Lossless SD digital Player is computer audio now reborn?

[gstew]

Mine arrived late this week. After looking it over, I'll do pretty much the same mods as you did, Nige. First tho, I'll compare it stock to the SDTrans using the same supply, then upgrade the clocks and try it again. THEN I'll go to battery power.

I've already got some good NDK 22/24 clocks here and have located the 32kHz in the same series. Now I need to look for a good 8mHz clock... hopefully I can get it in the same NDK series.

I also found the cable to the screen and switches at Mouser and will order a couple of 4" ones. I plan to try powering the screen separately too, so I can cut the appropriate lines to provide the feed.

This is all down on my priority list, don't expect much on this before May or so.

EDIT: A couple more questions/comments on the EBAY SD Card Player....

Nige, does this thing actually take the stock power feed first through an 8v regulator, then through a 5v one before sending it through two 3.3v regulators, one for the processing and display side, the other for the audio signal generation and output? Seems like a lazy and probably not to effective of a way to do it, but then I am not an EE.

Nige, also, in the pictures, it looks like you still have the AC power input hooked up. Do you still need this even with the battery power feeds for some reason I haven't seen yet or have you just not taken the time to remove those yet?

Greg in Mississippi

[nige2000]

Mine arrived late this week. After looking it over, I'll do pretty much the same mods as you did, Nige. First tho, I'll compare it stock to the SDTrans using the same supply, then upgrade the clocks and try it again. THEN I'll go to battery power.

my first thoughts of it stock was "well it doesnt suck"
this board if fairly simple you could do all that in an evening

I've already got some good NDK 22/24 clocks here and have located the 32kHz in the same series. Now I need to look for a good 8mHz clock... hopefully I can get it in the same NDK series.

where did you source the 32khz ndk?

I also found the cable to the screen and switches at Mouser and will order a couple of 4" ones. I plan to try powering the screen separately too, so I can cut the appropriate lines to provide the feed.

ok ill let you test that one

This is all down on my priority list, don't expect much on this before May or so.

whats the priority list ill rearrange it for you :)
this is starting to look real promising

was reading your post on diya about the rpi been somewhat comparable to the sdtrans?
we did a bit of testing a couple yrs ago with lifepo4 powered rpi and beaglebone in conjunction with ians fifo reclocker and neither was as good as a modded pc and amanero with no fifo reclocker
the rpi was terrible without the fifo reclocker i2s signal is just bad

[gstew]

my first thoughts of it stock was "well it doesnt suck"
this board if fairly simple you could do all that in an evening

Good to know. I'm just curious how it compares to the SDTrans and want to have a good baseline.

Truth on the simplicity.... but "only in an evening" AFTER having all of the parts on-hand!

where did you source the 32khz ndk?

Mouser. I'll post a link tomorrow. I was looking at some various 8Mhz options too earlier, but put that aside to finish tomorrow.

ok ill let you test that one

That seems like a no-brainer to me, displays are noisy!

whats the priority list ill rearrange it for you :)
this is starting to look real promising

Sorry, work and some high-school student mentoring is going to keep me heads down for the next week or so, maybe some progress after that?

was reading your post on diya about the rpi been somewhat comparable to the sdtrans?
we did a bit of testing a couple yrs ago with lifepo4 powered rpi and beaglebone in conjunction with ians fifo reclocker and neither was as good as a modded pc and amanero with no fifo reclocker
the rpi was terrible without the fifo reclocker i2s signal is just bad

Hmmm... one of the other projects in-queue is an Ian reclocker setup feed by another Pi into a BuffII, mostly as an experiment on what makes a difference in the sound of that setup. So I'll have more perspective in the near future.

I REALLY have to be clear on the SDTrans vs R-Pi into the DAM that the SDTrans is clearly better. But the magic of the DAM comes through on both AND I find the setup with the R-Pi input more inviting to listen to with more PRAT and toe-tapping-ness.

Remember tho, my DAMs are basically stock right now, not even the standard V-Ref mods. So maybe things will change once I have improved the DAMs and I'll hear more of the R-Pi's warts.

