Further Isolation of the Silverstone PCIe - USB Card

[AlainGr]

Hum... The "thick" guy that I am is now confused I will post a photo later to show what I did... It will be easier for all to see if I did the right (or wrong) thing...

Alain

[CoenP]

If you didn't its about the USB grounding wire from the USB receptacle to the PSU and the one from the PE to the chassis.

H*ll no Coen. I never said anything about the wire from PE to the chassis ... Leave that be please.

Peter

ok!

(misunderstanding: in my NOS1 the USB grounding was attached to the PSU board with a little screw not a bolt like the PE wire)

regards, Coen

[AlainGr]

ok!

(misunderstanding: in my NOS1 the USB grounding was attached to the PSU board with a little screw not a bolt like the PE wire)

regards, Coen

I think that you described it better than I... A screw yes, attached to the PSU "board", not the PSU itself... Anyway, will post a pic later...

Alain

[Scroobius]

Hi Anthony,

My next thoughts are to the dexa clock guys.  Can you give this a try? 

I just may give it a try but I do not have a power supply. I do have a linear 5v supply for the Molex and a 3.3v regulator. So I would have to find something to drive the 3.3v regulator it should not be impossible.

If I were a betting man (I would be willing to bet Peter another bottle of whisky on this ha ha) my hunch would be that it will not affect the sound when the clock is a Dexa. The clock output of the Dexa is isolated & floating. The power supply is the Dexa and it too is floating. OK I know that is not the whole story but as this has to be all about jitter...........

Nick and I have discussed a couple of times that the Dexa improvement could be about power supplies but there again who knows?

Anyway I cannot do this straight away as I am busy just now but as soon as I can I will have a play.

Cheers

Paul

[acg]

Hi Anthony,

Did you disconnect the chassis ground wires in the NOS1 like Peter suggested in the " best sound ever"  disconnected ground thread?

If you didn't its about the USB grounding wire from the USB receptacle to the PSU and the one from the PE to the chassis.

regards, Coen

Hi Coen,

No.  I have not disconnected that wire.  I will try it at some stage.

Regards,

Anthony

[acg]

Hi Anthony,

My next thoughts are to the dexa clock guys.  Can you give this a try? 

I just may give it a try but I do not have a power supply. I do have a linear 5v supply for the Molex and a 3.3v regulator. So I would have to find something to drive the 3.3v regulator it should not be impossible.

If I were a betting man (I would be willing to bet Peter another bottle of whisky on this ha ha) my hunch would be that it will not affect the sound when the clock is a Dexa. The clock output of the Dexa is isolated & floating. The power supply is the Dexa and it too is floating. OK I know that is not the whole story but as this has to be all about jitter...........

Nick and I have discussed a couple of times that the Dexa improvement could be about power supplies but there again who knows?

Anyway I cannot do this straight away as I am busy just now but as soon as I can I will have a play.

Cheers

Paul

Hi Paul,

It would be much appreciated if you could give this a go.  Perhaps you could use a battery?

I _think_ that at least some of what I hear is to do with grounding rather than linear power or clocks, but who really knows at this stage?

Regards,

Anthony

[PeterSt]

(misunderstanding: in my NOS1 the USB grounding was attached to the PSU board with a little screw not a bolt like the PE wire)

Coen, remember, we are Dutch. We try to speak Dutch (dufficult enough already) and try to bring across what's intended. Now :

We as poor Dutch make the first mistake in naming a bolt a screw while it really is a bolt. Or ?
A screw drives into wood and a bolt needs a nut (or tapping).
We tend to differentiate between that small bolt and the larger one by naming the small one "small bolt" but with one word which doesn't exist in English (in Dutch each noun can be made smaller by means of adding "je" to it, so, bout (bolt) - boutje (small bolt).
Inherently we tend to call the small bolt "screw" (but still the small version - schroefJE) but we will never call the larger bolt a screw (young children maybe).

Peter


PS: And then there's also the parker (a screw which drives into sheet metal with same diameter from end to head).

[PeterSt]

Its a miracle this works. Except for the tx and treceive lines that are balanced all other pins are referenced to groud. Without a ground the voltage cannot be determined. I expect that a pin that is 'low' provides the grnd connection necessary for operation.

Hey,

I have been staring at this in stages, and I suddenly might see some key in Coen's remark here. So what about those differential lines now NOT being connected to ground (adjacent pins not connected anymore) while in the normal situation they would be and which is WRONG. Well, not the best.

What about those differential lines creating ground, while that ground could be used at the other side (on the card).

The other remark/question I can make/have is :
Is the card still running well when no USB cable is connected to the NOS1 ? If not, the ground comes from that side (obviously).

It is not formal thinking that it is allowed to let come the ground from elsewhere, though still (in my view) it can be for the better "depending on". Just think about what happens when two different potentials are bonded. So in this case it can be mobo ground vs (lab) PSU ground. Both won't be the absolute truth were it for absolute zero ground reference; only differential will do that (absolute middle). The fact that differential is involved is already fishey because it won't be the same as the other grounds. Notice though that I assume(d - also in text above) that the differential signals *form* ground, which is not necessarily true I think. Anyway :
When two potentials are not the same, a current flows.

Of course, you can try to be ahead of this all and bond all grounds as much as you can, but all you will be doing is shifting the problem out of the device (like towards NOS1 in this case).
Compare this with a cut ground between two parts of the device, but now the "no ground" comes from the other device (NOS1). I personally think this is for the better. But it depends where the ground is cut.

Peter

[acg]

Hi Peter,

Following is some information extracted from an Intel Powerpoint article about Differential Signalling:

Any signal can be considered a loop is completed by two wires.
One of the “wires” in single ended signaling is the “ground plane”
Differential signaling uses two conductors
The transmitter translates the single input signal into a pair
of outputs that are driven 180° out of phase.
The receiver, a differential amplifier, recovers the signal as the difference in the voltages on the two lines.
Advantages of differential  signaling can be summed up as follows
Differential Signaling is not sensitive to SSO noise.
A differential receiver is tolerant of its ground moving around.
If each “wire” of pair is on close proximity of one and other. electromagnetic interference imposes the same voltage on both signals. The difference cancels out the effect.
Since the AC currents in the “wires” are equal but opposite and proximal, radiated EMI is reduced.
Signals passing from one board to another are not subject to the local ground disturbances.
As frequencies increase beyond 1GHz, up to 80% of the signal may be lost, but difference still crosses 0 volts.
There are still loss issues for differential signaling but only come into play in high loss system. Most single ended systems assume approximately 15% channel loss.

So, it seems as though the differential pairs can be used successfully between pcb's of different and varying grounds (which is lucky) so I doubt that they ever were connected to ground on the USB board (not directly anyway).  The differential pairs all go directly to the USB Host Controller IC which according to my understanding has separate ground (or at least separate sections for analogue and digital).

I will try your suggestion about pulling the cable from the NOS1 to see if the usb still works shortly and let you know the results.

Anthony

[acg]

Peter, I just un-minimised the o/s (with the NOS1 unplugged) plugged in a USB drive to the Silverstone card (with my further isolation) and Windows Explorer opened up with the contents of the USB drive.  Unplugged the USB drive, plugged in the NOS1, changed XXHE to the NOS1 (from Primary Sound Driver) clicked play and am currently listening to music.

So the Silverstone with these isolations works as a normal USB port.

Anthony

[nik.d]

Great finding Anthony 
So next (elegant) step would be cutting the traces on Silverstone card itself (at gold plated leads) iof using PCIe extender, isn't it?

[acg]

Possibly.  If the traces were cut then we would also need to use the "condom" to separate the faceplate from pc ground, but that can be done.  Then there are experimentations with clocks on the card, which seem to be beneficial.

Tonight I have managed to figure out that the "CLKREQ" wire can also be cut which leaves only the "Wake" and "PWR Good" wires that have a signal to the Silverstone card (apart from the differential pairs).  I know that the "Wake" pin is optional in the PCI Express standard and therefore can be made redundant, but I need to do a bit of experimenting to see whether I can tie this pin to the PC ground (i.e. not connect it to the Silverstone card).  Then there is the "PWR Good" pin that I may be able to link to the lab LPS but I am not sure about that one.

Anyway, there is still a bit to figure out.

Anthony

[PeterSt]

Hi Anthony,

Your test with the USB disk is not right I think;
Now *that* can provide ground ... (uhm, assumed it is self powered I guess).
What I meant to try is to just have the card in and see if Windows still recognizes it. Don't say "yes of course or otherwise the USB disk wouldn't have worked" because plugging in that might just get it working.

Otherwise let's not make too much fuzz about this, but we can reckon that if it's NOT working anymore we have to change our thinking somewhat ...

Peter

[acg]

Sorry I was not very clear Peter but it was a flash drive or USB stick, one of those finger sized devices that run from the power of the USB port.  I don't think that this invalidates the ground test.

Anthony

[PeterSt]

Ah, OK Anthony !
Peter