for a chip to replace the 2707 and this clearly seems to work. Sadly, the implementation appears to be so poor that a complete redesign is probably required. Maybe the manufacturer would be interested in selling bare chips and drivers to diyers.

[vuki]

Originally Posted by jkeny

Finished most of the potential mods to this unit - here's my experience/impressions:
- The clean 3.3V supply to the clock is a must do - it clears up the sound substantially
- After I did this I put a scope on the existing 3.3V on-board - pic below - looked like 40mV peaks @ 80KHz?

Maybe one could use low drop linear regulator like lm1117 instead of switching one?

[jkeny]

Originally Posted by analog_sa

Most do not. I am tempted to try an ADuM4160. My current solution is to simply isolate the output of the dac chip through a transformer which means that all the PC ground garbage gets through to the USB chip, oversampler and dac - not ideal.

No an ADUM won't work with these new USB 2.0 devices - they run at Hi-Speed (480Mbit/s) so nothing I know of is fast enough to pass this through

Sadly, the implementation appears to be so poor that a complete redesign is probably required.

Don't be so harsh on the developers - I'm sure they designed the unit to be portable - mine with mods + Batteries is no longer so because it has been changed for a different goal - best sound. Still all boxed up it should be a neat unit - I never liked the idea of the unit plugging directly into the USB port anyway . It is prone to the same PS problems that arise in the Musiland also - switching regulators. But these are easily removed & when clean PS supplied instead the unit really sings so not a bad underlying design really.

Don't get too hung up about the galvanic isolation - it still sounds wonderful & I don't know how much better it would be with galvanic isolation. I do know that when I power my laptop from my Linear bench supply the sound improves so maybe that is the area to focus on cleaner, non-switching power supplies to the laptop/PC? If you really want galvanic isolation then go for a Squeezebox Receiver which is what I've done. Only problem is it's limited to 16bit 44/48KHz ( 24/96 when the SB Touch is released) but I reckon that modded this will sound wonderful also probably better than the HiFace!

[jkeny]

Hey Vale, Guido, you really need to try the clock PS mod & report your findings!

[Valeriano]

Hi John,

very nice job you did, congrats.
you like the sound now ?

as I told you in a previous PM, I replaced onboard regulators on my hiface with ultra low noise linear regs.. tps79318 and tps79333 from TI

pinout is the same except one which I lifted and connected to a bypass cap (as per datasheet).. I also replaced two inductors following the previous reg output with two zero ohm smd jumpers. a bit tricky mod due to size.

this improved the sound quite significantly
I compared mine with stock hiface via spdif (wm8804) .. modded hiface sounds better by a margin

my mods so far : external PS, onboard regs, i2s tapping

I know you're playing with nanotech batteries for quite some time.. it's time to order some A123 lifepo4.. do you have a source in EU?

and yes, next thing I'll try is clock on battery

maybe would be a good idea to design a simple battery based PS for every critical low amperage component (DAC, hiface,..) with builtin charger/switcher

PS: USB isolation at hi speed (480Mbs) is not available on any known chip (at least to me).. maybe we can focus on I2S isolation using ADUM ?

cheers,
Vale

[SamL]

Originally Posted by jkeny

If you really want galvanic isolation then go for a Squeezebox Receiver which is what I've done. Only problem is it's limited to 16bit 44/48KHz ( 24/96 when the SB Touch is released) but I reckon that modded this will sound wonderful also probably better than the HiFace!

Will be interesting to find out how the transport function of the Hiface in comparison with the SB. By right the Hiface with propriety software should be better. But... one never know.

[jkeny]

Originally Posted by Valeriano

Hi John,

very nice job you did, congrats.
you like the sound now ?

as I told you in a previous PM, I replaced onboard regulators on my hiface with ultra low noise linear regs.. tps79318 and tps79333 from TI

pinout is the same except one which I lifted and connected to a bypass cap (as per datasheet).. I also replaced two inductors following the previous reg output with two zero ohm smd jumpers. a bit tricky mod due to size.

Ah yes, forgot, I must try one of these for the 1.8V supply

this improved the sound quite significantly
I compared mine with stock hiface via spdif (wm8804) .. modded hiface sounds better by a margin

my mods so far : external PS, onboard regs, i2s tapping

I know you're playing with nanotech batteries for quite some time.. it's time to order some A123 lifepo4.. do you have a source in EU?

I ordered mine from Japan (they are now manufactured in China , I believe) from an Radio Controlled aircraft shop Air Craft - Electric R/C Aircraft World Shop. They don't seem to have the 2300mAh in stock anymore. I'm not sure about sources in EU. I'll PM you!

and yes, next thing I'll try is clock on battery

Yes, well worth it

maybe would be a good idea to design a simple battery based PS for every critical low amperage component (DAC, hiface,..) with builtin charger/switcher

That's exactly what I'm thinking too!

PS: USB isolation at hi speed (480Mbs) is not available on any known chip (at least to me).. maybe we can focus on I2S isolation using ADUM ?

cheers,
Vale

Yes that should work, but I haven't given it much thought.

[jkeny]

Originally Posted by SamL

Will be interesting to find out how the transport function of the Hiface in comparison with the SB. By right the Hiface with propriety software should be better. But... one never know.

Sam, I would fall in favour of the SB because it is isolated from PC by ethernet transformer & it has similar building blocks to HiFace a processor that controls Xilinx CPLD feeding I2S to DAC. I'll be removing WiFi & changing the PS, of course, with some other mods.