Coldbug fix for MSI P55-GD80, Trinergy, and Fuzion

Quote:

---

Originally Posted by K404

I was under the impression it was signalling, not power? :)

---

It's both actually Kenny if you think about it. Looking at the description, it sets the standing bias for the PCI/e bus reciever stages (presumably to facilitate an adequate level of voltage swing/ideal curve). IF it's increasing bias
to the receiver stages, yes it may be potentially harmful long term. I don't think a shorted value is strictly necessary actually. All vendors just copied what Sham did because they were under pressure to do so.


In this instance, a larger PEGR BIAS change is only required during post I think. Had companies of found this out earlier, I'm sure they'd of added an automated routine to change the value for a few seconds. Mind you, it's usually EVGA that do that kind of stuff. Most other boards are completely devoid of anything to do with active compensation or signal controlling circuitry, which is one of the reasons I sometimes struggle to see what they're asking you to pay for on their top-level boards other than the bling, a :banana::banana::banana::banana: load of chokes and a funky box.

As a precaution, what I would do, is only use this mod while benching sub-zero. If you return to air/water cooling - reverse the mod.

later
Raja

Thanks Raja :D

Have EVGA set it as a pulse?

Is shorting the resistor on MSI boards shorting out the entire resistor network, or just reducing the voltage/ has the signal amplitude become the supply voltage? Whats the signal strength like before and after?

Im... possibly over-simplifying the circuit here :p:

Quote:

---

Originally Posted by K404

Thanks Raja :D

Have EVGA set it as a pulse?

Is shorting the resistor on MSI boards shorting out the entire resistor network, or just reducing the voltage/ has the signal amplitude become the supply voltage? Whats the signal strength like before and after?

Im... possibly over-simplifying the circuit here :p:

---

I don't think it's setup in 'pulse' value as you put it on the EVGA board, I think the value at post remains static (don't quote me on that though, Sham's a bit of a whizz at fiddling with this :banana::banana::banana::banana:). I have not looked at EVGA's full implementation so can't comment on whether shorting the resistor results in open to ground - depends on what else they have in-series to provide the 63 step control.

On other boards, yes, it's fully open because the resistor is to ground. The drive voltage is likely VTT (it supplies just about everything on these boards). I can't comment further as I have no idea how things are setup internally, although I do know these are differential pairs (most have a reference voltage set around the mid point), so I'd think the change in bias just results in satisfying a required time oriented bus data transmit sense during post (change in slew behaviour).

The board engineers know a lot more than I do and have access to stuff that most of us don't. Unless there's soemthing radically different to how it appears this stuff works, the straight short to ground is a typical quick fix manoeuvre without any real finesse or consideration over long term implications.

later
Raja

Ok, had a bit more time to look at this now that I've got some work out of the way. The resistor sets BIAS comp, which is not quite the same thing as setting the standing quiescent BIAS for a FET -BIAS comp is used to attenuate (trim by freqeuncy) the differential signal bias across the transmission line to mitigate noise on an inactive/un-terminated line during transmit (so that noise does not cross-leak into an 'active' reciever thus corrupting a logic interpretation).

As such, the long term effects of full grounding may not be nasty (only Intel can answer that properly though). That all being said, there is likely to be a sweet spot for any given operating frequency and temperature to wring the most from the PEG bus and it will vary from board to board and CPU to CPU depending upon layout and trace capacitance (though more would need to be know to understand how it all works fully during a transmit).


later
Raja

my CB/CBB mod on fuzion
http://img39.imageshack.us/img39/4107/cbmodfuzion.jpg
will test it later with clarky
:p: