Something cool for Raspberry Pi/ODROID: I2S/DSD isolator HAT with native DSD decoder

Having, then, mistakenly bought the DoP decoder board :confused:, what are the currently available DSD-capable DACs, in HAT form for the RPi?

Sometimes it is difficult to determine what is actually currently available from another country, even with the wonders of Google Translate.

I find it challenging to know for sure whether something like the TerraBerry HAT from Japan is actually available for shipment to the U.S. much less actually DSD compatible and with a proper driver available for the specific software I wish to use (Moode 3.7).

Ditto the super cool looking TauDAC from Germany, but is this vaporware? My attempts to contact the developer via their web contact form have gone unanswered.

In my case I think Ian's forthcoming (soon?) ES9018 board is probably my best bet, or maybe the previously announced but seemingly dormant Soekris DAM1231 project will pan out. Allo also seems a likely candidate to potentially offer such a DAC HAT, and I'd certainly love to see Mamboberry take a stab at it.

According to Tim @ Moode, the driver availability is no trivial piece to this. So for instance would Ian's forthcoming DAC board likely work with an existing ES9018 driver such as the one for Twisted Pear?
 
Member
Joined 2003
Paid Member
@TioFrancotirador : Thanks for your advices. I don't think I will use Pi zero. It is probably better than my RPi 2 in terms of noise. But it doesn't have ethernet connector, which I think is a must for stable streaming HiRes files.

About skipping 3.3V from LT3042 to AVDD of 502DAC and powering it directly from Salas Reflektor D, it is very good idea. But, by doing so, we will make a trade off between local regulator (better signal road, low res) with better external regulator (assuming Salas Ref D is better than LT3042 PSU). Personally, if local regulator is good enough, and LT3042 is a really good one, I won't change it.

@Greg Stewart
You should be careful when trying to feed 3.3V to RPi via GPIO. I think I have read somewhere that you shouldn't put too much current through 3.3V GPIO, otherwise it will damage RPi.

trung224,

Sorry for the slow response, I'd forgotten I wanted to make a couple of comments.

My thoughts on local LT3045 versus remote Salas are basically the same as yours... but a very short-connected set of Salas Micro Reflektors might be a good tradeoff. I hope to try this at some point.

Also I agree on the Pi Zero... I'd be MUCH more inclined to use it if it had on-board wired ethernet. I've stuck with RPi 2Bs to avoid the 3's WiFi.

OTOH, I think the caution on feeding the 3.3V via the expansion header had to do with the limitations of the on-board switch-mode regulator. After removing it and replacing it with a linear regulator, it shouldn't matter FROM A CURRENT LIMITATION perspective if you feed it into the insertion point you identified OR the expansion connector pins. OTOH, it is probably theoretically 'better' to use your insertion point, as that is closer to the 3.3V power users. I was just considering the expansion pins as an easy way to get the 3.3V into the board... but since there are NO 1.8V expansion pins, I might as well feed both directly into the board.

Greg in Mississippi
 
Member
Joined 2003
Paid Member
Sometimes it is difficult to determine what is actually currently available from another country, even with the wonders of Google Translate.

I find it challenging to know for sure whether something like the TerraBerry HAT from Japan is actually available for shipment to the U.S. much less actually DSD compatible and with a proper driver available for the specific software I wish to use (Moode 3.7).

Ditto the super cool looking TauDAC from Germany, but is this vaporware? My attempts to contact the developer via their web contact form have gone unanswered.

In my case I think Ian's forthcoming (soon?) ES9018 board is probably my best bet, or maybe the previously announced but seemingly dormant Soekris DAM1231 project will pan out. Allo also seems a likely candidate to potentially offer such a DAC HAT, and I'd certainly love to see Mamboberry take a stab at it.

According to Tim @ Moode, the driver availability is no trivial piece to this. So for instance would Ian's forthcoming DAC board likely work with an existing ES9018 driver such as the one for Twisted Pear?

MikeyFresh,

Fab detective work. Both look to be very interesting RPi DAC HATS!

On the Terra-Berry (the English page worked well for me)... while I like the idea of the AK4490, I DON"T like the idea of the AK4137 SRC. BUT it looks to be a well thought-out and engineered board & I bet it sounds good. It also will probably work best with an output stage, not just connecting the DAC output via coupling caps to RCAs as they appear to do with their RCA connector daughter board.

I have a couple of RPi HAT prototype boards here that I intend to use to connect an JLSounds AK4490 DAC to a Pi. I'll report how that goes when I get to it. I plan to use a couple of small OPA1641-based Balanced-to-Single-Ended converters as output stages... the dual version of that chip (OPA1642) is sounding very nice in the Dial DAC design.

