Hi,
I am the proud owner of a Perpetual Technologies P-3A DAC (I just love this device!). (Yes I also have the P-1A but surprisingly adding it actually degrades the sound - confirmed by blind testing 3 different people)
A few years ago I upgraded the P-3A by buying a fantastic jitter reducer named M2Tech HiFaceEvo 2. (which I highly recommend too, Though they have a newer model now)
However, this upgrade was not complete as I am only using the inferior toslink to connect the duo instead of i2s.
The problem is the i2s connector on the M2tech is HDMI and on the Perpetual Technologies it is a 5 pin din connector (PT manual says it is compatible with the Audio Alchemy standard).
I also know that the signal itself is not necessarily exactly compatible in terms of voltage as mentioned here:
https://www.m2tech.biz/wp-content/uploads/2017/06/App001-Using-hiFace-Evo-I2S-Output-PrB.pdf
I asked an audiophile technician I work with to create a cable for me to connect the 2 and he asked me to do some research.
So if anyone here did something like this or have any idea on how to go about it or any information that can help I would appreciate it a lot.
Thanks,
Ben
I am the proud owner of a Perpetual Technologies P-3A DAC (I just love this device!). (Yes I also have the P-1A but surprisingly adding it actually degrades the sound - confirmed by blind testing 3 different people)
A few years ago I upgraded the P-3A by buying a fantastic jitter reducer named M2Tech HiFaceEvo 2. (which I highly recommend too, Though they have a newer model now)
However, this upgrade was not complete as I am only using the inferior toslink to connect the duo instead of i2s.
The problem is the i2s connector on the M2tech is HDMI and on the Perpetual Technologies it is a 5 pin din connector (PT manual says it is compatible with the Audio Alchemy standard).
I also know that the signal itself is not necessarily exactly compatible in terms of voltage as mentioned here:
https://www.m2tech.biz/wp-content/uploads/2017/06/App001-Using-hiFace-Evo-I2S-Output-PrB.pdf
I asked an audiophile technician I work with to create a cable for me to connect the 2 and he asked me to do some research.
So if anyone here did something like this or have any idea on how to go about it or any information that can help I would appreciate it a lot.
Thanks,
Ben
If the dac connector wired the same as this one:
If so, and if the following is correct for the other unit you have :
hiFace Evo uses an RJ-45, 8 pins connector with the following pinout (left to right):
1) SDATA
2) GND
3) LRCK (FS)
4) GND
5) SCLK (BIT CLOCK)
6) GND
7) MCLK
8) GND
Signals are LVCMOS, that is CMOS levels at 3.3V supply. They are TTL-compatible, that is, they can directly drive a 5V TTL input stage without need for level translation. The hiFace Evo’s I2S output is drive by a high current driver, so it can directly feed an input when no level translation is needed.
Then it looks like making a cable might be pretty straightforward. However, it would probably be best to keep the cable very, very short length. Like 2-3 cm if you can manage. LVCMOS I2S bus signals are most commonly not designed to drive long cables at all. If you look inside the hiFace and see what chip or chips they are using to drive the I2C outputs, then depending on that maybe some other cabling options could be possible.
Assuming typical LVCMOS drivers, all the grounds from the hiFace connector should probably be twisted with the signal that comes before it in the connector order (e.g. SDATA on pin-1 twisted with GND on pin-2, and so on). All the grounds at the DIN connector end should be connected to the DIN connector shell. It may work better to separate the the twisted pairs somewhat from each other if possible. You can experiment to see if you think it makes any difference. Also, instead of twisted pairs it could be that small 50-ohm coax cables for each I2S signal and its ground would work too. Again it may be best if trying that to see if moving the cables around, say, bundling them together, or else separating them from somewhat from each other, has any audible effect.
If so, and if the following is correct for the other unit you have :
hiFace Evo uses an RJ-45, 8 pins connector with the following pinout (left to right):
1) SDATA
2) GND
3) LRCK (FS)
4) GND
5) SCLK (BIT CLOCK)
6) GND
7) MCLK
8) GND
Signals are LVCMOS, that is CMOS levels at 3.3V supply. They are TTL-compatible, that is, they can directly drive a 5V TTL input stage without need for level translation. The hiFace Evo’s I2S output is drive by a high current driver, so it can directly feed an input when no level translation is needed.
Then it looks like making a cable might be pretty straightforward. However, it would probably be best to keep the cable very, very short length. Like 2-3 cm if you can manage. LVCMOS I2S bus signals are most commonly not designed to drive long cables at all. If you look inside the hiFace and see what chip or chips they are using to drive the I2C outputs, then depending on that maybe some other cabling options could be possible.
