Compatible I2S source devices


Hi Patrick,

I´m curious about your route, I few months ago I did post here some ideas about using a interfeca like yours and someone ask me why I will use too many conversion to send I2s, so today I’m using RPI 2 with Allo Reclocker and HDMI output to DS, also using some regulators as you and I can tell you it’s sound awesome, but maybe I’m missing something, could you please tell me your findings?




Hi Alexandre, yes I remember well, don’t know if it was me or not (it’s possible yesblush_gif), who argumented on reducing the number of transformations so targeting a more straightful I2S chain. So far, at the moment the Kali Reclocker was limited and had some problems if I remember well. We was waiting for a solution from Audiophonics (which come far far later as expected) and from Allo. Seems now we have both. But the time being, I decided to test the Armature Hecate board and was stunned by the result so I decided finally to build something around.

Besides that, a point which had its weight in choosing the Hecate was its galvanic isolation we do not retrieve on the Kali boards (I hope to be right as I did not look at their new spec recently) and the fact I had the project to power up the I2S/HDMI module with 2 (small) banks of SuperCaps.


How is this solution better than the Mano Music Streamer?


@Speed Racer I can’t say if it is better or lesser. And it sounds to me it is not easy to provide an answer if we just compare what we see on displayed spec sheet while in some case a less featured solution can sound better for lots of reasons (quality of cables, case agencement, engineering …).

So, just to remind, it is a DIY project, not industrialized, I tried to put the best components I can (I guess I used some times a mass to crush a fly) in an empirical context (no precise measurement tools to confirm the choices). The price can be the engine which lead to go on the DIY way instead of buying an “on the shelf” (and improved) solution, but it is even not proven that a DIY is cheaper at equal feature (surely be most customizable).

I don’t know which interface the ManoStreamer is using but now what I see as differences, is:

  • ManoStreamer use a galvanic isolation for its LAN connection, and it is may be important, I just went to plug the 12V of NanoPat external PSU to my ethernet switch and I feel that there’s a gain, a more relaxed presentation^^. 1 point for the ManoStreamer

  • The NanoPat has an external PSU in charge to perform a first regulation stage (with all radiating components). The streamer box embeds only the 2 last regulation stages chosen for their ultra lower noise characteristics.

  • The NanoPat, on paper, cost me a bit less if we do not count the time (just sum the components of the streamer, plus those in the external PSU with two 220V/36VA and 220V/8VA transformers + 2 regulation cards at twenty Euros approximatively).

  • Don’t know if the Mano streamer is isolated on its output, The Hecate board has an isolation.

  • The NanoPat is built by me, it’s my child now, I love it. +1 point for him wink

Patrick said - The NanoPat is built by me, it's my child now, I love it. +1 point for him wink
I'd like to adjust, it's not only my baby as I rely a lot on the Audiophile community.

I thanked previously PSAudio, Paul, Ted and all other participants of this long thread (Alex and lot others).

Something on which I’m not totally at my ease as I’m wondering how a such DIY thread can, in an certain manner, cause harm to PSAudio? I mean, may be less LAN Rover or PSAudio Bridges sold ?? I don’t know how Paul considers this point and sometimes I say to myself that this I2S thread should be closed by the moderator, how many “bad” ideas are transmitted that lead to Not buy a PSAudio product ?? May be not so much, but…
Anyway its a pleasure, to exchange with open minded people of PSAudio.


Patrick, you’ve done a wonderful job of documenting your journey! Not sure I followed everything, but it has been fascinating to see the extent to which you have gone to solve your “problem”. I don’t think you have any need to worry about PS Audio, Paul has been very open in his interest in what other tinkerers are doing with his products or those that interface with them. I think it is highly unlikely that your efforts will put a significant dent into PS Audio sales since the knowledge required to do this as well as the effort would eliminate a good portion of the folks here. Personally, I might have been interested in something like this 20-30 years ago, but now I’m more interested in the off the shelf solution. It’s reassuring to see their are folks out there who have the ability to think and create!


@pmotz thank you for the encouragements. Actually I thought about all the DIY solutions presented here, not only mine, but you’re right the amount of work, for a non professional, to come up to a functional solution is huge.

I’m discovering the few discussions on Octave and Bridge 3, sure that PSAudio is preparing something with a very high level far in advance on all we can try to produce at the DIY level cant-believe-my-eyes-smiley-emoticon_gif


@patrick thank you for sharing the journey in 1st place! I have just ordered the same regulators as you are using. I have 5v 1,5A linear PSU which is intended to feed Hecate 5v and 3,3v via low noise regulators from the same 5v output. Is this OK or you have experience that 3,3v requires (or in other words really benefits) dedicated psu even with the ultra low noise regulators you are using?

