An Open Letter to "Feathers"

[feathers]

Uploaded for your viewing pleasure..

http://www.icetec-uk.com/icetec/temp/melted_3-1.htm

I've only recently switched to a new CPU and that was only ordered because I thought the 2.4ghz had been damaged by a Swiftech MCX4000 heatsink.

For your info:

The DD Maze 3-1 was melted on the day after Christmas. Didn't have a spare waterblock to replace it and so I had to fit a Thermalright SLK heatsink. Eventually I installed a Swiftech MCX4000 air heatsink to test with peltier. It worked upto a point but of course air-cooling of peltier isn't very good. While making adjustments to the MCX4000 it crushed the CPU and bent some of it's pins. After that the CPU didn't work. A replacement was ordered (3ghz HT 800) and when it arrived.. It too didn't work! Fault was traced to an incorrect setting in bios and so it transpired that both CPUs were working!

I put the new one up for sale on Ebay (cos I wasn't impressed by it's overclocking). Got to the last day of auction and then I cancelled and decided to give the new chip another try. Ran it for a week at 3.5ghz @ default core voltage and now it's been increased to 3.7ghz.

The old P4 2.4 is gathering dust. It's one tough little CPU and it's happy to run at 3.4 so long as it's kept cool.

The waterblock was melted by a 130 W peltier.

Now using a DD Maze 4-1.

[bigben2k]

Originally Posted by feathers

BigBen2k:

Can you give more info on: "I'd mention the "Cascade SS" (one of 5?, which I have) and "Cascade XXX" (in development) water blocks" ?

...

"The other issue is that while pelt cooling is indeed extreme, it's really not energy efficient, and a phase change system is much more desirable, in that aspect." - Agreed... But a suitable peltier and PSU can be bought for £100.

Nice site! ( www.icetek-uk.com )


Going over the info you posted, you caught a lot of good info. You did however miss one point: the cold plate (aka spacer) purpose is to allow you to clamp the peltier to the block, and is required because that clamping pressure is beyond what a CPU can handle. That aside, you got the other purpose of the cold plate correct. (more info on this here: http://www.procooling.com/articles/h...__just_how.php )

You also appear to have missed that Intel specs call for an operational temperature of no less than the freezing point (not a concern if you're an enthusiast, but it is if you offer related products).

As for the cost, I was referring to the operational cost, not the purchase cost. Yes, you can slap something together pretty cheap, but the power usage is much higher than with a phase change system. A pelt can require more power than it can move, while a phase change unit would only require 1/3 of the power it moves. In most extremes, you use both.

The condensation issue has the potential to ruin a CPU and/or a motherboard, if no proper precautions have been taken. You may have gotten lucky, but it is quite possible to loose one or the other, to a single drop of water. For that reason, we recomend coating the whole motherboard and related peripheral cards with "conformal coating" (well, I recomend it!), as detailed here:
http://www.procooling.com/articles/h...per_coolin.php

That aside, water cooling will still outperform an air cooling solution, as has been measurably demonstrated here:
http://www.overclockers.com/articles373/wbsum.asp
and here:
http://www.overclockers.com/articles876/

(So to say that water cooling isn't worth it, is "innacurate"). There's also the quiet aspect of water cooling, that is so attractive, over an air cooled one.


The "Cascade" water block idea first appeared here:
http://forums.procooling.com/vbb/showthread.php?t=6101

then evolved through here:
http://forums.procooling.com/vbb/showthread.php?t=6666
(and in the OCAU Forums)

and is available here:
http://www.employees.org/~slf/lrwb/

Stew (Cathar) and I talked about it, and aside from the "XXX" version that's in development, it's not going to be beaten any time soon (so we're both working secretly on different ideas... ).

[feathers]

Thanks for the comments, BigBen.

Constructive criticism is always appreciated. Your comments will certainly result in a few changes and additions to the Icetec website.

"but it is if you offer related products" - What products?

Re: condensation.. We've probably gone to the other extreme on the Icetek site of making it all seem as the there are no real risks (I suspect this is the result of a strong desire to redress the balance and to show people that peltier cooling isn't as difficult or risky as they were led to believe). I will add a piece on the dangers of water and even condensation to some extent. What I won't do is tell people that the slightest air-leak resulting in condensation will destroy their PC (as others have done). This is simply not our experience of minor condensation (nor of our customers). If customers follow the same steps we outline in our tutorial section then they should get a pretty good insulative seal.

The conformal coating is a good idea and it will certainly be added to our shop (as well as recommended to customers).

I followed the links for the jet waterblock. Very interesting. I've also bookmarked the link to the website and I will be reading up on it tomorrow (later today).

Thanks!

Raf Baker.

