Heated die simulator II
 

BillA and pH, I've read the latest on OC vis a vis die simulators; the big copper block with the small die footprint. Is this still, in your opinion, the best configuration for a heated die simulator? Are there better ways to deal with the area heat issues than having the entire mount heated?

Bump :)

I'm having thoughts of using something other than copper: for some reason, I forgot why we're using copper, and I came across some info about a steel "flux bloc", in my Googling.

A steel block wouldn't have the same thermal properties, but may help in having a finished surface that doesn't require as much maintenance, not sure. Thinking about a hard plating here.

Anyone?

no, have you read the relevant threads in the "Water Cooling Simulator Discussion" section ? (scroll to the bottom and select From the: Beginning)
lots there
my old stuff (pages gone from H)
http://www.thermal-management-testing.com/Hdie1.htm
http://www.thermal-management-testing.com/Hdie2.htm
pic of die modded to take a 0.083" OD RTD
http://thermal-management-testing.co...die%20wRTD.jpg

pH has something going for a die,
I'm trying to scrounge up a bit of 2" thick phenolic sheet
- anyone ???

Sorry, Bill I looked at the wrong thread and then looked at the "linkage" from procooling which took me to overclockers.com. It looked like the copper block with heater and insulation was the tech to use. Have you or pH considered using an evacuated chamber for your die simulator? I mean it would solve the issues of insulation to the WB. And it would check for leaks. I do approve of the RTD - which vendor are you using? I know I need to learn more - thanks for teaching me. I'll get back to my background reading now.

I'm having problems sourcing RTDs
the one in the photo was ordered 4-wire, came 3, and I think the 'rework was just to splice in the 4th lead outside the sheath
- not many want to jack with 4 wires in that sheath
I tried to order from consens, but never received anything (because I asked for certs with alpha and beta values defined ??)

did a bunch of work looking at a vacuum box, porting, etc.
did not seem to be worth the effort, the whole thing would have to be inside an environmental chamber - whew !

I didn't think it would need to be in an enviromental chamber since the radiated thermal energy is so much less (10-3) than the conducted energy. The problem was the thermal breaks on the heater leads and the temp control on the lab chiller. nicht wahr?

hmmm
my equipment room is 85 to 95°F (the chillers are unhappy @ 95), so the vacuum box and contents would also be at 85-95 ? no ?

breaks are not so bad with 4-wire (load sense circuit) psus
with the present environmental chamber everything in and out is 4-wire (3 psus, 4 RTDs, except for 6 TC channels), and the chiller control uses the coolant inlet temp in LabVIEW to control a Kepco SN488-122, one channel of which controls the chiller
(wired, not yet operational)

a vacuum box - to me - is a huge amount of work to gain what ?
|absolute| knowledge of the heat input ?
but as long as I use the same die (insulation), my 'error' is constant

I deal with this right now in the testing of wbs and hsfs
wbs are on the heat die, hsfs are on an Intel ttv
the 'efficacy' of the heat die I can ascertain, with the ttv nothing (no means of measuring the actual dissipation, air flow and mean temp)

one part of the C/W 'mess'

All I've got :-

"mldhab


Registered: Jan 1970
Location:
Posts:

quote:
--------------------------------------------------------------------------------
Originally posted by X-Silver
mldhab
Please do post more pics
--------------------------------------------------------------------------------



OK you asked for it

OK, here are a few pictures:

First is the heat die housing, broken apart:

http://www.jr001b4751.pwp.blueyonder.co.uk/Housing.jpg

Here is a picture of the pieces put together:


http://www.jr001b4751.pwp.blueyonder.co.uk/Redcross.jpg (Red cross, Les)

this is the housing with an unmachined copper slug in place:

http://www.jr001b4751.pwp.blueyonder.co.uk/Housing3.jpg

This is the machined heat die slug, the green arrow shows a .050" hole for a thermocouple in a small shelf to the side of the die face, the upper surface of the slug is relieved to limit its contact with the upper assembly cover, the bore for the heater is centered lengthwise in the copper slug:

http://www.jr001b4751.pwp.blueyonder.co.uk/DieSlug.jpg

This is a heater (two wires for heater power, other two wires for thermocouple in the heater):

http://www.jr001b4751.pwp.blueyonder.co.uk/Heater.jpg

You can search the different Overclocking oriented sites and come up with several different approaches to 'Heat Die' construction.