For the 5V regulator input are you using 5V PSU or 6+V psu?

I have to decide by tomorrow which (if any) linear psu to order to finish (without the casing) my i2s streamer for DS DAC so i would really appreciate your experience here :slight_smile:

thank you!

P.S. i have noticed big capacitor in parallel with input to 3,3 regulator - what specs is it in what is it’s function?


@ Maniac Sorry for the late answer, I was on the road. Hope it is not too late for you.

If think it’s a pity to feed the Hecate 5V and the 3,3V module with the same PSU. This because, the Hecate board put in the game many care to isolate its output (I2S) from its Input (USB). They use some top notch galvanic (magnetic cells) isolation and a full dedicated 3,3V. The small black box on the board produce a 3,3V from the 5V with a separated ground ((it seems to be this component). All numeric signals are isolated too. Thus is you re-use the same 5V PSU to power both the USB 5V of the Hecate and the I2S/HDMI module (even through the 3.3V low noise regulator), you’ll share the same ground for both. The HDMI module must be grounded with the 3,3V of the I2S output of Hecate. Don’t know if I’m clear, sorry.

If you can, use your 1,5A Linear PSU to power only the Hecate board on its USB input and another 5V PSU (100mA is sufficient) to feed the 3,3V LT3045 you bought. The LT3045 accepts up to 20V. The LT3045 is a Low Voltage Drop Out, only 300 or 500mV, so a 5V in input should be sufficient.

Here are the options, you have, and, to understand, keep in mind the I2S ground of the DirectStream input is connected to the I2S/LVDS ground of the HDMI module, then to the I2S ground of the Hecate output:

  1. Re-use the 5V PSU you use for the Hecate Input. As just said before, you’ll link all the grounds everywhere, so, you lose all the benefits of the isolation implemented on the Hecate board.

  2. Put a new transformer + a diodes bridge + a cap (to filter the 100Hz sine ondulations). The LT3045 card will regulate to 3,3V as expected. The USB ground and the I2S ground are no longer linked (like in previous option). But in some manner, I’m wondering if the fact to use a transformer will not bring some noise to the DAC (don’t forget the DAC’s I2S ground input is connected at the end). On this point it’s not clear in my mind, as I said I’m not a god in electronic, especially about EMI and so on, sorry.

  3. Use a battery (6V, 9V or even 12V) to power the LT3045 3,3V regulation board that will power the I2S/LVDS HDMI module. The battery’s ground is, as well, connected to the isolated 3,3V of the Hecate. By this way, there’s no ground loop possible, the ground takes end at he Hecate’s I2S output level. This sounds a good option, the battery is not heavily loaded (only 30mA) but it is not the most convenient, I mus admit.

  4. Use 2 banks of SuperCaps. The first is “energized” by a PSU, while the second is used as PSU to feed the LT3045 3,3V and then the I2S/LVDS HDMI module. You need a load-balancing mechanism to switch from 1 bank and then to the other bank when this last is discharged. This option seems the best, and should be simpler to implement than a LPS1 UltraCaps from Uptone Audio. The 3,3V/30mA we need should not require big SuperCaps and working at 3,3V is an advantage. It is easy to find affordable SuperCaps at 5V, while higher voltages are more difficult to find or setup. Using SuperCaps instead of a battery (option 3) is a proven advantage in audio if you look at the feedbacks on the LPS1 UltraCaps. Many UltraCaps have a very low internal resistance compared to the internal resistance of a battery.

The NanoPat is, for the moment, built on the Option 2 design, although my initial aim was to build the Option 4 dull_gif. I’ve put a low ESR audiophile capacitor 3900uF 16V from Panasonic (the one you noticed on the picture) in order to filter the last stage. I believe the LT3045 has its own filtering cap, so I don’t know if the Panasonic brings something or not. In my open questions in a previous post, I still have the interrogation about the importance to power the Hecate’s input (as well as the SBC that feed the Hecate with data) with as much care as we take for the HDMI module…

You didn’t talked about the SBC or computer you use and the USB cable between it and the Hecate board. How do you separate the 5V of the computer to the 5V to feed the Hecate ? wink

EDIT: NanoPat is built on Option 2 (not 3)


Thank you Patrick, the PSU i have today have two separate outputs - 5V and 12V. I use 12V to power switch in my audio chain and 5V to power Rpi3 and Hecate - i have made custom short cable for PSU to be able to connect up to 3 USB cables to PSU.

I also have JCAT USB splitted cable connected on one end to Hecate and other two ends as following : data connected to RPI (via AQ Jitterbug) and power end to the same 5V PSU as RPI3.

Theoretical problem is the PSU is only 1,5A while rpi3 can consume up to 2,5A, but in reality it seems that rpi running only RoonBridge is not consuming much as it works stable in this setup :slight_smile:

Other theoretical problem i am afraid of is that Rpi and hecate are fed from the same PSU thus ground pollution might affect Hecate input.

My friend is going to buy something from audiophonics and i want to use this opportunity to buy one more PSU (as postage from Audiophonics is about 20€ here to Slovakia).

The options for new PSU i see are (current PSU will be used for 3,3V I2S only as you have suggested):

  1. 5V PSU 3,5A - would power both RPI and Hecate USB

  2. 2x5V 1,5A - one rail would power Rpi and other Hecate USB

I would be happier with option 1 as it have enough power to feed rpi even under heaviest load and not overloading the psu, but i am not sure if this shared power between Rpi and Hecate would not negative impact on the Hecate SQ? What is your experience please?

Option2 would allow to have dedicated PSU for all components however the rpi3 would still run from 1,5A PSU as today making me nervous if it will not overload the PSU once the load (like some OS upgrade on Rpi etc.) would happen…


EDIT: just to add what i am trying to achieve - i am happily using Bridge2 as my main reference which is connected to ethernet via TP-Link optical transceivers and i am trying to build streamer which would have better SQ thank the Bridge2 - which may be pretty hard but i still hope :slight_smile: I suspect that having external streamer would benefit mainly by removing load done by B2 on the DS Dac internal PSU and removing EMI/RFI caused by B2 inside the DAC… Having abibility to stream DSD128 would be also nice benefit…


Maniac Sorry I think I can’t report a relevant feedback on SQ differences if you feed the RPI and the Hecate by two separates PSU versus both feed by a common linear PSU.

Actually, I tested the replacement of an IFI Audio iPower 5V/2,5A PSU by a basic switching 5V/2A PSU for the SBC (RPI3 or NanoPI) + a separate AQVOX 5V/1A dedicated to the Hecate. The difference was not obvious in this case. The real jump occurred when I upgraded to the NanoPat configuration with the 2 LT3045 regulators (thought, a common PSU for both SBC and Hecate). Besides that, there are too many differences coming with the NanoPat design in more than the LT3045 (my AQ Jitterbug removed, a new very short/silver USB cable, the self made separate triple PSU …).

However I believe the recommended 2,5A PSU for the RPI3 are overrated when we use it like us for an audio bridge purpose, without WIFI activated nor big eaters USB devices connected (see this page). The Hecate seems to not consume a lot. But, to be frank I would be nervous like you at each OS upgrade for example. And all measurements I took did not reveals the peaks…

Did you think about the option to build a double PSU with a 2x9V transformer and such regulation here ? You could feed both the RPI3 and the Hecate. On the datasheet of the module, they say to use a 3300uF filter caps which should be sufficient with a 9V AC as input (seems risky to go below 8V). You could keep your current PSU for the network switch and the HDMI output.

I’m curious to get your feedback wen you’ll complete your project (especially the comparaison with the BridgeII) happy-048_gif


@patrick I was out to mountains during new year so I was not able to reply sooner. So i ended up ordering two PSUs from Audiophonics - one 5v 3,5A (for RPI) and one dual 5v 1,25A which will be used for Hecate - one rail for 5V USB input and other rail for 3,3V GD Audio I2S interface. I was trying to save some money by buying just one PSU, but it seems really the best solution will be to have it separated. I am still waiting for delivery of the Ldvo regulator boards. The only issue I have is that now the cumulative cost of this solution is on level where I could buy ready made Singxer SU-1 USB :slight_smile:

Yesterday I was comparing the RPI3 + Jitterbug + Uptone USB Regen + Hecate via SPDIF and bridge2 (both streaming from Roon) - the bridge2 is still better, so I am now curious once the regulators for Hecate will be delivered and connected how this will end up :slight_smile:


so the regulators have finally arrived today, i have connected it all togehther and strange things happen :slight_smile:

i have desoldered original cables from gdaudio module and soldered new ones with pin connectors for HECATE i2s pinout.

i have connected 3,3V and GND (separate pcb holes in the gdaudio module to 3,3V regulator + and -)

then i have connected module with hecate with i2s data, bck, wck and gnd (have not connected MCK)


i2s input 1 on my ds dac does not detect any signal

i2s input detects native (and if i recall corrrectly also DoP) 64 dsd, however shows as dsd128 and plays only left channel . changing volume in roon to anything less than 100 makes noise instead of music.

pcm does not work at all on any i2s input

strange enough the native dsd still plays via left speaker even when i2s DATA connector is not plugged in or even I2S GND not plugged in, or MCK plugged in (i was trying to identify the problem thus disconnecting/connecting the pins)

bridge2 works fine

i am too tired now to troubleshoot further i just hope the i2s inputs on my dac are ok as i do not other i2s device to test their functionality

i have also tried to use 5V power source without the 5V regulator in a way but no change)

the spdif output plays fine

i have tried ropieee and also dietpi with roonbridge

does the behavior say anything where the problem could be?


Does your source hook up the 5V to the HDMI connector? If so you are fighting the 5V power supply in the DS… If yours power supply is just a tidbit higher voltage than the DS then yours will be trying to power the DS. If your new supply possibly sinks current as well as provides it, you may be dissipating any extra current from the DS’s 5V supply.

maniac said

i have also tried to use 5V power source without the 5V regulator in a way but no change)

Hi Mania, I do not well understand what you mean. The AudioGD HDMI module must be powered by a 3,3V source, so do you mean you tried to feed it without your new 3,3V regulator and you applied a 5V on the HDMI module ???

The first time I tried the HDMI module it didn’t worked too, I made a mistake in the connections with the Hecate header pins. Are you sure you didn’t invert an I2S cable, the order on the Hecate board is inverted to the order on the HDMI module.

May be check too that the LT3045 3,3V regulator produce well the 3,3V with only 5V in input. It is given as a low voltage drop out regulator (350mV) but check it. In my case I put a 8,5V in input (I only had a 6V transformer available).

You said: “i2s input detects native (and if i recall corrrectly also DoP)”

I’m unable to check this point, the Hecate Board produces DoP only on its coax SPDIF output. On the I2S output, I always see native DSD 64, 128 and 256. I discarded the 256 (with Roon’s DSP) while the DirectStream does not support it (may be a further imrovement Ted ? 65_gif). The PCM works on all frequencies (192 is maxi for me) and outputs as PCM.

Do not get discouraged, you are close to the target 41_gif


thank you Ted, Patrick.

yes the GD-Audio i2s module is powered with 3,33V coming out from regulator (no 5V going to the i2s module, just 3,33V and I2S signals from HECATE)

So to the connections up:

HECATE ------ GD-Audio module

GND PIN ------ GND (the one where other i2s junctions are)

DATA PI ------- DATA

BCLK ------- BCK

LRCKR ------- WCK

MCKL ------- not connected

Then i am powering GD-Audio module with “+” and “-” 3,3V connections from 3,3V regulator

That’s all, the USB power to HECATE is delivered Either by 5V regulator, or direcly from RPI - with no differen

I have tried to re-solder pin header to hecate just to make sure there is not bad connection, also measured HDMI I2S input pins if there is no shortcut etc. As I have also second unused GD-Audio module i have measured impedance between i2s pins (of cosrse on my desk disconnected from hecate and DAC) and the impedances there look the same so i think the module should be OK.

I still have only DSD64 playing left channel only (be it DoP or Native setting in Roon) , while DAC shows DSD128

PCM does not play at all.

@Patrick if it would be possible for you could you make a more detailed photo how is power to GD-Audio module connected and which specific pins on Hecate board connects to which junctions on the GD-Audio module?

i will try to attach my photo here, just don’t be scared - as part of evening playing with kids we made temporary Lego cases for this project :wink:



Hi Maniac,

Your table (HECATE —— GD-Audio module) is correct for me.

I do not see clearly where your red cable is going on the Hecate board, it seems to be on the left side of the header. On the left side I’ve plugged strictly nothing.

Actually what is the red cable, is it the MCK?

From what I see the brown should be the GND. On my Hecate board I plugged the HDMI ground to the 2nd PIN from bottom on the right side of the Hecate header.

Did you make your connections in conformance with the schema below (right side of the header is the one closest to the SPDIF output)?



I add an note. Audiophonics told me they got some issues to have a working HDMI module while the whole setup (Hecate, I2S links and HDMI module) where not enclosed in a box and may be not isolated from external interferences.

So all my prototypes where tested in a metal case (even not grounded, thought…) and I got no issue except some clicks and pops at some occasions. The final version of the NanoPat is ok at this point.


Yup, it’s in order with that scheme, i am just using GND pin under the DATA pin. The red one is MCLK indeed.

I will test other module during the weekend and will look for some metal casing - thanks for the tip!

The only additional thing i tested today was connection from my notebook and it had the same symptoms as from RPI so i can consider RPI/SW to be OK. Oh and had also to re-solder the GND cable as i found it broken (brown one - near the i2s module - it can be seen on the fullres photo). It seems the cables I have are fragile, i may use your approach to keep the cables on the 2nd module as they are and just solder them to the Hecate i2s pinout. Will report back, hopefully this time with good news :slight_smile:


Actually I would like to replace the AudioGD I2S wires by few silver plated I2S cables I bought + shield them with a copper cage, but I not had patience and was afraid to broke something on the HDMI module’s PCB…

Good luck !