SDTrans 384?

[gstew]

Hehe, we know we are crazy but we have our limits & every now & then we take stock of where we are & ask "how did we get here" :)
We have always maintained that we shouldn't be using a PC as the start point but so much progress was being made on this front that it's hard to get off that particular path & start a new one

This may be the crossroads - some testing & auditioning will tease this out but I also expect that our experience with the PC audio mods so far puts us in good stead for improving a SD card player - Let the games begin :)

BTW, I agree with all you say - the design principles of that board look right - it's just down to implementation details but in looking at it there doesn't seem to be any issues there either?

LOL! I haven't gone down the battery path yet, but when you see my digital source test-bed, you will know that I too, am crazy by standard measures.

I do have plans for some mods to the SDD... mainly adding PPS bypass caps (similar to what are on the back of the board Rick showed), separately powering the different sections, battery and/or that crazy light-powered thing as power sources, and remoting the LCD on a cable, possibly with it's own power supply. And of course good mechanical mounting and damping.

But modifying the heck out of my DAM DACs comes first!

Greg in Mississippi

[jkeny]

JKeny,

The Soekris clock is variable and adjusted to make his short I2S FIFO work, but the output will be close to the standard 45/49 frequencies used for the 44.1 and 48 sampling rate families, depending on which family is being selected.

I doubt it varies far from the standard clock frequency.

Yes, but this is clock speed you are talking about that has been derived from the one physical onboard clock on the Soekris board, PLLed within the FPGA chip & then output as the audio clock latching the shift registers in the ladder DAC output stage - it's not the raw, unprocessed output of physical clocks which can be used in a synchronous way to time both converter (SDTRANS) & DAC (SOekris) - this would be the optimal, lowest jitter clock configuration which also avoids the issues of multiple clock domains

OTOH, I doubt taking the higher-jitter clock from the DAM across the isolators and then across wiring into the SDTrans will improve on their lower-jitter NDKs directly connected. I could be wrong, but this is not an experiment that is high on my list.

Yes, I would consider this a low priority experiment. A much more useful approach would be to find a suitable DAC which used 22.2792 & 24.576MHz clocks & use one pair of physical clocks as the synchronous clock signal for both SDTRANs & DAC

You CAN use external clocks on the SDTran easily enough. I vaguely remember one of the implementation of the pricey Pulsar clocks was on an SDTrans... I bet it was transcendental!

Greg in Mississippi

Hmm, I always wonder about clocks like the Pulsar - I find improved PS to any clock goes a long way to improving the sound as most are sub-optimally powered anyway

[nige2000]

you can take the soekris onboard clk signal as master to feed back to source
not sure if thats worth trying
maybe i should try?

Nige, is this working? I seem to remember hearing different opinions on that.

Greg in Mississippi

yea doubt it will be better either
but i might be able to try

[jkeny]

Hehe, we know we are crazy but we have our limits & every now & then we take stock of where we are & ask "how did we get here" :)
We have always maintained that we shouldn't be using a PC as the start point but so much progress was being made on this front that it's hard to get off that particular path & start a new one

This may be the crossroads - some testing & auditioning will tease this out but I also expect that our experience with the PC audio mods so far puts us in good stead for improving a SD card player - Let the games begin :)

BTW, I agree with all you say - the design principles of that board look right - it's just down to implementation details but in looking at it there doesn't seem to be any issues there either?

LOL! I haven't gone down the battery path yet, but when you see my digital source test-bed, you will know that I too, am crazy by standard measures.

I do have plans for some mods to the SDD... mainly adding PPS bypass caps (similar to what are on the back of the board Rick showed), separately powering the different sections, battery and/or that crazy light-powered thing as power sources, and remoting the LCD on a cable, possibly with it's own power supply. And of course good mechanical mounting and damping.

But modifying the heck out of my DAM DACs comes first!

Greg in Mississippi

I would suggest a simple, quick & cheap experiment of using a 3.3V LiFePo4 as external 3.3V power into the SDTRANS may well reap some instant rewards & then go down the path of modding the Soekris ala Nige's menu of tweaks (again, most involve battey power)

[nige2000]

<SNIP>

why is this board supposed to be good?

Nige,

My take...

1. Very simple processing in the FPGA, limited types of files, no data transformation, no reclocking, just take what is on the card and feed it out as an I2S signal, so minimal noise created in playback.

2. Good use of the hardware and board layout. The clocks are close to the FPGA, it only outputs I2S, this goes directly to the buffer chips for direct output, to an LVDS chip for the PS Audio HDMI output, and to an I2S->S/PDIF conversion chip (Cirrus, as I remember).

3. Reasonable distribution of power regulation right at the local consumers along with good filtering. It's a plus that you can power all of the sections separately for higher performance.

But it is basically just a digital turntable, no metadata, having your music only on SD cards, and some care has to be taken to keep from corrupting them (from the manual, don't turn off or eject the cards while they are being accessed). It is not as convenient or as flashy as most computer digital playback. I'm good with that, but am finding the process of loading SD cards a bit of a pain.

Greg in Mississippi

P.S. I got mine on the tail-end of the last GB, it was the base unit without the upgrade caps (on the picture of the back of the unit that has been posted) and no remote. It cost me about $350 USD, with the GB discount.