Assuming typical LVCMOS drivers, all the grounds from the hiFace connector should probably be twisted with the signal that comes before it in the connector order (e.g. SDATA on pin-1 twisted with GND on pin-2, and so on). All the grounds at the DIN connector end should be connected to the DIN connector shell. It may work better to separate the the twisted pairs somewhat from each other if possible. You can experiment to see if you think it makes any difference. Also, instead of twisted pairs it could be that small 50-ohm coax cables for each I2S signal and its ground would work too. Again it may be best if trying that to see if moving the cables around, say, bundling them together, or else separating them from somewhat from each other, has any audible effect.
Last edited:
Apparntly the article I linked is too old and relates to old M2Tech models. The new ones (like the one I have) use an HDMI cable using the PS Audio standard for I2S. I am guessing that changes things.
Regarding length of cable:
Does it really need to be short as 2-3CM? that seems unrealistic from a practical standpoint. I would guess ~ 10 CM is the shortest that is doable.
Does the length implicate sound quality or is it just a matter of working or not working?
I am thinking of taking an HDMI cable and removing 1 connector and connecting a DIN connector instead. (and maybe shorting the cable in the process if needed)
Regarding length of cable:
Does it really need to be short as 2-3CM? that seems unrealistic from a practical standpoint. I would guess ~ 10 CM is the shortest that is doable.
Does the length implicate sound quality or is it just a matter of working or not working?
I am thinking of taking an HDMI cable and removing 1 connector and connecting a DIN connector instead. (and maybe shorting the cable in the process if needed)
The newer M2Tech uses LVDS signaling (differential) on I2S output so it probably does not work directly with your dac. With LVDS the length of the cable can be much longer (maybe up to 3m).
Maybe you could post pics of the connectors on the back of you units so we can see what they are. Also it might help if you have manuals for the units that describe the interfaces, then maybe you could post links to the manuals or and or scan the pages with relevant information and post those.
Otherwise you are asking us to go dig up details you probably already have.
Regarding the modern HMDI I2S interfaces, usually they are LVDS (Low Voltage Differential Signaling). If your dac uses LVCMOS I2S signals (Low Voltage Complementary Metal Oxide Semiconductor), then the two sets of signals are not directly electrically compatible. It takes some active circuitry (some chips that need some power) to do the signal conversion. Such chips could possibly be built into a cable nowadays, but it would then have to be a very custom engineered cable, not something that a tech could just make up one of. What someone might be able to do for you is put the electronics in a little box that goes behind your dac. It would have a very short cable for the LVCMOS signals.
Otherwise you are asking us to go dig up details you probably already have.
Regarding the modern HMDI I2S interfaces, usually they are LVDS (Low Voltage Differential Signaling). If your dac uses LVCMOS I2S signals (Low Voltage Complementary Metal Oxide Semiconductor), then the two sets of signals are not directly electrically compatible. It takes some active circuitry (some chips that need some power) to do the signal conversion. Such chips could possibly be built into a cable nowadays, but it would then have to be a very custom engineered cable, not something that a tech could just make up one of. What someone might be able to do for you is put the electronics in a little box that goes behind your dac. It would have a very short cable for the LVCMOS signals.
In term of taking photos of the back of the devices there is no need. The M2Tech uses HDMI for the I2S and the Perpetual Technologies uses a 5 pin din connector that you have already posted the schematic for.
This is the schematic for the PS Audio standard used by the m2tech:
https://forum.psaudio.com/t/compatible-i2s-source-devices/1336/457
Regarding the Perpetual Technologies, I couldn't find a schematic. But inspired by Markw4 to go dig, I did find a schematic for an Audio Alchemy DTI pro 32 which is compatible with the P-1A and has both an I2S input and output:
Edit (fixed link): https://elektrotanya.com/audio_alchemy_dtipro32.pdf/download.html#dl
So I hope now this gives the pros here enough information to figure out what is needed to make such a cable and what length it needs to be.
Also, if the I2S interfaces are indeed not compatible, do you think modding the P-1A by replacing the original I2S input with one matching the PS Audio standard is a viable option? Would that be easier or harder than making a cable with active circuitry?
Thanks for all the help everybody!
This is the schematic for the PS Audio standard used by the m2tech:
https://forum.psaudio.com/t/compatible-i2s-source-devices/1336/457
Regarding the Perpetual Technologies, I couldn't find a schematic. But inspired by Markw4 to go dig, I did find a schematic for an Audio Alchemy DTI pro 32 which is compatible with the P-1A and has both an I2S input and output:
Edit (fixed link): https://elektrotanya.com/audio_alchemy_dtipro32.pdf/download.html#dl
So I hope now this gives the pros here enough information to figure out what is needed to make such a cable and what length it needs to be.
Also, if the I2S interfaces are indeed not compatible, do you think modding the P-1A by replacing the original I2S input with one matching the PS Audio standard is a viable option? Would that be easier or harder than making a cable with active circuitry?
Thanks for all the help everybody!
Last edited:
You could use something like this:
https://hifimediy.com/product/hifimediy-i2sdsd-isolator-384khz-hdmi-to-ttl-interface/
https://hifimediy.com/product/hifimediy-i2sdsd-isolator-384khz-hdmi-to-ttl-interface/
HDMI pinout might have to be swapped around depending on whether the converter board uses the PS Audio standard.
There are differing standards, as can be seen at: https://docs.google.com/spreadsheets/d/1h5PMUBkldkpt1rCnAR4ZHYGZNeCe-vwIFyKWYMZWsX0/htmlview#
There are differing standards, as can be seen at: https://docs.google.com/spreadsheets/d/1h5PMUBkldkpt1rCnAR4ZHYGZNeCe-vwIFyKWYMZWsX0/htmlview#
Thinking about this idea, a question came to my mind about the suggested conversion. The whole point of using I2S is to reduce jitter to a minimum. Does adding an I2S conversion box as suggested run the risk of introducing new jitter, or will it keep the signal intact?
Jitter can be hard to keep minimized. It isn't just what is commonly referred to as jitter either. Technically jitter is variation in timing as viewed in the time domain. Phase noise is similar variation as viewed in the frequency domain.
Often what is called jitter in some of the literature is more or less the same as what may also be referred to as far-out phase noise. That is to say, phase noise that is relatively more far removed in frequency from the carrier or clock frequency. OTOH, close-in phase noise averages being much closer in frequency to the nominal carrier or clock frequency. The distinction is made because close-in and far-out phase noise tends to have different effects in data converters and in other applications such as radar.
In any case, getting jitter or phase noise very low tends to take a some effort, and its easy to have it degrade. For dacs that offer HDMI I2S inputs, often the I2S signals pass through more than one stage of processing. First they are usually converted from LVDS to LVCMOS format. Then in many cases the signals may pass through a CPLD chip which is programmed to handle signal routing, clock division, etc. Such processing most likely is going to have at least some adverse effects on jitter/phase-noise. The question is how much exactly?
Answer might be that it can be hard to say without trying it and seeing what you thing sounds best. A lot depends on the particular implementation of all the involved circuitry. Can't speak to it much more generally than that. Sorry.
Often what is called jitter in some of the literature is more or less the same as what may also be referred to as far-out phase noise. That is to say, phase noise that is relatively more far removed in frequency from the carrier or clock frequency. OTOH, close-in phase noise averages being much closer in frequency to the nominal carrier or clock frequency. The distinction is made because close-in and far-out phase noise tends to have different effects in data converters and in other applications such as radar.
In any case, getting jitter or phase noise very low tends to take a some effort, and its easy to have it degrade. For dacs that offer HDMI I2S inputs, often the I2S signals pass through more than one stage of processing. First they are usually converted from LVDS to LVCMOS format. Then in many cases the signals may pass through a CPLD chip which is programmed to handle signal routing, clock division, etc. Such processing most likely is going to have at least some adverse effects on jitter/phase-noise. The question is how much exactly?
Answer might be that it can be hard to say without trying it and seeing what you thing sounds best. A lot depends on the particular implementation of all the involved circuitry. Can't speak to it much more generally than that. Sorry.
First of all, thank you for all the great insight and information. It really helps to understand this better.
You mentioned this:
"First they are usually converted from LVDS to LVCMOS format."
Does this imply that in the M2Tech the conversion goes the other way around? If so can a mod be made (to the M2Tech) to get the LVCMOS out directly, (using a very short cable of course) instead of converting to LVDS and back?
You mentioned this:
"First they are usually converted from LVDS to LVCMOS format."
Does this imply that in the M2Tech the conversion goes the other way around? If so can a mod be made (to the M2Tech) to get the LVCMOS out directly, (using a very short cable of course) instead of converting to LVDS and back?
Hard to say. It could be difficult depending on how hardware is designed for each unit.
Regarding LVCMOS and LVDS, the former is the most common hardware signal format used for dac chips and dac digital circuitry. OTOH, LVDS is used when things like I2S signals have to travel over longer distances and or when improved noise immunity might be useful.