Innovatek Block

[Joe]

The thermal elements I am looking at have an array of thermal sensors across the plate. with monitoring them, knowing the total heat output, knowing the contact patch size, and knowing that the setup is 100% insulated, you can get the actual heat density and load thats being sent to the HSF. From there you can get what the face temp is.

"your friend Chip gave me some Cusil, need it ?"

hehehehe Friend.. yeh... just about as much of a friend as... well lets not get into that hehe Besides, why would I want to deprive Mr. CuSil of your goods?

I also got some of his CuSil... you see how it performed in that WBRU correct? ... awesome material ( well shitty block design more like it)

[BillA]

"an array of thermal sensors across the plate" - you're after the heat flux rate ?

it seems more direct to accurately measure voltage and current (simultaneously), that's the applied power
- assuming no (or very low) losses, that's then also the inputted power

"knowing the total heat output, knowing the contact patch size, and knowing that the setup is 100% insulated, you can get the actual heat density and load thats being sent to the HSF. From there you can get what the face temp is"

sure about that calc ?
R(subscript)thetaJA will yield a junction temp, but you're talking about calculating the substrate temp, one thermal interface removed from the sensors

I do a bunch of this kind of testing, but I sure would not attempt the calcs (with any degree of confidence)
good luck

[Joe]

Its months off into the future anyway. I got some more critical things my moneys going into besides a simulator to test 40$ HSF's


Yes the math isnt perfect, but I also havent sat down to work it all out either, this was all just off the top of my head.

Edit: I thought your tag line was "be cool"

[EMC2]

Originally posted by Joe
I am guessing this has changed again since then cause reading from under the core on the new XP/MP cpu's is pretty pointless (and from all measurements so is the thermal diode)

Curious why you made the comment about the thermal diode? (I agree with the under the core completely).

Why not put a probe right on the long side of the die edge like this:

[EMC2]

And then isolate it like this:

(admittedly not as accurate as you might be able to get with a die simulator, but inexpensive and "in system")

[Joe]

I am not sure about what thermocouple you are using, but I dont have ANY and have never seen one small enough to sit next to a core and still have room for goop and stuff to hold it there. Normally they are connected via fairly stiff solid conductor wires which makes it hard to keep from moving around.

its a nice idea but unfortunately I somewhat doubt you can get a T/C thats small enough and able to be secured well enough.

[EMC2]

You can actually get them very small and fairly inexpensive. A good example of the small size is the TC probes that ASUS sells for use on their m/bs (most have an extra input cnx on the m/b for them).

The bond at the die side isn't for the main holding force, it's strictly for thermal interface purposes. You epoxy (prefered) or superglue them at two spots along the substrate surface away from the die. The lead stiffness actually then helps in holding the end in place.

Oh, and the other trick is to put a strain relief tie to the m/b past the h/s body to prevent accidently "snag and jerk" olympics.

[Brad]

just buy an AXP and do a smbus mod

[Joe]

wait wait wait...

Asus boards do NOT use thermocouples... they use Thermistors. dony confuse the 2 as they are totally different devices. Also last I heard Asus boards were so inaccurate that if you got within 20 C of your real temp you are on a roll.

[BillA]

rain on the parade has arrived """"""""""""""

Joe's comment on TCs vs. thermisters is correct - but does not go far enough
(I not referring to the Asus board reference, thats a different (bad) deal by itself)

do a spot of work with the manufacturers of these devices:
you want to know the devices' and instruments' "stated accuracy"
(for a type T TC it is +/- 0.5°C or 0.4%, whichever is greater, thermisters run arount 1.5^C, and much greater - to 3^C)
THEN, to this you must add the effect of the meter's "stated accuracy" (these are calculations, not sums)
and the numbers will render damned near moot your measurements

I just had a BIG conversation with Omega's engn dept about 1 tenth degree variations, and they laughed at me

so do appreciate that the VERY BEST POSSIBLE temps you might read are +/- 1^C

valid readings of greater accuracy are possible, but ever more difficult
+/- 0.1^C will cost ~ $500 (with USED equip, RTDs only)
+/- 0.01^C will cost ~ $1500 (with USED equip, $2500 is more probable)
[I am now using sub-minature custom made - and calibrated - 4-wire RTDs,
and a calibrated meter with 0.001^C resolution; to get 0.01^C accuracy]

so EVERY TIME you see "34.6^C" think Bull Shit
the real number might be between 32 and 37

[EMC2]

Who said anything about actually using the m/b circuitry for the die temp measurements? The additional comment about the hookup was why ASUS made them, the ASUS reference was an example of a thermistor that was small enough to illustrate the point that you can get them that small.

Regarding accuracy - yes it is true that thermistors can not be had that are as accurate as top line 4-wire RTDs. However calibration to 0.5C for a select range is quite feasible for the temperature range needed, the key word being calibration

Unregistered - regarding your .001C calibrated instrument giving you .01C accuracy - as we both know it's an established calibration standard that you need TEN TIMES the accuracy for your calibration standard. So your numbers don't surprise me at all. And yes, I have talked to Omega on more than one occasion over the years. We regularly require accurate temp measurements of electronics at work with sub 1F for some our projects.

As far as thermistor accuracy (stated) that can be had, I suggest you go look through your Omega Temperature catalog at thermistors again or talk to Omega (hint, TENTHS of a degree C for the element itself).

[EMC2]

BTW Joe - you never did comment on why you made the comment regarding using the embedded thermal diode of the XP.

ps. BTW, TC's stink for accuracy. Their only positive point are inexpensive and a very wide temperature range.

[Brad]

I also don't like reviews where they use an in socket probe, and decide one heatsink beat another because one is 1C better

[BillA]

EMC2

in fact my coolant temps are read with thermisters, YSI 700 series at $170 a pop
even with these however the mfgrs stated accuracy is +/-0.2^C
but by probe selection and matching,
and individual calibration to a specific indicator (2 Digitec 5810s with 0.01^C resolution),
better can indeed be done

now please explain what ANY of this (or your referring me to an Omega catalog)
has to do with the thermistors that OCers are actually using ?

(hint, nothing)

or did I misunderstand, and it was Joe you were referring to the catalog ?
one's finger can be made to work, the real problem is a total absence of calibration

EDIT: too harsh a tone to this post
the real problem with OCers' temps is, as EMC2 observed, (no) calibration
the numbers are pie-in-the-sky,
comparisons are of the nature mine-is-bigger-than-yours,
and calculations based on these 'temps' are trash

and yet another of my poor jokes:
the Overclockers' Mantra

all temps are crap, but Watts are worse
ooooohhhhhmmmmmmmmmmmm

[surfinguru]

So Schoolie, are you only using the waterblock or the entire kit? If you're using the kit, how would you rate it overall? (Ease of installation, quality, etc.) Just curious as I'm expecting my kit to arrive Friday. Thanks!

[schoolie]

I'm only using the InnovalCool block. Sorry, I can't comment on the kit, but I think I've seen review floating around.

[surfinguru]

Well, thanks anyway for the quick reply!

[EMC2]

Joe (and unreg) - Who said anything about actually using the ASUS m/b inputs for accurately reading the temps?

The ASUS thermistor probe was used as an example of small thermistors with the additional comment about their m/bs as an aside (notice that part was in parentheses like this) as to why ASUS sold them. Go back and read my post, notice I said, "a good example of the small size". (admittedly I did in my haste and sleepy mode put TC, my error)


Unregistered -

The conversation does not have anything to do with what the typical OCer is using for temp measurements.

I started the discussion and posted the drawings as a result of Joe's post above where he was talking about measurement methods, AMD, thermal die simulators, etc.. The intended application was for Joe's testing of waterblocks. Let's stay in that context please


And yes, I was referring you back to Omega. I did that only as a result of your statement regarding accuracy of thermistors, your recent discussion with Omega, and your comment

that the VERY BEST POSSIBLE temps you might read are +/- 1^C

However, I see you have since backed down from that position ( I would like to see you cal your finger to +/-0.5C, lol )

( I saw the un-editted version unreg... we'll let it go )


Back to the discussion regarding methods of temp measurements for testing purposes -

In the case of waterblock testing Joe, if you really wanted to get an accurate measurement of the delta temp between the water going into the block and the die temp, you could use a wheatstone bridge circuit with two thermistors in the bridge. (of course, as a frabricated method - to save dollars)

ps. Joe, in a few minutes I'll send you something that shows you why I prefer resistive temp measurement probes over TCs.

