My diesim

[JFettig]

Hey guys, I got my die simulator made, its the squairish one in th back behind bigbens. I now have to figure out how I want to insulate this thing.


A little more info here: http://www.wbta.us/forums/index.php?showtopic=50&st=0


Jon

[pHaestus]

Is the hole for temp probe in the center of your die?

Bill suggests phenolic resin 2" thick for insulation. Pricey though!

[BillA]

and yesterday my MCW6000 played copper kettle, 177°C

PAY ATTENTION

[Anemone]

ouch!

So you were playing with a dual core Prescott? LOL

[BillA]

no, I have burned up/melted down this poor abused heat die 3 or 4 times
sometimes equip failure, this time my (operator) error - no circulation, boiled then blew up
but yea, high heat loads

[Incoherent]

Originally Posted by JFettig

Hey guys, I got my die simulator made, its the squairish one in th back behind bigbens. I now have to figure out how I want to insulate this thing.
A little more info here: http://www.wbta.us/forums/index.php?showtopic=50&st=0


Jon

Jon, you used 1/2" x 1/2", what is the sensor location relative to die surface?
Are we any nearer to agreeing on a standard die size, sensor location, copper appears to be the material of choice.
.

[BillA]

if that hole is close to the die face, it will have a definite impact on the heat flux
better would have been to test w/o a hole, establish a baseline, then drill or grove the die and re-measure for the offset

Inchoerent
I would suggest not using inputs from 'the geek' to establish a consensus re testing equip and procedures
Jon's program/aspirations/intended accuracy is somewhat limited, much more so than with pH or JoeC (let alone Swiftech - and now other mfgrs are doing sophisticated testing as well)

[Les]

Originally Posted by Incoherent

Jon, you used 1/2" x 1/2", what is the sensor location relative to die surface?
Are we any nearer to agreeing on a standard die size, sensor location, copper appears to be the material of choice.
.

Unresistered
Is your "sensor location relative to die surface" still 2mm?
This needs to be standardised

[BillA]

some guys who know a whole lot more than I do about testing looked at the die setup and concluded that it was reasonable - for what it is intended to do; provide a 'calibrated' heat source

no assessment was made of the accuracy or correspondence between what I measure as the 'die temp' and the actual die face temperature, just the measurement's repeatability

an offset ledge can be used as I do, or a groove cut on the face, or a hole drilled as Jon did
all will work, each will have a different temp offset

I think, before dimensions, an agreement is needed (or not ??) on where to place the sensor
one might also ask agreement between whom ?

[Les]

Originally Posted by unregistered

I think, before dimensions, an agreement is needed (or not ??) on where to place the sensor
one might also ask agreement between whom ?

I would agree to anywhere
IF, the position is indicated
Which you do not

[bigben2k]

FWIW, some data here:
http://forums.procooling.com/vbb/showthread.php?t=9445

The issue is partly the distance, and partly the diameter of the temp probe.

[Incoherent]

Originally Posted by unregistered

if that hole is close to the die face, it will have a definite impact on the heat flux
better would have been to test w/o a hole, establish a baseline, then drill or grove the die and re-measure for the offset

Inchoerent
I would suggest not using inputs from 'the geek' to establish a consensus re testing equip and procedures
Jon's program/aspirations/intended accuracy is somewhat limited, much more so than with pH or JoeC (let alone Swiftech - and now other mfgrs are doing sophisticated testing as well)

I just want to make a die sim which is as similar as possible to as many others as possible so that some sort of cross comparison can be made of any given block I might make. Right now I am using a Barton sized die, but I suspect few others are. I am fishing for some kind of "between those who are interested, standard", which I desperately hope pHaestus ends up using, since his setup will be definitive.

[pHaestus]

Incoherent:
I agree on all counts

I am inclined to go with the "copy BillA" approach, simply because he already has a fully functioning setup and he is available for discussion.

[BillA]

I think pH is close to having a die made, multiple copies ?

[pHaestus]

sure why not? The more the merrier; right? I think JFettig even volunteered to make 'em

[BillA]

pH
I've found several missing dimensions in the die dwg I sent you, I'll need to get them the next time I have a wb change - towards the end of the week ?

Les - why I did not define the RTD location/distance

[Les]

Incoherent
Are you USA. Europe or New Zealand based?
Possibly all 3.
Whichever, you are only occasionally stupid
Far from stupid this time

[pHaestus]

Whichever, you are only occasionally stupid

Truly praise of the highest order!

I thought Sweden? Guess he gets around a lot though...

[Incoherent]

Originally Posted by pHaestus

I am inclined to go with the "copy BillA" approach, simply because he already has a fully functioning setup and he is available for discussion.

Being 10x10mm? That would be sensible I think, the existance of a lot of pretty reliable data is a big plus too. The only down side is the tendency recently for dies to get bigger, which might swing the comparisons around a bit. Still, it's almost a worst case scenario for a block (small die) so ....

Originally Posted by Les

Incoherent
Are you USA. Europe or New Zealand based?
Possibly all 3...

Actually I need to update my details. I'm a Kiwi, resident of Sweden and was working in the States on a 1 1/2 year contract up until this year. Now very much living in Sweden.

[Cathar]

It would seem to me that lately the trend will be for larger dies with all these dual-cored CPU's on the horizon, at least until the chip mfgrs fix their shrinkage issues. >200mm^2 is near unviable though for immature processes which is what the CPU makers deal with since they are the ones pushing the boundaries more than most, so 140-160mm^2 as a middle ground? Say 12x12mm, or 12.5x12.5mm? Intel seems to like square dies lately, whereas AMD of late seem to be going highly rectangular. :shrug:

[bigben2k]

What diameter probes are you going to use pHaestus?

[Les]

The 10x10mm seems sensible
Can be related to Bill's data

Incoherent
Note your update signature update, cricket and RL?

[BillA]

one could consider the size as irrelevant if better units were used with "°C/W",
as in °C/W/cm²

do we have any data demonstrating wb performance differences between a larger and smaller die ?
(I do not, have not run as an independant variable)
Cathar ?

[Les]

Originally Posted by unregistered

one could consider the size as irrelevant if better units were used with "°C/W",
as in °C/W/cm²

No and No.

Statement of die size is definrtive

Using °C/W/cm² is not

[Incoherent]

Originally Posted by Les

Note; update, cricket and RL?

Have totally lost track of both League and Union, the games are too short to catch through multiple time-zones over the last ten years or so. Avid cricket follower where possible, often hard to find coverage. It should be easier now though, some stability finally.

Originally Posted by unregistered

one could consider the size as irrelevant if better units were used with "°C/W",
as in °C/W/cm²

Not so sure about that Bill, a block could be optimised for a specific size.
Anybody on for a thermistor matrix in a diesim block? Could be used in reverse to analyse the (un)evenness of the heat distribution over the CPU die too. Calibrate that!