P.P.S. If you think all the battery powered stuff you guys do is viewed as crazy, take a look in the later posts in that thread on how they experimented with setting up a light-powered source for the SDTrans384... shine bright lights onto solar cells to power the unit!

was just wondering why the sd trans is supposed to be better than the other sd card wav players that are cheaper
were going to be modding the boards anyway?

http://www.ebay.com/itm/Improved-FPGA-L ... Sw8cNUO9KX
http://www.ebay.com/itm/111721479669?_t ... EBIDX%3AIT

[gstew]

Yes, but this is clock speed you are talking about that has been derived from the one physical onboard clock on the Soekris board, PLLed within the FPGA chip & then output as the audio clock latching the shift registers in the ladder DAC output stage - it's not the raw, unprocessed output of physical clocks which can be used in a synchronous way to time both converter (SDTRANS) & DAC (SOekris) - this would be the optimal, lowest jitter clock configuration which also avoids the issues of multiple clock domains

<SNIP>

Yes, I would consider this a low priority experiment. A much more useful approach would be to find a suitable DAC which used 22.2792 & 24.576MHz clocks & use one pair of physical clocks as the synchronous clock signal for both SDTRANs & DAC

JKeny,

Actually the DAM clocking scheme is as follows:

"Clocking and FIFO
The DAC have a low jitter digital controlled oscillator (SiLabs si570), data is sent though a short FIFO and the FPGA and uC work together to measure incoming bitrate and adjust clock as needed, basically a digital PLL with very fast lock and very slow filtering. So the DAC itself only need serial data, word clock and bit clock, no master clock is needed, it will sync to whatever you feed it. "

So the clock frequency on the variable clock unit is varied slightly to make the FIFO work. AFAIK, it does not go through a PLL in the FPGA and I'd hesitate to call how it works a PLL, just a variable rate clock that keeps the FIFO from emptying (hopefully... there are some recent comments on the main DAM thread suggesting it may not work as expected in all cases, such as with an R-Pi I2S source which may not have a fully compliantly-clocked I2S signal).

Also, do you have any suggestions on that 'suitable DAC'? Of course, for anything run off the SDTrans using their MCLK signal, you will be running synchronous... that is one of my planned experiments, trying some EUVL & Curryman ES9022/23 & JL Sounds AK4490 DAC cards with the SDTrans. But that comes after I work the DAM mods.

Hmm, I always wonder about clocks like the Pulsar - I find improved PS to any clock goes a long way to improving the sound as most are sub-optimally powered anyway

I found Ian Canada's comments after trying a Pulsar in his FIFO most convincing. I think he does about as good of a job as anyone on powering them in his setup... and he might have used a LiPoFe4 on it.

Still, at the price for them, I won't be buying a set anytime soon, even after they become available again!

Later!

Greg in Mississippi

[nige2000]

I have just bought a SDTrans384 after reading the whole DIYAudio thread. Among others, I was impressed by Jonathan Carr (from Lyra cartridges) post stating he had heard the best digital sounds with this transport.

I also read a post in that thread from 'elecon', Stig Bjorge from Norway who is the owner of Lyra - Jonathan Carr's Boss. He said it was the best digital player he has heard.

Not half bad references.

Jonathan

what hastens me is is we dont know what people consider to be a good digital audio reference and im dubious about these reference points more so than anything else

sure we jump through a lot of hoops but we have jumped through them
we do have good digital audio references that has to be among the best of them

so will probably have to wait on adolfo's impressions

[gstew]

was just wondering why the sd trans is supposed to be better than the other sd card wav players that are cheaper
were going to be modding the boards anyway?

http://www.ebay.com/itm/Improved-FPGA-L ... Sw8cNUO9KX
http://www.ebay.com/itm/111721479669?_t ... EBIDX%3AIT

Nige,

I'd looked a lot at the first of the two you listed. And I also seriously considered both the QA550 and the larger and more expensive QA660 at one time. But as I've learned more about digital playback, I find the devil is in the details. I see that the SDTrans designers were very focused on getting the absolutely best sound from their unit and have paid attention to the details*... and evidence of that is in all the different powering schemes they have posted with the unit and their attendant DACs, up to and including the crazy light supplies. They (AFAIK) made very conscious decisions on things like the very limited processing for a limited number of source file types, the low-powered processors, the fairly expensive NDK clocks, all of the regulators, good layout, etc. They've also worked with at least one Japanese hi-end DAC designer to integrate their product with their DACs. I know we can mod the EBay ones, but we can't fix all the areas where they didn't think it mattered or compromised or didn't have making the absolutely best-sounding product in mind. We can't rewrite the FPGA code or straighten out the board layout.

Now seriously, I don't know whether one couldn't mod one of the EBay ones up to the level of the SDTrans. But I bet those same mods (where needed) would take the SDTrans even higher.

Greg in Mississippi

P.S. And I am not afraid of modifying eq, even expensive eq. One of my projects last summer was an extensive set of mods on a Sony HAP Z1-ES, a $2k USD unit, so I could have a easy-to-use and pretty good go-to digital source. But I chose that unit because it already had a bunch of things done pretty right... and was designed and packaged such that it was a fairly simple to mod unit, for a fairly complicated all-in-one player. AND I could have had a very expensive boat anchor very easily!

*P.P.S. I consider the SDTrans to be the ultimate expression of what has us do things like run MQN on an optimized server with a modified motherboard. I consider it to be that tweaky, based on what the designers posted while they were developing it and all the various and sometimes crazy mods or setups they showed. All it needs is really good power supplies.

[jkeny]

Right, it seems like it requires reading that SDTrans threads on DIYaudio again :)

[jrling]

+1 Greg. Spot on.