The TauDAC-DM101 also looked very interesting. I do like the idea of the Lundahl transformers as outputs... that has worked well for others with similar DAC chip (and that is something I plan to try with my HotRod Dial DACs too).

One thing I didn't see on either (though I didn't look too hard) is cost. I expect both to be north of $200 USD... those Lundahl LL1588's are almost $200 a pair by themselves.

Others have pointed out some other interesting higher-end RPi DACs, one from Dion Audio and the other from SabreBerry. Not sure if either ever became widely available or are even still available.

I suspect there's a bunch of them out there that are good ideas, but haven't made it into the mainstream yet, have died out, or are still in gestation, like the Soekris R2R RPi DAC.

One gets into some issues with these IF one is trying to get better sound than the current crop of all under $100 USD RPi DAC HATS:

1. Board real-estate & power for all the goodies... I think Allo.com has done a great job packing a lot of good HW into their top-end DAC HATS, the Boss and the Piano 2.1. BUT if you want to go up to the next level of DAC chips, they will generally work best (or require) a a separate output stage, then you start to need more board space (like the stack-on boards used on both the Terra-Berry and TauDAC). AND/OR you need more power supplies, like the +-15V for the Dial RPi DAC. My experience with the Dial has me convinced that to get to the next level of SQ for these units, DAC chips that use separate output stages will be needed... one can only do so much with a built-in CMOS-based OpAmp with the negative rail fed from a charge pump as used in the PCM51xx/52xx and ES9022/9023 series. I'm really curious how Ian's ES9018-based HAT will turn out... I know he'll likely use an I/V stage (though the higher-end ESS chips CAN be used in voltage mode to good effect). BUT I don't think he plans to use anything other than the basic 5v nor do I think he's planning to use +- supplies, which will mean coupling caps. If I'm right on these, I wonder that stock setup will sound compared to it hacked with a +-15v I/V stage or FET buffer stage or transformer output?

Also, another thing that I believe helps elevate the Dial DAC is that it uses no ceramic capacitors, only electrolytics, mylars and polyprops. There is a 'something' to the sound of that DAC and my HotRod version of it that make them both sound more akin to my modified Sony HAP Z1-Es (which also uses no ceramic caps around the DACs and output stages) versus the standard crop of DAC HATS available currently. AND these again take more board space! And cost.

2. Power supplies. You've seen the picture of my current best RPi DAC setup with 3-4 LPS-1s & 3 more linear supplies. I'm powering a $250 or so stack of RPi + add-ons + DAC with nearly $2000 of supplies that take up a LOT of real-estate. Not acceptable for many... though I'm LOVING the sound I'm getting!

3. Those add-ons. Yup, both an isolator & a reclocker will be needed for the best sound. MAYBE Allo or Ian or someone else will release a single-HAT-compliant board that does both. But that's another 1-2 boards in the stack with more cost & more power supplies required.

4. Cost. All of these goodies add cost. I suspect the TauDAC with the Lundahl transformer output stage will come in about $400... for that it BETTER sound good. BUT it will also require good power supplies to work at its best. Maybe the low-end of these setups that take one above the PCM51xx/52xx / ES9022/9023 DACs is the Dial setup at about $210 USD for his DAC card plus a +-15V power supply and a 5V power supply. Compared to what I'm doing, that is not much. BUT I suspect that's above the high-end of what people will put into these setup... at least until more hear what can really be done with them.

When you add all these up, IF you want to go above the SQ of a standard 5V powered RPi DAC, maybe you are better off doing a RPi -> isolator -> reclocker stack and using something like Ian's RPi I2S board to feed to whatever DIY DAC you want to use. I'm doing something lilke this now for a Soekris DAM DAC... and will be adding more when I get bench time to finish them. Bigger, more expensive, BUT you can take them farther because you are not as board-space limited.

I don't have an answer here nor any prediction of how this marketplace will play out.

Anyone got any better thoughts? AND should we take this all to a separate thread and leave poor Ian's thread alone?

Greg in Mississippi

P.S. MikeyFresh, I forgot to mention that the TP Buffalo DACs have no specific RPi drivers, the DAC configuration is handled by an on-board microprocessor... if you connect one to an RPi (whether directly or through an isolator and/or a reclocker), you just use the basic I2S output driver that's on all of the current RPi audio distros.
 
Last edited:
MikeyFresh,

Fab detective work. Both look to be very interesting RPi DAC HATS!

