100 sq mm Fluxdie worklog.
4 Attachment(s)
Finally been able to start work on the evolution of the fluxblock.
I am just posting these pictures to try and keep some kind of record of progress, also a strong motivating factor having this progress public. Measurement data is a fair way away yet but having got to this stage I can start connecting up and calibrating thermistors. Initial intentions are to characterise the behaviour of this kind of fluxblock setup, get some numbers for the MCW6000 and do a series of TIM tests at this new (for me) die size of 10x10mm. After that we'll see what other blocks become available and start accumulating measurements. I am interested in evaluating the negative impact of heat spreaders for example. The theory is the same as with the earlier fluxblock but the arrangement of the sensors is different. Instead of a die with a single sensor, there are two thermistors spaced 5mm apart, the top one 2mm from the die surface. The fluxblock is now only 4mm high (12mm before) and has only one, centred, sensor instead of three. There are three reasons for these changes. 1. The experience with the previous arrangement showed that my accuracy was enough to use a closer spacing of fewer sensors. This also has the effect of 2. reducing the total heat path length, enabling me to keep the temperature of the heater down at a safer level especially with the smaller cross sectional area of the 100mm^2 vs 144mm^2. 3. This arrangement makes it easier for me to keep the Fluxblock aligned, despite it being smaller and fiddlier. The Lexan clamping plate also serves as the primary insulator. Except for four clamping feet there is no contact with the die, there is a minimum of 1mm air gap between the die surfaces and the inside of the insulation clamp. I am in two minds regarding the Lexan. The previous die ended up embedded inside a bubbled, melted lexan block due to an unscheduled meltdown. If the temperature is kept below ~140-170°C it maintains its integrity. This is not particularly high but at the same time I have not been able to find any rigid material that comes close to polycarbonate in terms of low thermal conductivity. I have decided to go with it for now, mainly because I have some and I like its transparency. I have been doing some (very) simple modelling of this setup and I am confident it will work OK. More will become clear over the next few days/weeks. |
That is looking good. I am planning a similar version and was also going to use lexan being it is all I have other than wood.
Are you still going to use the resistor for the heater? |
What about that fibre-glass board stuff? You know, that brown wood-like appearing stuff that they use in foundries to hold various bits and pieces. Incredibly heat resistant, strong, and low thermal conductivity.
|
Phenolic resin? If so I can't find it anywhere.
|
Rang my machinists and had a chat.
They recommended to me this blue heat-shield plastic that is used in molding processes. Very high compressive strength, and overall a very rigid plastic they assured me. "High Temperature Insulator Sheets" "Asbestos Free Glass Reinforced Polymer Composite" Made by: D-M-E Company 29111 Stevenson Hwy Madison Heights Michigan 48071 Thermal Conductivity: 1.9 BTU/hr/ft2/in-F @ 75F, and 2.1 @ 225F 550F maximum working temp. Will convert to metric when I get a chance. |
By my reckoning, it works out to around
0.069 W/m-K, thermal conductivity in metric, which puts it into the same realm as polystyrene (styrofoam) for insulative purposes. |
Quote:
1.9 BTU/hr/ft2/in/F = 0.274 W/mc about the same as wood (oak~ 0.2 W/mc(Kryotherm)) |
Seems to me lexan is not that far off. 1.35 BTU @ 2F. Delrin is 1.6 BTU @ 2F
|
Quote:
|
Quote:
1 BTU/hr = 0.2931 W Watt/hr being summarised as merely Watts, as per convention 1.9 x 0.2931 = 0.55689 This is per foot squared, but W/mK is per meter squared, so over a larger area, need to multiply above by m^2 divided by ft^2 3.2808' per 1m 10.764ft^2 per 1m^2 0.55689 x 10.764 = 5.994 Now tranmission distance is per inch, but in metric, it's per meter, so need to divide by the number of inches per meter. 5.994 / 39.37 = 0.15225 Now difference is per degree F, but metric is per C. Need to divide by the number of F per C, which is 1.8 0.15225 / 1.8 = 0.0846 Okay, so that's a different value to above, but still styrofoam sort of values. 0.0846 W/m-K 5 points to who can point out what's wrong with the above. There probably is something wrong is all I'm saying. |
Ok, got it wrong. Meant to multiply by F per C.
0.15225 x 1.8 = 0.274W/m-K Which is what Les worked out. ~Wood sorts of levels then. |
How strong does it need to be?
