Test Bench V2. Want your suggestions!
 

I learned a hell of a lot building my first test bench. However I am not satisfied with it's construction and usability. What I want suggestions on is:

1) die sim construction
2) acceptable flow meter
3) acceptable pressure meter

and anything else you might want to chime in on.

Equipment I have right now:

1) Cole Parmer Thermometer
2) Cystal Fonts CF633
3) Gen-X Mak4 1190GPH Pump

That is pretty much everything I have that I could use in the new bench. I have a 24VDC 15 amp power supply (2 12V in series) and a little 12V 3amp supply.

My current die sim is in the attached pic. I have a YSI 703 probe on there on the right. That is 5/32" diameter and is just to big. I want to pick up some 423's or similar. Should be a little better. I would raise the step the probe is in above the rest of the die slug to better represent die temp. On the left is a CF633 probe for redundancy. I have a watlow 1/4" x 2" 40VDC 150watt max heater cartridge in there running at 24VDC to get approx 70watts. I would like to change this to a AC heater cartridge found on mcmaster and make it variable from 50watts to 150watts. I havn't figured out how to do that with a AC cartridge though. I seem to remember some saying they can be run DC though?????

My radiator fan combo is a 1990 ford F-150 heater core with a cardboard shroud with a 86CFM Panaflow 120mm fan. It actually works pretty good at keeping water temp approx 2C over ambient.

I need input on pressure differential measurement. I have pondered using a manometer? Something like this: http://testproducts.com/safecart/pro...18fa3505d2b804
Can that be used with water though? I assume water will never get the meter anyway. I have little idea how to best hook that up. That is the least expensive route I can think of.

Once I get a drawing of the new die sim I have in mind I will post it here. I am also wanting to know what die size to go with? 10mm? 11mm? 12mm?

The dP meter you found would be fine (no water doesn't get into it, if you mount it right), but it has two drawbacks:
1-it's battery operated.
2-It doesn't have an external output.

There are plenty of similar units on eBay that would do the same, and more, and are cheaper.

You want to watch out for the resolution, because a resolution of 0.1 PSI is going to be pretty useless. (this one can do 0.01 inches water, so it's fine).

The range is also important; this one is max 60 inches WC (~2 psi), which is good, but personally, I'd like it to be able to handle more, i.e. ~180 inches, or 5 psi, for those odd very restrictive blocks, at high flow rates (assuming your pump can handle it).

There's a critical piece of information missing about this product (no surprise there); the response time i.e. how long it takes for it to register a change (of X magnitude) in the pressure. I like to know what it is, because I'd like to quantify any oscillation (aka variations), on the 60Hz frequency, along with the resonent ones at 120 Hz and above, if possible. Entirely optional.

Otherwise, you can check the calibration yourself with a good old fashion manometer ("U" tube, always advisable; it should be part of your checklist before the start of each testing session).

I favor a model with an output, because I'm looking to log all of the data with a PC, but it's really up to you.

jd
1) I sent you the dimensions of my 'old' heating die
after testing with it and an 'industry' test vehicle, junk my heater die
sorry for the bad news
2) a 1/2" magnetic flow meter
3) a dP transmitter, Foxboro 823DP, Rosemount 1151, etc.

you need to take a longer range view
http://cgi.ebay.com/ws/eBayISAPI.dll...e=STRK:MEWN:IT
http://cgi.ebay.com/ws/eBayISAPI.dll...e=STRK:MEWN:IT
all purchased recently

I have been looking at the Rosemount 1151 for a long time now as they are easy to find. i am going to assume the DP on the model number means differential pressure? I see many model numbers....

How much loss of accuracy is a 1" mag flow tube over a 1/2"? I can find 1" pretty easily but rarely come across a 1/2"...

Any idea's on a useful heat die? Is the copper slug/heater cartridge combo still ok to use or is it just the design of your old one?

Attached is a drawing of my current die sim slug. I thought it worked out OK but I would have rather had the probe step above and isolated from the rest of the slug body. Once I get a revised drawing I will post it.

1" is too big
that is the heater die design I no longer like
raise the pedestal, add 2 or 3 very small holes for thermocouples
(you are making a calorimeter)

Attached is the revisions I was going to do except the holes for the probes would have been 1/8". Is that something closer to what you suggest? I was already going to install 2 probes, were would a 3rd go? Also should the holes go into those steps or be drilled right into the pedestal itself?

This is also going to require a different thermometer than the one I have. It only exepts YSI 400, 500, and 700 series probes. Smallest one's I can find are 1/8". I hear the 500 series has smaller but I cannot find them anywhere.

So the list so far:
1) Thermometer with 1/16" or smaller tc's?
2) Differential Pressure transmitter (probably Rosemount 1151)
3) 1/2" magnetic flow tube flowmeter.
4) Redesigned die sim

Jaydee I would suggest something like this (see attachment) which is what I think Bill is driving at. (?)