On the Terra-Berry (the English page worked well for me)... while I like the idea of the AK4490, I DON"T like the idea of the AK4137 SRC. BUT it looks to be a well thought-out and engineered board & I bet it sounds good. It also will probably work best with an output stage, not just connecting the DAC output via coupling caps to RCAs as they appear to do with their RCA connector daughter board.

I have a couple of RPi HAT prototype boards here that I intend to use to connect an JLSounds AK4490 DAC to a Pi. I'll report how that goes when I get to it. I plan to use a couple of small OPA1641-based Balanced-to-Single-Ended converters as output stages... the dual version of that chip (OPA1642) is sounding very nice in the Dial DAC design.

The TauDAC-DM101 also looked very interesting. I do like the idea of the Lundahl transformers as outputs... that has worked well for others with similar DAC chip (and that is something I plan to try with my HotRod Dial DACs too).

One thing I didn't see on either (though I didn't look too hard) is cost. I expect both to be north of $200 USD... those Lundahl LL1588's are almost $200 a pair by themselves.

Others have pointed out some other interesting higher-end RPi DACs, one from Dion Audio and the other from SabreBerry. Not sure if either ever became widely available or are even still available.

I suspect there's a bunch of them out there that are good ideas, but haven't made it into the mainstream yet, have died out, or are still in gestation, like the Soekris R2R RPi DAC.

One gets into some issues with these IF one is trying to get better sound than the current crop of all under $100 USD RPi DAC HATS:

1. Board real-estate & power for all the goodies... I think Allo.com has done a great job packing a lot of good HW into their top-end DAC HATS, the Boss and the Piano 2.1. BUT if you want to go up to the next level of DAC chips, they will generally work best (or require) a a separate output stage, then you start to need more board space (like the stack-on boards used on both the Terra-Berry and TauDAC). AND/OR you need more power supplies, like the +-15V for the Dial RPi DAC. My experience with the Dial has me convinced that to get to the next level of SQ for these units, DAC chips that use separate output stages will be needed... one can only do so much with a built-in CMOS-based OpAmp with the negative rail fed from a charge pump as used in the PCM51xx/52xx and ES9022/9023 series. I'm really curious how Ian's ES9018-based HAT will turn out... I know he'll likely use an I/V stage (though the higher-end ESS chips CAN be used in voltage mode to good effect). BUT I don't think he plans to use anything other than the basic 5v nor do I think he's planning to use +- supplies, which will mean coupling caps. If I'm right on these, I wonder that stock setup will sound compared to it hacked with a +-15v I/V stage or FET buffer stage or transformer output?

Also, another thing that I believe helps elevate the Dial DAC is that it uses no ceramic capacitors, only electrolytics, mylars and polyprops. There is a 'something' to the sound of that DAC and my HotRod version of it that make them both sound more akin to my modified Sony HAP Z1-Es (which also uses no ceramic caps around the DACs and output stages) versus the standard crop of DAC HATS available currently. AND these again take more board space! And cost.

2. Power supplies. You've seen the picture of my current best RPi DAC setup with 3-4 LPS-1s & 3 more linear supplies. I'm powering a $250 or so stack of RPi + add-ons + DAC with nearly $2000 of supplies that take up a LOT of real-estate. Not acceptable for many... though I'm LOVING the sound I'm getting!

3. Those add-ons. Yup, both an isolator & a reclocker will be needed for the best sound. MAYBE Allo or Ian or someone else will release a single-HAT-compliant board that does both. But that's another 1-2 boards in the stack with more cost & more power supplies required.

4. Cost. All of these goodies add cost. I suspect the TauDAC with the Lundahl transformer output stage will come in about $400... for that it BETTER sound good. BUT it will also require good power supplies to work at its best. Maybe the low-end of these setups that take one above the PCM51xx/52xx / ES9022/9023 DACs is the Dial setup at about $210 USD for his DAC card plus a +-15V power supply and a 5V power supply. Compared to what I'm doing, that is not much. BUT I suspect that's above the high-end of what people will put into these setup... at least until more hear what can really be done with them.

When you add all these up, IF you want to go above the SQ of a standard 5V powered RPi DAC, maybe you are better off doing a RPi -> isolator -> reclocker stack and using something like Ian's RPi I2S board to feed to whatever DIY DAC you want to use. I'm doing something lilke this now for a Soekris DAM DAC... and will be adding more when I get bench time to finish them. Bigger, more expensive, BUT you can take them farther because you are not as board-space limited.

I don't have an answer here nor any prediction of how this marketplace will play out.

Anyone got any better thoughts? AND should we take this all to a separate thread and leave poor Ian's thread alone?