Also, that assessment has me questioning how good my SDTrans setup is. I'm slowly making plans to do an upgraded AC power setup for it along with a battery supply through the stock 5v input... to see if I can help it pull away.

And finally, maybe, just maybe all of my reference points are just not as good as your "modded pc and amanero" setup. Quite possible, I'm not being flip here... remember, I am out here in an Audio wilderness, no other systems to compare with and no other enthusiasts I know of any closer than Rick McI, 5-6 hours drive away. I have to work hard at being as objective and honest on what is good sound and what makes a difference in a vacuum... and maybe what I'm hearing and liking just isn't even in the same league as what you've got. Quite possible... and part of what has me on this investigation / experiment upgrade path right now.

I suspect at some point, when I think the EBay SD Player is sounding pretty good with battery power, that I'll need to schlep it over to Rick's place if he's amiable and we do a head-to-head with his probably un-recognizable by then SDTrans. Somewhat of a pain, but we will want to know how they really stack up against each other and if we have been able to turn the $85 wonder into a real contender.

Greg in Mississippi

[nige2000]

my first thoughts of it stock was "well it doesnt suck"
this board if fairly simple you could do all that in an evening

Good to know. I'm just curious how it compares to the SDTrans and want to have a good baseline.

the clocks on both are not of comparable quality id be very surprised if the sound is remotely close

Truth on the simplicity.... but "only in an evening" AFTER having all of the parts on-hand!

very true lucky here i can get a lot of stuff next day
delivery stuff from US and china takes quire a while

where did you source the 32khz ndk?

Mouser. I'll post a link tomorrow. I was looking at some various 8Mhz options too earlier, but put that aside to finish tomorrow.

ok ill let you test that one

That seems like a no-brainer to me, displays are noisy!

im sure providing a separate supply will be best, i was just thinking it would be another complication while carrying out other tests

whats the priority list ill rearrange it for you :)
this is starting to look real promising

Sorry, work and some high-school student mentoring is going to keep me heads down for the next week or so, maybe some progress after that?

im just eager to get a slightly modded comparison to the sdtrans

was reading your post on diya about the rpi been somewhat comparable to the sdtrans?
we did a bit of testing a couple yrs ago with lifepo4 powered rpi and beaglebone in conjunction with ians fifo reclocker and neither was as good as a modded pc and amanero with no fifo reclocker
the rpi was terrible without the fifo reclocker i2s signal is just bad

Hmmm... one of the other projects in-queue is an Ian reclocker setup feed by another Pi into a BuffII, mostly as an experiment on what makes a difference in the sound of that setup. So I'll have more perspective in the near future.

rest assure that ians reclocker will turn it inside out

I REALLY have to be clear on the SDTrans vs R-Pi into the DAM that the SDTrans is clearly better. But the magic of the DAM comes through on both AND I find the setup with the R-Pi input more inviting to listen to with more PRAT and toe-tapping-ness.

imagine the pi would have lacked much resolution and micro detail almost non-existant

Remember tho, my DAMs are basically stock right now, not even the standard V-Ref mods. So maybe things will change once I have improved the DAMs and I'll hear more of the R-Pi's warts.

are you using revision 1 dams stock? cause that would be a severe disadvantage

Also, that assessment has me questioning how good my SDTrans setup is. I'm slowly making plans to do an upgraded AC power setup for it along with a battery supply through the stock 5v input... to see if I can help it pull away.

And finally, maybe, just maybe all of my reference points are just not as good as your "modded pc and amanero" setup. Quite possible, I'm not being flip here... remember, I am out here in an Audio wilderness, no other systems to compare with and no other enthusiasts I know of any closer than Rick McI, 5-6 hours drive away. I have to work hard at being as objective and honest on what is good sound and what makes a difference in a vacuum... and maybe what I'm hearing and liking just isn't even in the same league as what you've got. Quite possible... and part of what has me on this investigation / experiment upgrade path right now.

qh that modded pc and amanero would have been rather basic maybe just lifepo4 powered pico psu and wouldnt be remotely in same league as the moddedpcs weve been using this past year or so

I suspect at some point, when I think the EBay SD Player is sounding pretty good with battery power, that I'll need to schlep it over to Rick's place if he's amiable and we do a head-to-head with his probably un-recognizable by then SDTrans. Somewhat of a pain, but we will want to know how they really stack up against each other and if we have been able to turn the $85 wonder into a real contender.

theres much to be said for simplicity, if we power it well give it good clocks it should be a contender

[nige2000]

EDIT: A couple more questions/comments on the EBAY SD Card Player....

Nige, does this thing actually take the stock power feed first through an 8v regulator, then through a 5v one before sending it through two 3.3v regulators, one for the processing and display side, the other for the audio signal generation and output? Seems like a lazy and probably not to effective of a way to do it, but then I am not an EE

yea bridge to 8 to 5 to two 3.3 but doesnt matter we not using it
i think the display/control 3.3 reg is doing some critical work too if i remember correct
.

Nige, also, in the pictures, it looks like you still have the AC power input hooked up. Do you still need this even with the battery power feeds for some reason I haven't seen yet or have you just not taken the time to remove those yet?

no its not needed i never took the wires out from when i tested it worked and figured out how to operate it

i ordered a second unit from aliexpress so i can examine it a bit easier and try figure out what exactly the ak4118 chip does or is it there for spdif?
there is also a simpler unit with i2s only with no ak4118 but it has no audio freq clocks on it which kinda has alarm bells ringing a bit
curiosity killed the cat and ordered one of them aswell

[gstew]

Nige,

I have some familiarity with the AK4118 via the AK4114 on the ESI Juli@, which is basically the earlier version of the same chip. It does both the S/PDIF and the I2S signal generation. I took a glance at the AK4118 datasheet and suspect that the S/PDIF could be shut down, but it depends on whether the chip is controlled via parallel interface (pins set high or low) or serial (register settings). Sorry, I haven't had time to go look at it more, but I doubt you'd get ANY output with the chip removed... that's how it worked on the Juli@ (and yup, I tried that!).

On simplicity, I would argue that the SDTrans is a simpler setup even with the greater number of chips and power supplies. Each of its chips are very single purposed and most are only as complex as they need to be. OTOH, the AK4118 on the EBay player is a multi-purpose monstrosity. On the SDTrans, you can cut the power to the S/PDIF and the I2S-over-LVDS HDMI outputs and those chips aren't making any power supply noise anymore. On the EBay player, we MIGHT be able to turn off S/PDIF in the AK4118, but I suspect it will still be doing other background stuff still.

And all of those power insertion points on the SDTrans means you can choose to go gonzo on multiple power feeds along with cutting off things you don't need (as Rick has done!).

Also the EBay player has to have more complex SW as it handles a greater number of file types... SDTrans' SW is a bit like MQN, very single purposed (of course the add of DSD did make it a bit more complex). And I suspect that ARM chip + memory is a much more electrically noiser setup than the fairly basic FPGA on the SDTrans.

Still, I don't know how much any of that will really matter sonically when the EBay player is tweaked out like yours. Perceived sound is the ultimate arbiter.

On the R-Pi into the DAM, I suspect it works as well as it does due to the isolation and reclocking that the DAM does... the Pi would NOT be a good candidate for a direct connection like you do with your tweaked USB interface and the EBay player.

BTW, I'm watching what you're doing with the direct-clocking very closely and that's now on my mod-list... but based on your comments on the DAM main thread, I'm now wondering if I should leave these V1's alone, sell them, and then get either the latest dam1021 or the OEM version. Is the V1 with mods as good as the V3 or is the V3 still ahead after both have been equally mod'd?

BTW, on reclocking and isolation not being the perfect cure-all that the "bits are bits, DBT it" crowd insists, see this very learned analysis by a great audio engineer who ALWAYS uses isolation and reclocking:

http://www.computeraudiophile.com/f27-u ... post411906

I've heard some others suggest similar, but never as clearly and completely... and as damning.

This is the 32kHz NDK I was planning to get for the EBay player: http://www.digikey.com/scripts/DkSearch ... 0127673765

Any good suggestions on the 8mHz one?

Starting to get back on the stereo stuff after today, should get the clocks ordered this week.

Later!

