Ultimate water block tester

[Annirak]

[EDIT: Seems that inductive heating of the base-plate is a viable technique after all. Definitely cheaper than a laser, and requires no optics.]

Some people will shy away from this idea because of the requirements involved, but considering the premium being demanded by Intel for the TTV, it seems reasonable to me.

Here is my proposal:
Since mounting seems to be such an issue surrounding water block testing, let's take it right out of the picture. We need a device which can apply heat to a target with no physical contact.

The devices which are useable are obvious at this point: they are any device which creates electromagnetic radiation. The problem that many forms of EM have, however, is how well they are absorbed by metal. (e.g. spork in a microwave = sparks, not hot spork)

Several devices come to mind, but many share that common problem. We're left with the laser, and several less likely devices (gamma ray emitters, grasers, and serveral others. X-rays are no use, they go right through)

I eventually came to the conclusion that a laser, particularly an IR laser, is the way to go.

And we're in luck! CO2 IR cutting lasers are in ready supply for industrial situations. If we can get a 100W IR laser, we should be ready to fly.

So, here's the setup:

1)Paint the bottom of the Device Under Test (DUT) matte black. (no, this will *not* interfere with the transferal of heat to the DUT)

2)Affix a thermocouple which has had its joint--if not its entire body--painted shiny silver to the centre of the DUT. It will have negligible effect on the radiation absorbed by the DUT, as it is a small percentage (0.1% ideally) of the surface area of the block.

3) We really have 2 options for the laser spot. 1: put the DUT in a totally reflective box, with insulation on the outside. 2: ensure truely MATTE BLACK coating of the mating surface, and use black-body radiation calculations to determine heat loss through IR radiation.

4) Measure water temps in the usual way to get Delta-T.

As long as the power output of the laser is consistant and the IR reflectivity of the paint is consistant, no one can argue that there is variation in the thermal interface. There isn't one.

IMHO this would allow a particular internal design to be tested regardless of mounting techniques. WBs could be evaluated for performance without concerns about mounting techniques. It's the way of the future, guys, let's get those lasers going!

[Brians256]

You know.... I'm not really sure if you're serious or not. There's been some crazy things flying around in this forum.

ObHumor: In other news, Jonathan Swift has a great proposal on how to deal with overpopulation in Ireland!

[Annirak]

Actually, I am serious. It would provide reproducable, accurate values for testing the performance of a waterblock, with no concern for mounting problems.

Does anyone have a better suggestion for eliminating mounting/TIM variances from test results?

[Annirak]

It's possible to obtain a CO2 laser up to 60kW. 100W is so small that there are laser diodes that will do it. This shouldn't be a problem.

It's direct application of a known thermal quantity... why shouldn't we try this? It can't be worse than what we have... And it'd eliminate the complex test cycle that requires multiple mounts, etc.

[Annirak]

Here's a "for instance" on the laser source:
YLR-LP series lasers
There's a 300W model with 10-100% output power tuneability, and +/-3% long term output power stability.
+/-3% is a whole lot better than any of the test-beds we've seen yet.

[bigben2k]

Ok, it's not a joke.

Is it practical though? What is the cost of such a laser arrangement?

[Annirak]

The one I linked? Probably more than is worth our while. Though I'd imagine 10k would get you one.

I never said this was a cheap solution, just a good one. I'll keep looking into sources for IR lasers.

[Albigger]

Originally Posted by bigben2k

Ok, it's not a joke.

Is it practical though? What is the cost of such a laser arrangement?

It definately seems practical and potentially very beneficial over current testing methods.

Cost could be an issue.


SpotIR by research inc has a 250W model (not sure if you can tune the output or not, would think so, they have a range of products)
http://www.researchinc.com/thermal/products/spotIR.htm

Still wondering on cost though...


Also would have to validate the wattage input and the black coating I think. Maybe get an insulated chunk of copper (known dimensions) paint bottom, and measure temp as a function of time to ensure what wattage is actually being absorbed by the block?

[Annirak]

Parallax whole system laser pricing
Parallax laser tube only pricing
OEM laser system pricing

Not definitive, there are sure to be others, but a start.

[Annirak]

Validating wattage isn't as important as ensuring consistency. That could be as simple as keeping a test-block for validation, and re-testing it every time. Measurements could then be done with respect to the test-block.

[Albigger]

Originally Posted by Annirak

Validating wattage isn't as important as ensuring consistency. That could be as simple as keeping a test-block for validation, and re-testing it every time. Measurements could then be done with respect to the test-block.

Wattage important if a goal is true C/W numbers, but consistency HAS to be there regardless.

[Joe]

While this is an extremely interesting idea (and completely outside the box thinking which I admire), my question is wouldnt this amplify the "non realworld" crap that we see already when it comes to testing? I mean its really completely different than what a user would see as a "realistic" test.

But it would be interesting to see this test done, and compare it against results people see with a given Die Sim.