After analyzing all the different approaches we could find, BillA and I came to the conclusion that nothing we saw met our requirement for an accurate method of measuring the energy input to the waterblock’s baseplate surface. This is primarily because of the inability to differentiate between:
1), the energy input to the heat die assembly and
2), the energy input to the waterblock’s baseplate

There are undoubtedly some circumstances which only require throwing a bunch of energy at a waterblock or heatsink, but the kind of analysis we wanted to accomplish requires an accurate, quantitative accounting of the energy at several points in the water cooling system, under various conditions. Needless to say, knowing exactly how much energy is being put into the system is of paramount importance to our ability to accomplish these studies.

We have both been accused of being anal and overly scientific (as if that were a bad word) about how we approach these things. In justification, I can only say that Bill and I come from environments that require a fairly extreme level of accuracy and total repeatability in any given analysis. When you work professionally in this kind of environment - finding, focusing upon, and figuring out how to limit the 'weak links' in an analysis regime is the first thing you try to accomplish. We independently came to the conclusion that the 'heat die' was that 'weak link' in a water-cooling analysis. A search for a solution to this 'problem' is the reason we have been working together for some time now.

No tool is ever exactly perfect for the job it is used for, but I believe we have come closer to having 'the right tool for the job' with this heat die than anything else I have been able to find. And, I certainly expect that the search for the 'perfect solution' will make this tool better, in its future iterations. Read that to mean that any suggestions are always welcome.

Consider trying to quantify the difference in efficiency of the heatdie to heatsink interface when checking two different thermal compounds. Inaccuracy of the energy measurements, in this circumstance, is magnified to the point where (at some point) the analysis becomes meaningless. So far, this heat die has kept us at a level of accuracy that makes the product of the work we've done worthwhile.


Last edited by mldhab on 11-16-2001 at 11:40 PM "

talcum
Ben and others did some speculation re a vacuum box, need a link to the thread ?
Ben ?

Can't find the thread, sorry, but some info here:
http://www.wbta.us/forums/index.php?showtopic=5

and

http://www.wbta.us/forums/index.php?showtopic=41

In short, maintaining a vacuum is actually incredibly difficult. The main issue is that you will more than likely be facing some kind of leak.

To make up for that leak, you'd have to maintain the vacuum with an active pump, i.e. an electric vacuum pump. Then...

When selecting a vacuum pump, you'll need to make sure that it has the volume capacity to cover the leak.

Of course you don't want any leaks in the first place, because it will throw off your results, to some extent.

So I thought it best to use an o-ring around the heat die, to seal that opening where the heat die protrudes out of the box. Otherwise, you'd need vacuum proof electrical connectors, to run the heater wiring.

Otherwise, I'll be adding a pressure gauge, just to get a rough idea about wether the vacuum level is maintained or not, or if there's any kind of leak failure. This involves threading a connector to connect a brass cross to the side of the insulated box, so that three connections are available:
1-the pressure gauge
2-the relief valve (there's no sense in maintaining a vacuum when not used)
3-the pump inlet

I'll also be coating the inside of the insulation box with a reflective mylar, just in case radiated heat turns out to be a factor (probably a complete waste of time).

Still in the works, so otherwise at your own risk. At this point, I have everything but the insulation material. The heat die is being fabricated by Jon.

After reading their thread, I'd recommend they check the Huntington and Ceramaseal sites for vacuum feedthroughs and Duniway Stockroom for vacuum equipment. They really want a convectron or Pirani gauge to check pressure. That will get them down to under 5 millitorr. They will also want to use soft copper tubing and swagelocks for their water fittings. I'd suggest a 10" dia chamber with Ceramaseal feedthroughs using a turbo pump with cold trap to rough. They might be able to get by with a Viton O-ring seal for millitorr pressure but CF gaskets should be used wherever possible.
Sorry, I started thinking like a lab rat again. It is possible, but not cheap or practical. Sorry for wasting your time.

You mean millibar, right?

Thanks for the tips. I realize that I won't achieve the highest of quality here, but I'll certainly give it a try and report results.

CF gasket? (I'm sure it's going to hit me in a few minutes)

"cold trap" ?

No, torr as in 760 torr to a bar or a millitorr is about 10E-6 atmosphere. CF stands for conflat (it's a type of UHV gasket using a metal to metal seal) And cold trap is a baffle for the pumped gas prior to the pump which is filled with LN2 to remove water from the pumped volume before it contaminates the pump.