Greg in Mississippi

[nige2000]

Nige,

I have some familiarity with the AK4118 via the AK4114 on the ESI Juli@, which is basically the earlier version of the same chip. It does both the S/PDIF and the I2S signal generation. I took a glance at the AK4118 datasheet and suspect that the S/PDIF could be shut down, but it depends on whether the chip is controlled via parallel interface (pins set high or low) or serial (register settings). Sorry, I haven't had time to go look at it more, but I doubt you'd get ANY output with the chip removed... that's how it worked on the Juli@ (and yup, I tried that!).

i was a bit eager to get sound out of it i havent done much investigation or probing but i did do a little experimentation on the i2s line outs that suggest it will not work without the ak4118
needs further investigation when/if the other sdcard player arrive ill try and figure it

see here this simpler board i was thinking i could feed into a f-f to realign the i2s to the dam master fixed clock bypassing the ak4118?

On simplicity, I would argue that the SDTrans is a simpler setup even with the greater number of chips and power supplies. Each of its chips are very single purposed and most are only as complex as they need to be. OTOH, the AK4118 on the EBay player is a multi-purpose monstrosity. On the SDTrans, you can cut the power to the S/PDIF and the I2S-over-LVDS HDMI outputs and those chips aren't making any power supply noise anymore. On the EBay player, we MIGHT be able to turn off S/PDIF in the AK4118, but I suspect it will still be doing other background stuff still.

well wav sound better than flac so less processing is better

And all of those power insertion points on the SDTrans means you can choose to go gonzo on multiple power feeds along with cutting off things you don't need (as Rick has done!).

i doubt any of these chips are going to even take a breath out of a lifepo4 however you can still get localisations of noise which may be cured by multiple power routing from cell but the reference will always be multiple cells

Also the EBay player has to have more complex SW as it handles a greater number of file types... SDTrans' SW is a bit like MQN, very single purposed (of course the add of DSD did make it a bit more complex). And I suspect that ARM chip + memory is a much more electrically noiser setup than the fairly basic FPGA on the SDTrans.

Still, I don't know how much any of that will really matter sonically when the EBay player is tweaked out like yours. Perceived sound is the ultimate arbiter.

stock there should be no comparison however when both modded they should be much closer
would be great if we could do a comparison

On the R-Pi into the DAM, I suspect it works as well as it does due to the isolation and reclocking that the DAM does... the Pi would NOT be a good candidate for a direct connection like you do with your tweaked USB interface and the EBay player.

iso and fpga reclock should bring it up to a certain standard but is still far from optimum

BTW, I'm watching what you're doing with the direct-clocking very closely and that's now on my mod-list... but based on your comments on the DAM main thread, I'm now wondering if I should leave these V1's alone, sell them, and then get either the latest dam1021 or the OEM version. Is the V1 with mods as good as the V3 or is the V3 still ahead after both have been equally mod'd?

r1 vs r3 ones on the sd card with fixed clock and the other is with my usb-i2s unit and adjustable clock otherwise more or less identical so i can only really make an educated guess that they are more or less the same when modded and remaining differences are due to the fixed clock and sdcard player

BTW, on reclocking and isolation not being the perfect cure-all that the "bits are bits, DBT it" crowd insists, see this very learned analysis by a great audio engineer who ALWAYS uses isolation and reclocking:

http://www.computeraudiophile.com/f27-u ... post411906

I've heard some others suggest similar, but never as clearly and completely... and as damning.

nice to hear an explanation any experimentation i did with i2s isolators and batteries suggested they were at best pointless

This is the 32kHz NDK I was planning to get for the EBay player: http://www.digikey.com/scripts/DkSearch ... 0127673765

Any good suggestions on the 8mHz one?

think i just ordered one with reasonable freq stability
http://ie.farnell.com/webapp/wcs/stores ... reId=10163

have these here too have not tried
http://www.mouser.ie/Search/ProductDeta ... LMB-8-LY-T

Starting to get back on the stereo stuff after today, should get the clocks ordered this week.

Later!

Greg in Mississippi

[rickmcinnis]

Greg,

Are you saying that one can remove the power from the IIS and SPDIF rail and still get sound?