Greg in Mississippi

P.S. MikeyFresh, I forgot to mention that the TP Buffalo DACs have no specific RPi drivers, the DAC configuration is handled by an on-board microprocessor... if you connect one to an RPi (whether directly or through an isolator and/or a reclocker), you just use the basic I2S output driver that's on all of the current RPi audio distros.

1. I like Ian's es9018k2m (hopefully it will be es9028q2m), because it has direct differential output from dac chip for external output stage. So I guess in the future I/V HAT can be connected to it or transformers :)
2. LPS-1 is good, but in the end it is limited to TPS7A4700 perferomance in the end since it is used for regulations. Try Salas Reflector D mini (it has FET mod included). It muuuch cheaper and should be even better.
 
Member
Joined 2003
Paid Member
1. I like Ian's es9018k2m (hopefully it will be es9028q2m), because it has direct differential output from dac chip for external output stage. So I guess in the future I/V HAT can be connected to it or transformers :)
2. LPS-1 is good, but in the end it is limited to TPS7A4700 perferomance in the end since it is used for regulations. Try Salas Reflector D mini (it has FET mod included). It muuuch cheaper and should be even better.

@TioFrancotirador,

I totally agree on Ian's ES90xx board and like that it has both those take-off points for external outputs & alternate clock inputs (note the u.fl in the picture). Already thinking of options...

On the LPS-1, I'm building up some Reflektor D Minis this week and already have appropriate raw DC available, so will be able to share some comparisons in the next few weeks. BUT one of the main benefits of the LPS-1s are that they are isolated from the AC line which appears to eliminate some very pernicious sources of noise in computer-based setups.

So far, they have won in EVERY comparison I've done for post-digital source usage... DACs, isolators, reclockers. Results are more mixed with powering my digital sources (RPis, SDTrans384) where the mix of tradeoffs has me powering the Pis with AC-connected linear supplies (they are the main reason I'm building some 5V Mini Reflektors) while I still prefer the SDTrans with an LPS-1. I've got more comparisons to do here, but the LPS-1 DOES improve over any AC-connected supplies I've tried for some applications.

And of course, one set of trials WILL be using 2 LPS-1s in series to provide sufficient voltage headroom to power one of the 5V Mini Reflektors. Should be interesting!

Later!

Greg in Mississippi

P.S. I posted some comparison results for various LPS-1 setups versus a good linear supply in this thread on Computer Audiophile awhile back:

https://www.computeraudiophile.com/...ence-to-output/?do=findComment&comment=618268
 
Member
Joined 2003
Paid Member
:)
I´ve received an info from the TauDAC producer
TauDAC-DM101: 399,00 €
TauOUT-TP101: 249,00 €

@dkdkdk,

Thanks for that update. WOW, I was off by 100%!

I think those prices are reasonable given what they have there. I likely won't be buying one, but hope many do!

I wonder about their decision to feed the Pi with the power for the DAC board... I haven't heard a situation where powering the Pi separately doesn't improve things. This DAC certainly deserves that!

Later!

Greg in Mississippi
 
At first I was looking onto that prices in desbelief cause of my monthly bombing emails requesting info on how the work on DAC is going on :rolleyes:.
Unfortunately the price for the DAC+OUT is too high for me, especially with consideration how much was the streamer (RPi) and what others (Allo, Audiophonics...) want for their HATs.
But I´m definitely not saying that TauDAC can´t outperform standalone DACs with the same price. Just don´t want to try it myself :eek:
 
:)
I´ve received an info from the TauDAC producer
TauDAC-DM101: 399,00 €
TauOUT-TP101: 249,00 €

My initial observation on this unit had it estimated at around $700, so it looks to still be about that much many months later, with some potential give or take for currency fluctuation.

Expensive in the RPi realm, but not crazy given the BOM, as Greg Stewart pointed out just those output transformers are an expensive part.

Was there any estimate/timeline on actual availability?

Did you use their website form to contact them?
 
@TioFrancotirador,

I totally agree on Ian's ES90xx board and like that it has both those take-off points for external outputs & alternate clock inputs (note the u.fl in the picture). Already thinking of options...

On the LPS-1, I'm building up some Reflektor D Minis this week and already have appropriate raw DC available, so will be able to share some comparisons in the next few weeks. BUT one of the main benefits of the LPS-1s are that they are isolated from the AC line which appears to eliminate some very pernicious sources of noise in computer-based setups.