I guess another question is where are you going to get optics whipped up that will shine the laser light in a 14mm square patch on the bottom of the device to be tested?

[Annirak]

Joe, the point behind this is that it allows people to determine the real performance of the internals of the block--irrespective of mounting. It doesn't make sense to be the sole test, but it is the best sanity check I can think of. This kind of testing gives non-platform-specific results.

To simulate die size? A standard lense to widen the beam, and a 14mm square aperture in a reflective shield applied over the center of the block. That has to go with the reflective box technique.

[Joe]

also there would still need to be calibration to figure out exactly what % of the light energy is transformed into thermal energy by the matte black target. Since not 100% will be transfered.

[Annirak]

That's a falacy. I specified an IR laser for precisely that reason. IR (particularly far IR like the ones I'm suggesting) essentially IS thermal energy.

[Joe]

So your saying that 0% of the light would be refracted away from the source by the lensing, that 0% would be reflected off the target... I don't buy that.

It takes some special paint to have 0% reflectivity, and to absorb it all. I know they make it for special instruments, but its not something you buy from the store.

Don't take this for a flame, I am just trying to look at all the areas that could cause trouble... I am a big proponent of debunking something right away to remove questions later on.

[Annirak]

You're right. Some is going to be reflected away. I don't honestly know how much. That was why I specified the shiny silver insulated box. It will get reflected/refracted away, but since the only non-shiny surface in the shiny box is the waterblock, you end up with all of it getting absorbed by the waterblock, no?

There is still the point that it's wise to do callibration runs. I have nothing against doing a callibration run. The simplest I can think of is the thermal rise on an insulated block of copper approach.

[Albigger]

Originally Posted by Annirak

...There is still the point that it's wise to do callibration runs. I have nothing against doing a callibration run. The simplest I can think of is the thermal rise on an insulated block of copper approach.

Can't think of a reason NOT to do this. Would be very simple and relatively low cost (get a copper block machined/ground to good dimensions with a few thermal probes) - at least relative low cost compared to the laser.

I suggest multiple probes because the heat distribution won't be the same throughout the block while the laser is on. Alternatively or in conjunction, maybe put the laser on for a certain amount of time and then turn off. Allow the (still insulated) block to come to temperature equilibrium and from the overall heat rise of the block and time the laser was on calculate the wattage.


some irony - I've noticed at least a few of these lasers are or can be purchased with water cooling systems

[Voodoo Rufus]

This is a pretty cool idea with some potential I think. Could an IR thermal sensor be used for temperature sensing so as to avoid contact entirely for thermal measurements, or do they not come that accurate?

*waits to read more on this subject*

[Joe]

Photonics isnt my point of interest mostly, so I dont know much about hot to measure light output from a source in watts' to ensure your laser is running at 100% power.

I think this should deffinitely be tried, its a very cool idea. This would deffinitely make testing a very multi talented field... were now you dont only need to be good at electronics, plumbing, data gathering, etc... Now you also need to be good at photonics, surface coatings, etc...

Wild stuff!!

[Albigger]

Originally Posted by Voodoo Rufus

This is a pretty cool idea with some potential I think. Could an IR thermal sensor be used for temperature sensing so as to avoid contact entirely for thermal measurements, or do they not come that accurate?

*waits to read more on this subject*

I think they come that accurate, but could be wrong. also anything between say the sensor and the device you are taking the temp of can influence the measurement, but if we're going all out here why not put the whole waterblock in a vacuum? That way ensure no loss to ambient and the laser doesn't need a medium to travel through anyway.


But, in order to maybe keep this viable (affordable) I wouldn't think thats the road to go down. Also, in concert with maybe lowering the cost, I have a question:

Was going to wait to bring this up but since Joe brought up the light->energy thing:

can not any bulb (provided enough power) be enclosed with a high-emissivity "reflective box" to focus energy onto a waterblock base? Does it HAVE to be a laser? Surely cheaper than a laser but would definately require (constant) calibration, if even possible.

[Incoherent]

LOL

Good idea.

Your enemy is cost.

The optics required to get a square spot focused on the base of the block are not trivial but are certainly doable, the requirements are not severe.

Biggest problem is establishing how much energy you are actually placing on the block. Certainly possible to measure it.

Go for it.

[Joe]

Also something I would want to look at is atmosphere for the test. you may need to do it in a helium env, or some other light inert gas as a thicker atmosphere could impact a test on a given day. Based on humidity, etc...

I know lots of laser testing done for power generated by the light of a laser needs to happen in a Vacuum or some other very light and dry atmosphere.

[Annirak]

Do a callibration run before/after each test, measure humidity levels while in progress?

[Albigger]

Originally Posted by Annirak

Do a callibration run before/after each test, measure humidity levels while in progress?

Could but requires more measurements/calibrations.


If making a 'reflective box' anyway, may not be too difficult to add a fill bleed valve to fill the chamber with whatever gas is available (or pump the chamber empty)? Seal the ends against the waterblock and the other against the laser somehow?