Expanded PVC sheet possibly? Not sure what kinda thinkness you can get. |
Quote:
The material needs to be fairly strong, I don't think expanded foams would cut it. I have considered wood, oak would not be totally unsuitable in my opinion, it's thermal properties are rather good considering but I worry about fire and water. Quote:
Lexan starts softening about 160-170°C has thermal conductivity of 0.21W/m*°C. Mechanically this sounds interesting Cather. One disadvantage is that materials containing glass tend to wear tools out, but thats not a huge deal. If I can find some of this I'll probably use it. Until then I'm going with Lexan. |
hmm . . .
have we not been here before ? having burned up a heat die multiple times, I suggest that phenolic laminate is the cost effective material of choice it will char overnight, but it will not burn; I used foil faced rigid urethane around it |
Quote:
Meantime I intend to incorporate some form of overheat protection. Good intentions, we'll see if I actually do that. :) |
You've got a lower limitation: your resistor's operating temp range maxes out at 155 C.
http://www.arcol.co.uk/pages/product...t/fpa500_6.pdf |
1 Attachment(s)
Quote:
Well aware of that Ben, my target operating temperature is below 100°. The problem is to avoid what happens when things go horribly wrong... |
Quote:
(a place I could visit on the west side of Houston, where I need to go next week anyway) size(s) ? |
Well, I am impatient.
I got the baseplate made today and have whacked everything together and mounted the MCW6000. Nothing but the water temperature sensors are calibrated, and them with an old (pre meltdown) calibration. So far the waterblock behaves fantastically, keeping the die colder than the water with a C/W of -0.07°C/W. Awesome! ;) ... Obviously some work needed but the test shows that everything is working as it should, the clamping plate works well, the fluxblock alignment is excellent (no more buggering around with trying to keep it lined up whilst "docking" the WB), the sensor cable arrangement is better, the die seems good flatness-wise. Issues are heat, I am not happy with it. I may shave off some mm from the base of the die to reduce the heat path length. I inserted the thermistors without thermal compound because it was a pain to push them in, air gets trapped; having modelled a steady state scenario where it was unnecessary. Of course in reality response time suffers, in this case greatly, I liked the very fast respose of the earlier fluxblock, so I will redo them. Things to do: Reduce die height and reshape to reduce secondary losses. Reshape clamping plate to accomodate this. Reinsert thermistors with thermal compound, maybe drill a breather hole. Calibrate thermistors, 0-100° Rescript software to function with new thermistor arrangement. Fine tune calculations to allow for heat shadowing and other non linearities. Model this. A fair bit of work left to do before meaningful results but it seems to be progressing well. Edit. Bill you have PM. . |
Quote:
|
Glad you know!
jd: it's limiting because it's at the far end of the heater. the temperature increases from the top of the heat die, right up to the heater. While we'll all try to keep the die temp low, depending on the length of the die, I can imagine the temperature at the other being able to reach 150 deg C. It's not an issue if the heat die is pretty slim (as it is here). |
2 Attachment(s)
Quote:
Quote:
Although area is OK was in grave doubts about Fin efficiency values. Ditched modifying "Kryotherm to deal with pins" . New Model Use Flomerics to calculate "h(convection)" Consider "Pin Free Area"(Apfa) and "Pin Area"(Apa) separately, use Fin Efficiency(Fe) Calculator and calculate h(eff) --- h(eff)= (h(conv)*Apfa + Fe*h(conv)*Apa)/(Base Area) Convert "h(effective)" to Rwb using Waterloo Convert Rwb to C/W(T in) using Kryotherm.. Attached predictions for the MCW6000 on 12x12mm and 10x10mm dies Frightening coincidence with data. Not sure whether am using correct dimensions for the MCW6000 :- MCW6000 60x60x5mm bp, 281 (9x2x2mm) pins with 6mm(ID) inlet Modeled as 60x60x5mm bp, 289(9x2x2)pins in17x17 array Used Flomerics D=0.006, h=0.009,and r=0.02 to calculate "h(convection)" . "r=0.02" chosen as value to give minimum Rwb value(beermat Iteration?) - See "Pick & Mix". The problem is Waterloo only handles isoflux. Edit: Attached an extract of Excel (Hope it opens) |
Quote:
|
Quote:
|
0
|
All times are GMT -5. The time now is 05:50 PM. |
Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk... Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...