It is a variation on the fluxblock concept and something that Groth came up with independently a while ago. It will give you a rather accurate idea of the face temperature. Adding the "mini Fluxblock" as shown in the drawing gives the same ability that I have with the fluxblock, namely the ability to measure TIM performance. The small holes are for thermistors/thermocouples/whatever's your poison, the big ones for heater cartridges.

I would recommend a 12mm die because I am biased, that's what I have been using:) I would like others to do the same for data comparison reasons

Yes,think should be standardised.
Am understanding Billa still uses 10x10mm Die(Link).As shown in attachments, this leaves me perplexed. Think is a priority to sort.

Prescott 1: 112mm
Prescott 2: 135mm
Venice: 84mm
San Diego: 115mm

110mm would be about the average size but dual core cpus confound the problem with their size and their rectangular cores.

Glad to see expressions of a standardized (edit: "guideline") heat die area!

Can either of you (Incoherent / Les) comment further on a 12 by 12 versus other sizes? In other words, how would you justify it?

Since I'm still building my rig, would there be an interest in a test series of a number of blocks under different size dies? I had planned to run a series at 10 by 10 and 12 by 12. I'd consider adding another size, if there's enough interest.

My preference would be 10x10mm,biased because most of my modeling based on 10x10.Additionally the 10x10 may be more representative, than larger sizes, of the possible heat-flux concentration(s) from a CPU.
I, certainly,would be interested in "a test series of a number of blocks under different size dies" - Link

Your new diagram shows something that's 1 1/4" tall; I think Bill was trying to hint at something that's slimmer.

I'm assuming also, that Bill would rather see a shorter heater cartridge, but I can't tell.

Are we trying to make something like RoboTechs die sim? http://www.leesspace.com/WB-Testing.htm

Maybe higher rise with smaller and more holes for probes?

2 cartridges? Can anyone explain how to hook a AC cartridge to run on DC? I would like to very the wattage from 50watts to 150watts. If I can do this with AC all the better as all the cartridges I can find are AC. What length to use? Smaller the better?

I can build more than one size die sim. I can do a 10x10, 12x12 and 14x14.

As I understand, heater cartridges are a simple resistor. should only care about voltage and current rating.

AC heater probably means the wattage they spec is when used at normal line voltage.

Go to http://www.mcmaster.com/ and page 454 shows how they are listed.

I guarantee that you will be able to use these heaters with DC Jaydee. Problem is is that the resistances of these are rather high so you need a high voltage DC power supply to get the power output you want.
Using several of them with AC in different configurations would give you some control. eg 3 50W heaters gives 50W, 100W, 150W and then 1 + (2 in series) gives 75W, 3 in series gives 16.7W. This is a bit of a stupid way to do it though with complicated switching so you don't short circuit your power supply by accident. A Variac (a variable transformer) would be ideal for this but I suspect they aren't cheap and it is difficult to measure power accurately.
Low resistance heaters using DC are far easier to control, thats why I use these, but I would actually suggest something like this with a STRONG caveat that you have some kind of over temperature protection and design the heat die so that there is a minimum of material between the heater and the tested heat exchanger to keep the temperature of the load at a minimum (they have different ratings at different temperatures.)

That power resistor is iteresting. Does anyone have any objection on using them?

power resistors complicate the secondary loss control
use heater cartridges on DC
a 450W AC cartridge will need ~70VDC @ 3 or 4 A

using AC will not permit the precision you want

That's why I wanted to use DC in the first place.

This new bench is going to take some time to gather up. Money is not falling off trees right now. First thing I am going to pick up is a dP transmitter and use it in my current bench.

Hi.

The system not yet works, but at the moment, I am using a resistance similar to the commented by Incoherent. 10Ohms 10x10mm core.
If you wanted you could diminish the losses better isolating it, cutting the zone of leftover copper etc. The heat point is closely together of the water block, in this point I have de LM35 sensor.

I do not want to occupy much space with the images, I put the Link to you of the forum.

In the beginning of this message, there is an image of the internal part of the resistance. She is very thin. At the final, the first assembly.

page 1

And here possibly the final system without still mounting the electronics. I have taken advantage of to use the electronics of an electronic balance to measure the force of tightens.

page 2

Nice work dnkroz.

Google suggests you are using titanium. ?. Is that true? I like what you've done.

10x10 vs 12 x 12.

Probably best to go for 10x10, it is probably more representative of future dies plus the points that Les has already mentioned.
When I get up and running again I will use both I think. I had a meltdown which imbedded my previous die into a lexan plate, the next version will be similar to what I have pictured above.
I have another move to complete and a social life to rebuild in the meantime.

Conroe at 65nm will be Intel's workhorse for a good while, and is said to be around 125 to 150mm^2, so that is a good ball park. At least I would not go higher, as larger chips are unlikely, for the desktop anyway.

Edit: I also agree with Les about smaller dies. 90% of the heat on the die si coming from < 50% of its surface area. With a simulator, its probably much more evenly distributed, so a smaller simulator die then the processor being represented makes sense.

Great work dnkroz!

I really like the block mounting plate that you have.

Is that a pressure sensor under the heat die?