[Joe]

Originally posted by EMC2
BTW Joe - you never did comment on why you made the comment regarding using the embedded thermal diode of the XP.

ps. BTW, TC's stink for accuracy. Their only positive point are inexpensive and a very wide temperature range.

Wow man, who busted out the crack today???

T/C's stink?? I can get some lab calibrated T/C's with +/- .01C accuracy. I can get standard grade T/C's with +/- .1C that are calibrated....

You think Thermistors are accurate??? ahhahahahahahahah they change value by how much wire or solder is used to connect them to the reader.

EMC, I have seen 4 XP's read 4 different temps on the same board, same Vcore, same HSF, same thermal paste and same ambient temp. By a differences of 5C +/- Same batch XP's even Same thermal paste application method, same fan RPM.

SO you tell me whats so accurate about them? Also what about AMD's claim that they havea +/- tollerance of 3Deg C?

T/C's are not accurate.. hahah thanks that was funny

Edit: I am talking about Thermistors that Geeks are using just FYI, I know lab Thermistors, Thermal Diodes, Thermal Transitors, adn T/C's are in an entirely different ball park than what geeks use.

[EMC2]

You have to look at the entire measurement system Joe, not JUST the sensor. TC's require an extremely accurate backend (read the PM I sent you earlier).

Regarding the wire/solder comment - I think you are getting 2-wire RDTs and thermistors mixed up there RDTs run in the 100 to 140 ohm range for sub 100C temps. So yes, unless they are 4-wire RDT's to allow for correction of lead resistance, there is an issue.

Thermistors on the other hand are in the Kohm range (some in the 50K - 100K ohm range). Wire length and solder joints have almost zero affect then. And TCs have thier own entire set of warts regarding hookup and system environment induced errors.

Regarding the XP's thermal diode readings - how was the diode being read?

Absolute accuracy aside for a moment - the discussion (at least on my part) was about you doing waterblock testing. If you are comparing waterblocks, using the same processor and setup, along with accurate measurements of the inlet water temp to the block... True, you may not be able to do any calculations that require absolute temps, but you can do valid comparisons.

I haven't seen any AMD specs on the temp accuracy of the onboard diode. The only spec I've seen them put out is for the ideality factor of the onboard thermal diode, 1.002 to 1.016 (that is a bit wide, but "normal").

As far as what "geeks" may or may not be using, as I said before, at least on my part the discussion was centered on methods for waterblock testing... a different realm I would hope

[Brad]

/me is going to say the same thing again

if you are using one athlon XP, for all tests, I doubt it can be out by too much from heatsink to heatsink

[Joe]

Brad I still say testing with something thats clearly stated to have no better accuracy than 3C +/- is like testing with a compunurse next to the core....

If you test that way, then please dont give me temps with X.x since the .x of the degree mark will be totally error, as the thermal diode is incapable of that kind of resolution.

EMC2... you remind me of somone else I know in many ways ( stubborn arrogance )

If you are running a ThermoCouple its a given that you are using something calibrated on the back end to measure it accurately. Dont post saying how "hard" it is to read it accurate, since you CAN read it accurate with the right gear. Just cause the tolleraces are close doesnt mean its any less of a testing product, just means you cant be such a monkey with the testing gear.

What about Thermistors... a coating of oxide on the connections, a cold solder joint, different wire length... All can throw off the readings. Also I have seen Neon bulbs throw off some Thermistors by 10 Deg C! ( given not with professional thermistor gear but with the geek stuff)

The XP diode is read via SMBus things... ( talk to pH hes the man with that).

[BillA]

"EMC2... you remind me of somone else I know in many ways ( stubborn arrogance"
watch it there EMC2, you're on thin ice now . . . .

EMC2 (and any others interested),
I have a TC measurement analysis ? as part of a rather involved thread

care to comment ?

[EMC2]

Joe - we've both presented our views via pm and I've given my reasons why I hold my view. Basically at this point, we have different views. May be better to leave it at that on this topic after this. As such, last comments for our discussion by me

Closer tolerances mean smaller error margins and greater likelyhood of errors induced into the measurements by external factors. Note also I never said TCs were invalid for use as a testing method.

I already commented on your remarks regarding thermistor readings being affected by wire length, etc. in the PM. When using the proper thermistors, they don't have any more affect than they do with good quality TCs, for the technical reasons I gave you in the pm.

Take your thermocouple and solder connections between the probe wires and the meter leads and you will can have errors induced, if it is not done properly. Use the wrong type of wire for the connections and you really have problems. Don't use an isothermal block to make the two connections at and you have problems.

Don't on the one hand (thermistors) apply "ghetto" equipment and setup conditions while on the other hand (thermocouples) apply high quality lab grade equipment and setup conditions. You should at least set consistent conditions

As far as who I might remind you of - that's a two way street (albeight a different street)


Unregistered - I have the impression from comments I've seen on these boards that he won't ban someone because they have a disagreement with him or get involved in a discussion with him as long as it stays out of the personal realm and remains rational

On your other thread - maybe tonight. Have to go earn some $

[pHaestus]

There are some issues with testing with XP diodes, no doubt. For the kind of testing that BillA and Joe are talking about, they are not suitable. In fact, there is no way to test performance of waterblocks with a processor reliably. You are going to have to move to synthetic setups to get any hope of reproducibility.

Since the XP diode readers seem to keep surfacing in this thread, I will now talk frankly about the diode readers that can be made from maxim ICs. There a few obstacles that can stand in the way of their use in testing, some surmountable and some not.

For one thing, can they be calibrated? I think that the best way might be something like the following: Take your CPU and connect it to the reader, then move that CPU into water of different known temperatures. You should be able to construct a reasonable calibration that way. Now comes the kicker: you'll have to code some software that is capable of accepting this calibration equation rather than just an offset if it isn't simply a constant. Do the calibration after removing and replacing the wires: does it change?

Next, carefully look at your lead wires from CPU to reader. You will find that their length and their gauge likely will affect the readings you get. You won't be able to notice this until you try testing a heatsink at a wide range of heat loads. You'll notice that if your leads are much too long then the relationship between deltaT and heat load will not be constant. I know that C/W values in temselves aren't particularly reliable, but you'll find if you shorten the leads then they become constant. The shift in the deltaT for having leads too long is easily as large as the difference in heatsinks. Can you set the system up perfectly? Can you do it several times in a row? And how closely can you build the reader to the Maxim specs? Can you compare numbers with other people? Probably not unless they all get their diode reader from the same manufacturer. And have the same lead wires. Of the same length. Does the kind of solder on the socket pins matter? This is getting fun...

Here is something more troubling: my experience is that the diode temperature is lower by 4-5C than the temperature of a DOW probe epoxied under the core of a ceramic Duron "Morgan". How can this be that the temperature is hotter away from the source than at the source? It isn't exclusive to my reader: Joe C's $150 maxim test board shows the same phenomenon. Why? I am still thinking about that one.

And finally, to really complicate things, while the diode reader isn't good enough for serious testing, it is good enough to pick up on inconsistencies in methodology. Apply the thermal paste differently? Use more pressure on one of the replicates? All that will show up. And since the response time is very rapid and the resolution of the readers is only 1C (unless you hack the SMBus drivers for lower sampling rate) then the effect is marked. Is this a good thing? The effects after all were always there we just couldn't see them before. Can we come up with truly consistent paste application? Maybe Joe and Bill have thoughts...

Having explained why the diode readers are not suitable for high end testing, I still think that they are far superior to ANY of the alternatives available to overclockers. It picks up on temp changes from things like dragging windows, a bad application of thermal paste, a kink in a hose, and many other variables that the in socket thermiostor or a compunurse cannot. Its temperature compression is also quite a bit less than either a thermistor on the side of the core or the insocket probe. And since it connects to SMBus, you can easily add in a relay connected to the OverTMP function on the reader and turn off power to the computer (PWR_GND) if the CPU gets above a certain point.

As far as variability in the diodes of chips themselves, I haven't ovbserved that. I have tested out I guess 4 different XP CPUs (3 1600+ and a 1900+) and at 1400 MHz and 1600MHz (1.85V) they all produced similar delta Ts (within 1C) when using the same heatsink. Now that is just anecdotal evidence mind you but it certainly is more consistent than you will get when comparing different CPUs with a compunurse that you have to pull off and reattach.

I don't understand how Joe, Joe C, BillA, and other serious testers keep at it. Bill's adage is true:

"All temps are crap...."

And yet it is the job of a few unfortunates to try and make sense of them. The choice of some hardware sites to review neon lights instead starts to make sense. Maybe it is best to simply evaluate cooling gear based upon how far you can throw it?