I'm sorry I did misunderstand though. I meant put the entire device, die simulator and waterblock, under vacuum. That way there's no convective cooling from either the WB or the die. But Bill's right, it's not practical unless you work at a lab and do the work after hours.

Ive been looking here there and well almost anywhere and various insulation schemes are mentioned, but Ive yet to see anyone employ a simple vaccum chamber, sure youd need a "custom" top for each of the hose penetration changes, and youd have to address keeping the tubes from collapsing under a vaccum, but its insulative value would be quite high.

just thinking out loud, like what I might use an AC vaccum pump for when not drawing a vac on an AC. :p

of course youd still want to insulate it to cut down on radiative heat transfer
http://www.opticorp.com/IR_low.htm ???

bigben2k is actually working on a vacuum insulated die sim.

aha, so he is

thanx for the clue ;)
Id missed the forums over there, now Ive just registered

Nice to know that someone is interested; I was starting to think that it was a wacky idea!

Yeah, it's still in the works, as originally planned. See the forums here for more info.

yeh thats what they told James Dewar :p
Makes perfect sense to me ;)

I gather your trying to make just the die simulator in a vaccum?
(except for the face)
or are you tossing the whole kit and caboodle in a chamber?

Just the die simulator.

If I can get the equipment though (aka a bell jar), I'm not past running a few test series with the whole thing in a vaccum, in an effort to quantify the secondary losses, but that would be a one time thing only.

There's so much work involved, just getting started, characterizing the whole test bench and all, before any actual testing is done... but I'll get through it.

see I was thinking a good strong box with a 1\2" acrylic top
the "ports" for the electrical pass through and vac being permanent and in the box body
and the top having the pass through for water tubes, you make as many tops as hose configs you have and suspend the die and block from the tubes, there being extra heater cartridge wire in the bottom of the box to pull it all out and work on.

leaving just a good tube connection (NPT vs barb) and tube support against the vaccum (an internal wire with a plastic tube would induce a different flow, but you could always use a reinforced braided or more rigid automotive rubber hose for the last 3 inches)

Im looking to buy one of those 1151's we where talking about in the other thread what is the smart range indicitive of?

actually making the whole thing out of thick acrylic would be nice theater
a little LED edge lighting too :p

Hum, yeah, like a transmission hose, or power steering hose; anything that's a high pressure hose, really. I see.

I haven't searched really hard yet, but I've been looking (on and off) for a vaccum pump. I've figured out that it should be able to pull a vaccum as close to 29.9" Hg, and have a flow rate that can make up for any leaks; that's the part that's stomping me right now. Either way, I have to repeat the same condition for each test series, so if all I can maintain is 26-27", then that's what I'll have to do.

I have no clue about the 1151 range question, but if I come up with something, I'll let you know.

see I figured this was a perfect matchup to doing phasechange work where you build your own, drawing a decent vacuum being a key component of charging a system properly

as far as the repeatability of the vacuum, well while thats nice as you get closer to a hard vacumm its also not as important having dimisihing returns in effect Id gather
one of those things where past a ceratin point "close enough for rock and roll" likely takes effect

thanx for your input on the 1151 guess I need to do more digging
thought it might be well known

http://grs-sei-amd64.stanford.edu/re...port_jan17.pdf

example
which is completely encased in a vaccum chamber at 0.037 tor

thats a TIM (Thermal Interface Material) Test rig at Stanford
look at the pretty pictures then dig down to page 11

damn, wrong button :p

oh well might as well do something here


http://tinypic.com/ego1vd.jpg


yours for a measly $24000 dollars :p

ditch the lasers (RTD only)
do the machining yourself, already own the data aquistion system have a constant temperature fluid setup already, pickup used goodies off ebay a ghetto DIY around $5000?

they detail the parts with prices
I happen to already have an autoclave\vaccum chamber that might hold a smaller rig like that, but a simpler acrylic box to just isolate a die simulator would be the place to start

The lazer is for the thickness measurement of the tim material, since nothing else would measure 1 um with any kind of usefull accuracy.

While the load measure is good, it actually does leave out the TIM thickness.

Interesting; I'll have to study it some more. I like the flux block implementation; 5 RTDs in a row, on each side!