Have you actually tried this?

Or on second thought are you suggesting removing the chips for SPDIF and LVDS? That could be worthwhile.

Seems impossible to me.

[gstew]

i was a bit eager to get sound out of it i havent done much investigation or probing but i did do a little experimentation on the i2s line outs that suggest it will not work without the ak4118
needs further investigation when/if the other sdcard player arrive ill try and figure it

see here this simpler board i was thinking i could feed into a f-f to realign the i2s to the dam master fixed clock bypassing the ak4118?

On your current board, you MIGHT be able to tap the I2S prior to the AK4118, but I doubt it will work with the AK4118 disabled.

I found what looks like your in-coming board on Ebay, very curious how the clocking on that will work. I guess we'll have to wait until it arrives to find out.

i doubt any of these chips are going to even take a breath out of a lifepo4 however you can still get localisations of noise which may be cured by multiple power routing from cell but the reference will always be multiple cells...

stock there should be no comparison however when both modded they should be much closer...

would be great if we could do a comparison

Agreed

iso and fpga reclock should bring it up to a certain standard but is still far from optimum

Did you see Barrows' comments on the main Soekris thread where he mentions comments from "digital engineers who he respects"... I assume one of them is John Swenson who I referenced... but I'd been wondering if the FPGA-based reclock was as good as the HW setup on something like the Ian FIFO. Apparently not. I guess I'll have to try a setup with an R-Pi into an Ian FIFO and then into the DAM bypassing the DAM's isolators.

Another thought is that the isolators on the DAM are likely not the ultimate choice... John Swenson prefers the IL7xx GMR ones instead of the radio frequency Si ones Soren uses for lower jitter and radiated noise. I will have to swap in the IL equivalents on my Soekris to see what diff that makes.

nice to hear an explanation any experimentation i did with i2s isolators and batteries suggested they were at best pointless

Ted Smith, the designer of the PS Audio Directstream DACs has said as much too, just not as clearly and completely as John S. Ted also posted some interesting comments on the status of his current re-write of those DACs firmware discussing how is experimenting with various code configurations to lower the setup's jitter... and how he can see improvements in the jitter performance via simulation. Of course, he also says that the simulation is not reality and that the results have to be judged by listening. Refreshing to hear another highly-regarded digital audio engineer agree with some of the results us crazies have found!

r1 vs r3 ones on the sd card with fixed clock and the other is with my usb-i2s unit and adjustable clock otherwise more or less identical so i can only really make an educated guess that they are more or less the same when modded and remaining differences are due to the fixed clock and sdcard player

Whew. I'll keep my Rev 1 cards then.

think i just ordered one with reasonable freq stability
http://ie.farnell.com/webapp/wcs/stores ... reId=10163

have these here too have not tried
http://www.mouser.ie/Search/ProductDeta ... LMB-8-LY-T

Thx!

Greg in Mississippi

[gstew]

Greg,

Are you saying that one can remove the power from the IIS and SPDIF rail and still get sound?

Have you actually tried this?

Or on second thought are you suggesting removing the chips for SPDIF and LVDS? That could be worthwhile.

Seems impossible to me.

Sorry for not being clear here, Rick. What I was suggesting is that you could un-power the CS8406 S/PDIF output and DS90LV LVDS I2S-over-HDMI chips to reduce the noise they create just being there. I'm not sure if you can do it by removing the nearby noise filters (labeled NF10, NF11, and NF12? It looks as though they are in series with the power to the chips) or if you have to cut traces or just remove the chips entirely. You'd still need to keep power at the CN7 connection point to power the LC245A (via NF9).

On the direct I2S output, the LC245A chip looks to just be a transceiver chip that offers some drive and isolation (but not isolator-level isolation) on the I2S outputs. You COULD try taking the I2S from the inputs to that chip, but no guarantee that it'd be better... not sure if the FPGA direct output has much drive level.

I do need to confirm it does not do any reclocking. Based on a good read of the datasheet and a closeup picture I have here, it looks like it can't and it doesn't. But I'll have to check some connections on mine to be sure.

Greg in Mississippi