Tír Na HiFi

nige2000 wrote: anytime i used isolators for i2s on a complete battery system pc to dac(no pc +5v ) i found no additional benefit
and therefore considered isolators to be a cure of sorts for a noisy ps to the i2s device and that providing cleaner power was the prevention method
i was experimenting with that i2s isolation earlier and im still of that prevention type opinion for the most part
and that i may benefit from another cell and/or different wiring methods

i dont think thats why the intona didnt lock
gnd is still gnd and without a direct gnd connection between the usb to i2s converter and the soekris fpga gnd on both sides would still be connected the long way (through the earth gnding in most systems amps monitors etc)

i assume you mean the shift registers thinking -4 (-3.3v) is gnd?
is that what they call a virtual gnd?

Yea, I'd agree - using isolators seems to add some jitter/noise/skew to the signals so it's likely optimal to reclock the signals after the isolators. On the other hand, the advantage of isolators seems to be the reduction in noise before it becomes embedded in the signal & impossible to remove.

But here's the crux, as far as I see it - this reduction in noise has to happen as early as possible i.e. before it gets embedded in the signal - so if we take a typical PC audio configuration & the signalling conversions we have PC->USB->audio device ->I2S->DAC->analogue out - isolation at I2S stage is less effective than at the USB stage because the noise has probably already become embedded in the upstream signal. That's one of the reasons why the isolators on the Soekris seems to have little effect (I also think that a lot of the downstream signal manipulation in the Soekris is also probably responsible for the little noticeable effect of the isolators). Logically, the optimal place for USB isolation would be at the PC end i.e. on the card sending out the USB signals but there might also be a need for further isolation/cleaning downstream as the USB signal line seems prone to picking up noise?

Years ago I did experiments with isolating I2S signals & reclocking them - so everything of importance was on the "clean" side (clock reclockers, DAC i.e. it was done correctly - everything was isolated) - I found that it didn't make enough of a difference to warrant the work involved (two isolator chips were needed to cater for downstream signals: 4 I2S signals, 2 clock enable signals & upstream signals: clock signal) but the differences we heard in Tony's with the Intona in-line makes me think of revisiting this but at an earlier point in the signal stream.

I haven't looked too closely at the Soekris board - I was just thinking out loud about the possible reason for the failure to achieve lock with your Soekris board - it seemed to be related to ground issues?

So, you are speaking of the INPUT ground?

I assumed the output but I wonder if there could be an advantage there, also?

So how do you do this? Tie all of the I2S grounds into a bundle and that goes to the INPUT battery ground?

Can't see how that would bypass the isolators, though.

rickmcinnis wrote:So, you are speaking of the INPUT ground?

I assumed the output but I wonder if there could be an advantage there, also?

So how do you do this? Tie all of the I2S grounds into a bundle and that goes to the INPUT battery ground?

Can't see how that would bypass the isolators, though.

I don't know if you are asking me or Nige but I haven't really studied the Soekris board yet - however the general principle is to keep signal grounds & power grounds separated - the signal return currents are very much smaller than the PS return currents & it's easy to pollute signal ground currents with the PS ground currents - the only place they should be joined together is at the PS, as you said.

Isolators on the signal inputs obviously introduce complications in grounding schemes but shouldn't be too difficult to handle

Thanks, jkeny,

I was asking whomever would be willing to answer! As if I would ever turn down an answer from you!

I have heard nothing much good about the isolators and your explanation of their uselessness on the SOEKRIS board goes along with my experience. I think SOEKRIS stuck them in as part of the old "checking the boxes" routine; trying to make something that will be all things to all people and who could blame him? I suspect his business model requires him to sell a bunch of them.

I hear no difference between them being powered or not. AS Nigel has pointed out they become less and less important with a "clean" computer - I have long wondered if there was something wrong with them and for all I know there is. They have been this way from the very beginning. That is, they do not require power to pass the signal and I do not know enough about them to know what, if anything, that means.

dimdim, who has a SOEKRIS blog, said he destroyed his and replaced them. He never said how he knew he had destroyed them so I have no idea what happens. Do they simply no longer pass the signal at all? Can you tell me? (it will ease my mind).

If I said something that made sense regarding PS in my post it was luck! I am just trying to understand what Nigel is suggesting and how to implement it. Though thinking about what YOU said it would seem we would be either bypassing this grounds separation or blending them when they should be separate?

AS usual I am getting over my head ...

Sure, isolators become of less & less significance as the noise they are blocking becomes less & less but I don't know how low this has to be before isolators are irrelevant. One of the reasons for this lack of knowledge is that measurements of this low level of noise is difficult & even more difficult is to have a spectrum analysis of this noise

It's always difficult to decide what the cause is of what we are hearing - it's good to keep an open mind on these matters - for instance, the Intona & the Regen both regenerate the the USB signal & therefore I would expect that the noise spectrum is also changed when using these devices. To me, that this explains why two Regens sound different to one Regen or a Regen+Jtterbug or Intona+Regen i.e. because noise is not a static thing - it is a dynamic thing like a cloud which changes shape & density depending on what's happening dynamically in the signal handling aspects of the circuitry - as you change the signal, the noise also changes.

jkeny wrote: Yea, I'd agree - using isolators seems to add some jitter/noise/skew to the signals so it's likely optimal to reclock the signals after the isolators. On the other hand, the advantage of isolators seems to be the reduction in noise before it becomes embedded in the signal & impossible to remove.

before the intona testing i was pondering taking out these i2s isolators, surely any i2s device powered with solely 3.3v from lifepo4 couldn't possibly generate enough noise to need i2s isolators after the intona testing last wk and the i2s isolater testing i did earlier it must be coming from upstream which still astounds me tonys pc is well pimped out ps wise although it had a generic usb card in it which would have dulled its performance for everything

But here's the crux, as far as I see it - this reduction in noise has to happen as early as possible i.e. before it gets embedded in the signal - so if we take a typical PC audio configuration & the signalling conversions we have PC->USB->audio device ->I2S->DAC->analogue out - isolation at I2S stage is less effective than at the USB stage because the noise has probably already become embedded in the upstream signal. That's one of the reasons why the isolators on the Soekris seems to have little effect (I also think that a lot of the downstream signal manipulation in the Soekris is also probably responsible for the little noticeable effect of the isolators). Logically, the optimal place for USB isolation would be at the PC end i.e. on the card sending out the USB signals but there might also be a need for further isolation/cleaning downstream as the USB signal line seems prone to picking up noise?

makes sense
when i was experimenting making usb cables it was very apparent that shielding (to pc side only) was a must and there construction was at least every bit as fickle as interconnects if you say build a usb isolator into a usb card most cables will pick up a multitude of crap before it gets to the i2s device

Years ago I did experiments with isolating I2S signals & reclocking them - so everything of importance was on the "clean" side (clock reclockers, DAC i.e. it was done correctly - everything was isolated) - I found that it didn't make enough of a difference to warrant the work involved (two isolator chips were needed to cater for downstream signals: 4 I2S signals, 2 clock enable signals & upstream signals: clock signal) but the differences we heard in Tony's with the Intona in-line makes me think of revisiting this but at an earlier point in the signal stream.

I haven't looked too closely at the Soekris board - I was just thinking out loud about the possible reason for the failure to achieve lock with your Soekris board - it seemed to be related to ground issues?

nothing to do with the soekris the intona interacts only with the diyinhk i2s (well suppose to) when we bypassed the built in usb hub it is as stock config
the only thing i can think of was that the hub was powered and connected to the usb data lines when we bypassed it
but then again it worked that way with a direct cable from usb card
maybe its impedance sensitive or there's too many usb connections
didnt work with the power injector either

we were discussing gnd at the time because when theres pc's with clock mods connecting up stuff to it causes a glitch were the monitor goes all flashy and pc needs to be rebooted
and we were doing a lot of connecting:)

rickmcinnis wrote:So, you are speaking of the INPUT ground?

I assumed the output but I wonder if there could be an advantage there, also?

So how do you do this? Tie all of the I2S grounds into a bundle and that goes to the INPUT battery ground?

Can't see how that would bypass the isolators, though.

are you talking about isolating digital signals ?
or optimal wiring methods of the analog outputs?

nige2000 wrote:

jkeny wrote: Yea, I'd agree - using isolators seems to add some jitter/noise/skew to the signals so it's likely optimal to reclock the signals after the isolators. On the other hand, the advantage of isolators seems to be the reduction in noise before it becomes embedded in the signal & impossible to remove.

before the intona testing i was pondering taking out these i2s isolators, surely any i2s device powered with solely 3.3v from lifepo4 couldn't possibly generate enough noise to need i2s isolators after the intona testing last wk and the i2s isolater testing i did earlier it must be coming from upstream which still astounds me tonys pc is well pimped out ps wise although it had a generic usb card in it which would have dulled its performance for everything

As I said, I haven't really looked at the Soekris input stage design but these isolators are presumably on the input signals to reduce any noise coming in on the external I2S signals? Yes, Tony's pimped PC should be better than most for noise on USB

But here's the crux, as far as I see it - this reduction in noise has to happen as early as possible i.e. before it gets embedded in the signal - so if we take a typical PC audio configuration & the signalling conversions we have PC->USB->audio device ->I2S->DAC->analogue out - isolation at I2S stage is less effective than at the USB stage because the noise has probably already become embedded in the upstream signal. That's one of the reasons why the isolators on the Soekris seems to have little effect (I also think that a lot of the downstream signal manipulation in the Soekris is also probably responsible for the little noticeable effect of the isolators). Logically, the optimal place for USB isolation would be at the PC end i.e. on the card sending out the USB signals but there might also be a need for further isolation/cleaning downstream as the USB signal line seems prone to picking up noise?

makes sense
when i was experimenting making usb cables it was very apparent that shielding (to pc side only) was a must and there construction was at least every bit as fickle as interconnects if you say build a usb isolator into a usb card most cables will pick up a multitude of crap before it gets to the i2s device

Years ago I did experiments with isolating I2S signals & reclocking them - so everything of importance was on the "clean" side (clock reclockers, DAC i.e. it was done correctly - everything was isolated) - I found that it didn't make enough of a difference to warrant the work involved (two isolator chips were needed to cater for downstream signals: 4 I2S signals, 2 clock enable signals & upstream signals: clock signal) but the differences we heard in Tony's with the Intona in-line makes me think of revisiting this but at an earlier point in the signal stream.

I haven't looked too closely at the Soekris board - I was just thinking out loud about the possible reason for the failure to achieve lock with your Soekris board - it seemed to be related to ground issues?

nothing to do with the soekris the intona interacts only with the diyinhk i2s (well suppose to) when we bypassed the built in usb hub it is as stock config
the only thing i can think of was that the hub was powered and connected to the usb data lines when we bypassed it
but then again it worked that way with a direct cable from usb card
maybe its impedance sensitive or there's too many usb connections
didnt work with the power injector either

we were discussing gnd at the time because when theres pc's with clock mods connecting up stuff to it causes a glitch were the monitor goes all flashy and pc needs to be rebooted
and we were doing a lot of connecting:)

Yea, it was all very strange - when we pulled out the plug of the PS which was the charger for the Soekris batteries, it caused Tony's PC to hang. This hang occurred, even though there was no ground pin in the plug & the charger was disconnected from the batteries. When we pulled out the RCA leads connecting the DAC to the amp, the PC hung

The Intona would only work with the DIYinHK XMOS based DAC when it was connected through your DIY Regen hub & not when connected directly, yet Pearse's Intona worked when directly connected to his Meitner XMOS based DAC.

We couldn't get any 5V injector to work with the Intona, yet Pearse's Hynes supply powering the PPA USB card works fine

All very strange & slightly bewildering

jkeny wrote:As I said, I haven't really looked at the Soekris input stage design but these isolators are presumably on the input signals to reduce any noise coming in on the external I2S signals? Yes, Tony's pimped PC should be better than most for noise on USB

yes these are soekris built in i2s isolators
doubt they have any benefit if power is clean enough to xmos, as you said before damage has already been before it gets to i2s stage

Yea, it was all very strange - when we pulled out the plug of the PS which was the charger for the Soekris batteries, it caused Tony's PC to hang. This hang occurred, even though there was no ground pin in the plug & the charger was disconnected from the batteries. When we pulled out the RCA leads connecting the DAC to the amp, the PC hung

dont even talk to me bout it
it has wrecked my head most of last yr
funny enough pc doesn't actually hang it just buggers up display until reboot

on diyaudio i asked for advice on installing external clocks on motherboards in an attempt to solve these crazy modded mobo clk display issues
i linked to pangs blog as an example of what i wanted to accomplish
i got reply thanking me for the link because he was going to use it in his class as an example of what not to do
begs the question how something so wrong sounds so good

just thinking out loud
we probably don't even the amount of signals are going through domestic electrical wiring we can get several mb/s ethernet transferred through ac cabling with those "home plug" ethernet yokes
i think the neutral in domestic wiring is merged with earth in the fuse box so plugging stuff in or out might be giving a myriad of optional return paths and/or signal gnd differentials?

The Intona would only work with the DIYinHK XMOS based DAC when it was connected through your DIY Regen hub & not when connected directly, yet Pearse's Intona worked when directly connected to his Meitner XMOS based DAC.

We couldn't get any 5V injector to work with the Intona, yet Pearse's Hynes supply powering the PPA USB card works fine

All very strange & slightly bewildering

im stumped
needs more investigation
maybe Pearse and i can investigate sometime

pearse did note that he could not get a connection with his ppa usb cable without the regen either
there seems to be many connection issues like this on other forums

it seems fairly evident that there needs to be usb isolation somewhere between pc and the usb to i2s conversion

imagine the revelation the intona would be on an imac

nige2000 wrote: imagine the revelation the intona would be on an imac

Agree with all you said before this point but I think this is taking it a bit far - Mac users will probably not try an Intona so we will neve know if this is true :)