So far, they have won in EVERY comparison I've done for post-digital source usage... DACs, isolators, reclockers. Results are more mixed with powering my digital sources (RPis, SDTrans384) where the mix of tradeoffs has me powering the Pis with AC-connected linear supplies (they are the main reason I'm building some 5V Mini Reflektors) while I still prefer the SDTrans with an LPS-1. I've got more comparisons to do here, but the LPS-1 DOES improve over any AC-connected supplies I've tried for some applications.

And of course, one set of trials WILL be using 2 LPS-1s in series to provide sufficient voltage headroom to power one of the 5V Mini Reflektors. Should be interesting!

Later!

Greg in Mississippi

P.S. I posted some comparison results for various LPS-1 setups versus a good linear supply in this thread on Computer Audiophile awhile back:

https://www.computeraudiophile.com/...ence-to-output/?do=findComment&comment=618268

Greg, that is great! I am wonder your results.
Keep in mind that Ref D was created to fight also AC noise.
You should have at least 5V-8V difference between DCVin and DCVout.
So for 5Vout you should use 10VAC trafo, unless you will use Cree SiC diodes for rectification then you can stay with 9VAC (they drop less) and do not have to use snubber :) Another option is to use CLCLC. Some say it is the best.
As for the trafo avoid toroids and use split bobbin EI or R-Core instead. In those trafos primary and secondary are totally separated. In most linear PSUs toroids are used. As a rule of thumb you trafo should be at least 3 times stronger that the current you output.
Also overall Ref-D performance is dependent on C2 which filters noise from LEDs. For digital circuits I would suggest 1000uf/16V FPCAP nichicon R7. It has the least noise and filters out high frequencies very well.
BTW The best way to power Reflector D with LPS-1 would to skip onboard TPS7A4700.
 
Last edited:
Member
Joined 2003
Paid Member
My initial observation on this unit had it estimated at around $700, so it looks to still be about that much many months later, with some potential give or take for currency fluctuation.

Expensive in the RPi realm, but not crazy given the BOM, as Greg Stewart pointed out just those output transformers are an expensive part.

Was there any estimate/timeline on actual availability?

Did you use their website form to contact them?

Its gotta have a pretty expensive BOM... those WM8741 DACs are $12 each in quantity of 2000 versus $3.26 for the PCM1795 in the Sony HAP Z1-ES at the same quantity. The clocks on board appear to be the high-bux Crysteks. I suspect it does local reclocking, but hard to tell... DANG, I need a closeup of the board top and bottom!

And of course the $200 USD of transformers for the outputs.

When you consider some of the BOM costs, then compare it to a Twisted Pear Audio Buffalo-IIISE at $379 USD for just the bare board, AND understand the base cost of the transformers, those prices are not bad. BUT for me, the lack of flexibility in upgrading the power scheme (no separate power, only one power input) means I'm likely not interested AT THAT PRICE. Power is EVERYTHING for the best implementations... If I'm spending that type of $, I want to make it the best I can. As far as I can tell, I'd have to hack it somewhat seriously to do that. I'm ok to do that with a $45 HiFiBerry DAC+Pro. Or build up HotRod Dial DACs from raw boards with my own power feed scheme. The TP DACs have a great deal of power flexibility built-in, so one can optimize to their heart's (& ear's) content.

From what I see so far on the TauDAC, not so much. BUT maybe, of course, they have done such a good job it isn't needed... which IS possible. Need more details!

Just some thoughts.

Greg in Mississippi
 
Last edited:
When you consider some of the BOM costs, then compare it to a Twisted Pear Audio Buffalo-IIISE at $379 USD for just the bare board, AND understand the base cost of the transformers, those prices are not bad.

Not quite. I've had one for quite a while, and it wouldn't have gotten up and running if it had been a bare board that I bought. ;) The Buffs since version 1 have been assembled and tested.
 
Ian, So my replacement works!

Started comparing with a simple i2s source, a raw hifiberry digi+pro (no on board regulation, at least with my beta board).

It's surprisingly easy to hear the improvement even on headphones! It's digging down into the noise, bringing out subtlety. Easier to get into the volume, nice varied leading edge.

Learning that you are not isolating/mirroring all GPIO's. For example this digi+pro relies on GPIO5 (44.1x) and 6 (48x) to select the crystal -- so I set piCorePlayer to sample everything to 48. (It passes thru GPIO6)

I don't plan on using the digi+pro, but thought the digi+pro would be a pretty popular board to use with this...

Any thoughts?

Thanks!!!

PS. I'll send you that half broken board this week.
 

Attachments

  • IMG_0038.jpg
    IMG_0038.jpg
    824.4 KB · Views: 711
  • IMG_0040.jpg
    IMG_0040.jpg
    853.9 KB · Views: 684
Last edited: