Titanic questions

[snme2pm1]

Originally Posted by Madpoo

Correct... my thinking was that instead of shorting it entirely, you wire a POT with a similar resistance range in parallel, just enough range to give you something between 0 and 0.5x of the original resistance, instead of just 0 with a solder bridge.

Gives you some finer control over it...

If you did have such a mythical POT wired in parallel, as you enthusiastically turn it to where its resistance approaches zero it would be providing the path for most of the current.
I would not want to contemplate the hazard of passing the bulk of such current through the wiper of a mechanical potentiometer.
Though since were're talking mythically, such a device could be imagined to have magnificent specifications so as to fulfill the role without burning.
The relationship between sensed current and angle of rotation of such mythical potentiometer would be very unlinear, with effect crowded at the near zero Ohms end.
The above hazard can of course be diminished, but why would I want to spoil your idea with longer sentences about engineering realities!

[VBCurtis]

Originally Posted by snme2pm1

If you did have such a mythical POT wired in parallel, as you enthusiastically turn it to where its resistance approaches zero it would be providing the path for most of the current.
I would not want to contemplate the hazard of passing the bulk of such current through the wiper of a mechanical potentiometer.
Though since were're talking mythically, such a device could be imagined to have magnificent specifications so as to fulfill the role without burning.
The relationship between sensed current and angle of rotation of such mythical potentiometer would be very unlinear, with effect crowded at the near zero Ohms end.
The above hazard can of course be diminished, but why would I want to spoil your idea with longer sentences about engineering realities!

I fear you confuse madpoo, who is optimistically imagining solutions he freely admits may not be practical in reality, with LaurV who tried to critique your opinion. Pretty harsh to attack a guy who already admitted he doesn't know EE who is just opining 'cause it's fun with the same "I am an engineer, you're an idiot" tone you used for LaurV.

[flagrantflowers]

Originally Posted by snme2pm1

Iwhere its resistance approaches zero it would be providing the path for most of the current.

Could you put a small resistor in series with the POT?

[VBCurtis]

Originally Posted by flagrantflowers

Could you put a small resistor in series with the POT?

Since the POT with the required properties is mythical, we can surely put an arbitrarily small resistor in series to achieve a lower bound on the resistance of our alternate path. Or, we could just set a minimum level on the mythical POT.

How much power do you want to allow your card to draw? Is your answer in mythical watts, or real heat-producing watts?

[LaurV]

Originally Posted by flagrantflowers

Could you put a small resistor in series with the POT?

How small? The original one is already 5 milliohm, and has no potentiometer, wiper (cursor) contact, soldering joints (between pot and eventual serial resistor).

(I hope that everybody understand that all this discussion is just "for fun", we all (myself included) are trying more or less ironically to push the limits of the ridiculous. In reality, as it was already said in the thread by different people, things are not done like that, by "adjusting" the shunt resistors, and I really hope that nobody will start hacking into his/her GTX Titan to "make it more powerful" in this way)

[Madpoo]

Originally Posted by LaurV

How small? The original one is already 5 milliohm, and has no potentiometer, wiper (cursor) contact, soldering joints (between pot and eventual serial resistor).

(I hope that everybody understand that all this discussion is just "for fun", we all (myself included) are trying more or less ironically to push the limits of the ridiculous. In reality, as it was already said in the thread by different people, things are not done like that, by "adjusting" the shunt resistors, and I really hope that nobody will start hacking into his/her GTX Titan to "make it more powerful" in this way)

I only hopped in because when I saw that you could change things by soldering a bridge across that shunt, the first thing that leapt to mind was why bring the resistance to ~zero when you could merely reduce it a bit.

I had no idea there's that much current flowing through there, and I kind of doubt there is or the shunt itself would need to be pretty beefy. All it's doing is providing a fixed voltage drop (a very small one since it's mOhms). Same power sensing stuff you'd find in a breaker box or a voltmeter for that matter.

Shorting across it means the V across the shunt drops to near zero instead of whatever it was, and I guess that reading is handled in the BIOS somewhere to do whatever.

Suggesting a POT isn't too insane, is it? Saying the wiper couldn't handle the current, even all the way down, seems strange given that a tiny solder bridge is expected to do just that in the original example. So either I'm not grasping the actual circuitry, or...?

As retina suggested also, if it's a matter of power handling, take the exact same resistor on the board and wire that in parallel... you've halved the voltage drop across the shunt which may give you finer control, compared to the brute force "short across it" approach... that's all I was saying originally. LOL

Anyway, hey, we're all just having fun here, we can be friends. It's theoretical to me especially... I don't have a GPU and probably will never bother unless one were given to me or was so cheap I *had* to buy it.

[LaurV]

Originally Posted by Madpoo

I only hopped in because when I saw that you could change things by soldering a bridge across that shunt, the first thing that leapt to mind was why bring the resistance to ~zero when you could merely reduce it a bit.

Which is a very recommended thing to do, and we all agree with you here. I said in the very post that reducing it to almost zero will give you headache to take care what you giving to that card to do. Remember, "increase the clock, watch the temp..." etc. Repeat every time when you change the work type, as TF and LL will stress the card differently. You should even take care when you change the exponent, because LL of a 38M expo and LL of a 70M expo will stress the card differently (amount of memory used). That was "do it on your own risk" type of advice. You may have the surprise that you do LL on some expo, set a clock and voltage which do not turn the card red hot, and forget it, but then the LL finishes and the new exponent uses a different FFT which makes the card red hot.

Of course, limiting the power is a common sense thing to do, that is why it is limited from the factory. Reducing the resistor is preferable to shorting it, from a thousand reasons. No one arguing with you here.

Originally Posted by Madpoo

I had no idea there's that much current flowing through there, and I kind of doubt there is or the shunt itself would need to be pretty beefy.

Actually not. Or well... partially, and that is the reason why the resistance is so small, to keep the power dissipation in check. The power (P=UI=U²/R=I²R) dissipated is like 20² (Amps sqared) * 0.005 (ohm)=2 Watt resistor, for a 12 Volt * 20Amps =240Watt consumption of the card. Actually, a 5W metal film SMT resistor is used, that is why I said in the very first post that "the two biggest resistors you can see" there. They are huge, indeed. The suggestion to add another resistor over was OK, and that I was initially saying, too. I only "picked on you" for the "adjustable" part, which can not be done, or at least, not by playing with the resistor. The rest is, as you said, "having fun". Forgot to stick the tongue to you!

[LaurV]

For the part with the wiper handling the current, think about the fact that the resistance is R=, i.e. is direct proportional with material resistivity and length, and inverse proportional with the section. The contact point between the wiper and the horseshoe of the POT (technically the contact point between a spherical surface and a cylindrical surface) the section is very small. That means high resistance, in the tenths of an ohm (or hundreds of milliohms, say tens of milliohm in the best case). That point where the contact is, will dissipate a power an order of magnitude higher than all the rest of the contraption we can improvise there. It will get red hot and smoke. Try taking a 1.5V AA battery or accu, and short it with sharp tweezers and see what happens with the sharp tip of the tweezers when you make contact. It is instantaneously red and smoking. And that is not 20 Amps. THIS is the problem with the wiper (cursor), the contact resistance. Movable part, which has a ball, rolling cylinder, whatever (or is made of carbon for pots with very high resistance, to slide, which is not our subject here). That is why I said "power switches" in my former post. Dip switch or jumper won't do it, for the same reason (contact resistance very high).

[TObject]

Shortly after these fabulous GPUs were introduced it was shown, on mersenneforum.org, that to yield reliable LL results, Titans have to be down clocked (particularly in the memory department). Is this now outdated information?

Otherwise, why all the talk about overclocking: to churn out incorrect residues faster?

Disclosure, I do not own a Titan, although I am interested now that a few are available on the used market, from gamers upgrading.

[chalsall]

Originally Posted by TObject

Is this now outdated information?

No.

[Madpoo]

Originally Posted by LaurV

...That point where the contact is, will dissipate a power an order of magnitude higher than all the rest of the contraption we can improvise there. It will get red hot and smoke. Try taking a 1.5V AA battery or accu, and short it with sharp tweezers and see what happens with the sharp tip of the tweezers when you make contact. It is instantaneously red and smoking. And that is not 20 Amps. THIS is the problem with the wiper (cursor), the contact resistance. Movable part, which has a ball, rolling cylinder, whatever (or is made of carbon for pots with very high resistance, to slide, which is not our subject here).

Yeah, true, I guess the small off-the-shelf POTs you'll normally find aren't that high quality.

For some reason in my mind, I was envisioning the type of system used for devices that maintain electrical contact while swiveling around or can infinitely rotate. Of course in those cases I think they use concentric copper rings that maintain solid contact, not just at a single point.

Fair enough criticism of my idea. How about this one then... instead of a solder bridge, just a "whatever" length of wire, where the length is determined by what resistance you want? After all, we're talking about some small resistance anyway, so a coil of 20 feet of whatever gauge copper, perhaps? As long as there's no AC component to the power (there shouldn't be on a 12VDC regulated supply), I don't imagine a coil of wire would pose any issues besides an interesting electromagnetic effect. Might even wind it to counter that effect.

It's been a (long) while since I've calculated the resistance of wire based on the various factors (cmil, length, material, etc). I honestly have no idea what length at what gauge would approximate a few milli-Ω similar to the resistor already there.

Oh well. Like you said, you really have to be trying hard and willing to risk a few things to even bother with any of this. LOL More power to you (pun intended).