Hi, guys!
I have been having a problem with 2 of my DIY DACs for an extended period of time. The problem was with i2s connection btw a DAC board and a spdif to i2s converter in one case and USB to i2s converter (I2SoverUSB) in another case.
The sound was smeared in bass and treble and I could not do anything about it. The sound was smeared in comparison to a CD player which had a clear and precise sound.
This was despite the fact that one DAC has a synchronous reclocking flip flop after i2s connection just near to a DAC IC.
I read somewhere that proper i2s connection requires that every i2s signal wire has a corresponding ground wire and implemented it but after having done this the sound became very harsh without lots of air and treble.
What to do? And then it ocurred to me...
What I did was that I made a corresponding ground wire for every single i2s wire which was put very close to the signal wire and even twisted with it but (and this is the most important part) every ground wire was cut from ground plane on one side. In other words all 3 ground wires were connected to the gound from one side only. And only after having done this the problem of smeared sound was rectified and the sound became as life like and vigorous as on a CD player.
It turns out that if ground wires are connected to the ground on both ends then the sound is dull and lifeless but if I remove ground wires then it becomes smeared.
Of course we need at least one ground connection btw a DAC board and converter but I came to conclusion that this ground wire should be located far from the signal wires.
I would be interested in hearing other opinions on this matter and any possible scientific explanation of this phenomenon.
I have been having a problem with 2 of my DIY DACs for an extended period of time. The problem was with i2s connection btw a DAC board and a spdif to i2s converter in one case and USB to i2s converter (I2SoverUSB) in another case.
The sound was smeared in bass and treble and I could not do anything about it. The sound was smeared in comparison to a CD player which had a clear and precise sound.
This was despite the fact that one DAC has a synchronous reclocking flip flop after i2s connection just near to a DAC IC.
I read somewhere that proper i2s connection requires that every i2s signal wire has a corresponding ground wire and implemented it but after having done this the sound became very harsh without lots of air and treble.
What to do? And then it ocurred to me...
What I did was that I made a corresponding ground wire for every single i2s wire which was put very close to the signal wire and even twisted with it but (and this is the most important part) every ground wire was cut from ground plane on one side. In other words all 3 ground wires were connected to the gound from one side only. And only after having done this the problem of smeared sound was rectified and the sound became as life like and vigorous as on a CD player.
It turns out that if ground wires are connected to the ground on both ends then the sound is dull and lifeless but if I remove ground wires then it becomes smeared.
Of course we need at least one ground connection btw a DAC board and converter but I came to conclusion that this ground wire should be located far from the signal wires.
I would be interested in hearing other opinions on this matter and any possible scientific explanation of this phenomenon.
In one DAC frequency of MCLK is 16mHz and in another case I have a USB to i2s converter I2SoverUSB and there is no mclk there.
Very strange. But there are too many unknowns to do some good explanations. Maybe jitter changes significantly due to ground loops? Who knows.. Measurements with instruments would be the best whay to find out.
as regards jitter in one DAC I have a reclocking circuit right after i2s connection and if any jitter arises then it is eliminated by reclock according to the theory...
Hm... You see my second DAC is connected to a USB to i2s converter I2Soverusb which has synchronus reclocking onboard and here there is no synchronous reclocking after i2s connection. So there's no need to bypass reclocking. Need I repeat that on this second DAC I hear similar results when cutting the ground wires from one side?
You have a ground noise problem. Although it could be a bit more complicated than the descriptions that follow, one way to think about it could be summed up as: When you connect the I2S ground wires then ground loops are created and noisy ground currents start flowing in those wires and distort the I2S signals that the ground wires are wrapped with. When you cut the wires at one end then the ground wires act like electrostatic shields to help protect the I2S signals from being distorted by local electric fields.
Most likely noisy ground currents flow when the grounds are connected because of how power supply grounds are connected somewhere else in the circuit. Usually its best to fix problems at the source, if possible.
It might help a lot to see hi-res pics of how you have the I2S connections hooked up, and how your power supply grounds are connected. Also helpful to be able to see how long all the wires are, and how they are routed.
On the other hand, if you are happy with sound as it is now, you could leave the circuitry as it is. Up to you.
The DAC board and the converter are powered by separate PSU's and transformers and their grounds do not connect anywhere except in the i2s connection between them.
If the ground wires are only connected at one end, then they are shields, not grounds.
Current will not flow in a shield wire connected at one end.
Current will not flow in a shield wire connected at one end.
Last edited:
if we take an antenna which is connected only at one end then how does it work if there is no current flowing along it?
Radiated electromagnetic waves are emitted and or received by the antenna. According to Maxwell equations, the fields are coupled to (usually) resonant currents flowing in the antenna. For that action to take place the electrical length of the wire needs to pretty close to 1/4 wavelength of the electromagnetic waves or a longer multiple of that. That sets limits on frequencies at which an antenna can be effective.
In other words, its a different situation from currents which are conducted through wires without significant radiated energy involved. For coax and twisted pair type conductors, radiated fields should be mostly canceled out and thus not be an issue (without getting into more special types of cases).
In other words, its a different situation from currents which are conducted through wires without significant radiated energy involved. For coax and twisted pair type conductors, radiated fields should be mostly canceled out and thus not be an issue (without getting into more special types of cases).
Last edited:
Since I2SoverUSB board uses digital isolator for I2S the ground wires between the DAC board and I2SoverUSB board do not form ground loops as the isolated ground in I2SoverUSB is essentially part of DAC ground. The problem is probably caused by something else.
But please take note that my another DAC has a slightly different construction and there i2s connection is made before ground isolation (digital ground) but the sound changes in a similar way in that DAC when I manipulate with ground wires.
Gosh, I wish I could hear that. I mean, I've never heard a system do stuff like wrt I2S ground wires and I'd like to learn what that sounds like.
You see, I have an I2S connection made from 3, twisted wire pairs, with three ground connections at the source end and a single "nearby" connection on the other. It's
the best I can do without scraping the receiving board down to a ground plane on that single ground side.
By your experience, my wires that connect grounds could be not giving me the optimal "experience". I think it sounds good - but maybe if I break the ground connections between the two?
I dont understand how the signal inputs on the downstream side have a reference, without a ground, to the upstream signal drivers on that input board. Particularly if the input board is powered from a separate power supply, which is the case for my setup.
I use I2SoverUSB myself. It has a 'dirty' side powered by USB, and a 'clean' side which requires an isolated 5v power supply at the dac.
If a shared 5v power supply is used by the dac and the 'clean' side of I2SoverUSB, then a ground loop will be formed when the I2S grounds are connected. If two separate power supplies were used, but the ground of each power supply output were connected to AC line ground then that would also cause a problem. I saw one dac where a jumper was added to power the supposedly clean side from dirty USB power. That formed a big ground loop with the PC. In addition there are other possible ways to mis-wire I2SoverUSB.
Anyway, if the OP does not wish for help fixing the problem then there is nothing we can do beyond what has already been said. If he would like help then we need more information, such as hi-res pics of the build. It would likely also be helpful if the OP has a scope and DVM to perform some tests.
If a shared 5v power supply is used by the dac and the 'clean' side of I2SoverUSB, then a ground loop will be formed when the I2S grounds are connected. If two separate power supplies were used, but the ground of each power supply output were connected to AC line ground then that would also cause a problem. I saw one dac where a jumper was added to power the supposedly clean side from dirty USB power. That formed a big ground loop with the PC. In addition there are other possible ways to mis-wire I2SoverUSB.
Anyway, if the OP does not wish for help fixing the problem then there is nothing we can do beyond what has already been said. If he would like help then we need more information, such as hi-res pics of the build. It would likely also be helpful if the OP has a scope and DVM to perform some tests.
Last edited:
In my case the digital part of I2SoverUSB is powered by PC and the clean side is powered by a separate PSU. So there are no issues here. The grounds of these 2 power supplies are not connected to AC. Therefore, there are no ground loops. Unfortunately I don't have a scope but I do have a digital voltmeter.
That is incorrect. The clean side of the I2SoverUSB can be connected to DAC without ground loops provided that the digital isolator implementation of I2SoverUSB is done properly.
Okay. With the power switched off and the I2S grounds left disconnected, if you measure between the dac ground and the USB board ground with the DVM, what is the resistance? Also, is the resistance constant or does the ohmmeter show the resistance is changing for a few seconds before it settles?