[BillA]

jeez, if Ben can go there, why not me too ?

feathers
your site is a SALES site
you have limited mfgrs data on ONE TEC, no more (on ANY item)
no performance charts, nothing

I am suspecting you work with P4s only
- where clock throttling is saving your butt, but you don't even know it !

may I be so cruel as to suggest you put your 130W (Qmax) TEC on a OCed Athlon ?

feathers
you have come here to bs us about your technical expertise, but I am not buying
and your ignorance is so vast you do not 'see' why a 130W TEC cannot cool a P4 w/o throttling

proof ? -> here is some genuine TEC data


the MCW5002-T uses a 225W TEC
the MCW5360-T uses a 360W TEC

now you on your site are going to tell us that a 130W TEC is good to go ?

you need to trash your puffed up verbiage and start dealing with real numbers

(pleeze folks, the MCW5360-T is an 'industrial' product not available to OCers)

EDIT:
I should add that the "heat die load, Watts" is I*E, for 'Radiate' kinds of CPU Wattage increase by ~30% the values shown

[bigben2k]

I haven't even gotten to that part!

Yes, a 130W TEC is insufficient. Proof? Check out the icetek Forums, and the user that complains about wild temp swings, and performance under par. Heck there was a time when one was happy with an 80W TEC, but that was even before I came here! Bill's point is that you ought to demonstrate (as I do poorly) that you understand the numbers behind it all. Last I checked, DangerDen was selling 170 and 220W TECs (checking... they sell 226, 169, 80, 120 and 172).


Today's heatloads easily hit 80W, without any OC, and can hit upwards of 120 Watts. A TEC runs under volted (as is required, for the sake of efficiency), on top of which only a portion of its heat transfer power run efficiently (what did we estimate? 70 to 80%?) simply because of the application.

The "products" I was referring to, is what you are selling: you should demonstrate that you are aware of the thermal limitations (and properties) of the processors which your product proposes to cool.


(Bill: I thought you said that 3 gpm wasn't worth it? What's with the 11.4 lpm test results?)

[feathers]

The grumpy bastard was correct on a few points but wrong on many others.

We sell to both AMD and Intel users (although we only use Intel ourselves). We are aware of the heat problems associated with AMD CPUs (amongst many other problems) and we are aware that a TEC130 isn't suitable for ALL CPUs and as time goes by then we will have to switch to higher power units. This is why we're making arrangements to purchase devices upto 280 Watts from a Russian manufacturer.

The TEC130 has given very good performance since we first began selling it some time ago.

The performance data shown was based on an Intel CPU overclocked from 2.4 to 3.2

In case you hadn't noticed.. The website is new and the shop has only recently opened. This is the reason the stock-list is very small at the moment.

Those higher wattage peltiers are a priority 'yes'.
We will be adding them to our site 'yes'.
We will be selling 220W or higher pelts in future 'yes'.

It's interesting that UNREGISTERED (aka BillA) has chosen a SWIFTECH chart with which to make his point.

Swiftech went to some considerable expense to develop their air-cooled peltier heatsink (based on 220W pelt) only to discover that it's performance vs noise vs cooling was unacceptable (I know this from personal contact with Swiftech). No doubt BillA would have launched a similar attack on Swiftech for wasting time on such a flawed design?

"your site is a SALES site" - In your opinion. It's a sales site because it has a shop, right?

Do you seek to elevate your status within this community with that particular observation, Bill?

Or are you simply the next in line trying to establish dominance?

I suggest you learn to ease off and don't make too many assumptions.................

I am not going to apologise for owning a site which dares to sell computer hardware.

[feathers]

BigBen:

"The "products" I was referring to, is what you are selling: you should demonstrate that you are aware of the thermal limitations (and properties) of the processors which your product proposes to cool." - I can certainly agree with that. I am certainly aware that nothing remains the same for very long in this industry. I have worked very hard over the past few years.. Advertising, research, testing, contacting companies, acquiring new hardware, answering customer queries, preparing customer orders etc.. As I've said: The TEC130 has been selling for some considerable time but I am well aware that we can no longer recommend it for CPU cooling (I actually don't need a pompous ass to tell me that - BillA). Why don't we have these higher wattage peltiers available NOW? For the above reasons of workload and schedules. The site only went active last month. The shop only went active from March 1st. Since I am contact with different peltier suppliers then the decision has to be made as to which supplier best suits our needs. Coupled to this we are making arrangements to sell a new line of professional watercooling parts.

I should also point out for any other pompous arses out there that we don't only supply to computer users. Since we began selling peltiers several years ago we have been selling to universities and research labs. It is only over our more recent history that the emphasis turned to computer cooling.

(From this point on I will ignore any replies (such as BillA's) where the postee makes it clear that he is only interested in starting a fight. I don't have a problem with someone who raises issues in a helpful manner (always helps to add a touch of humour as well). I do have a problem with posts like yours BillA Unregistered. Even though you raise some relevant concerns - it is done solely with the intention of maintaining/elevating your status within the community).

I certainly welcome ideas, suggestions from well intentioned, friendly, lighthearted individuals.

[gruntledweasel]

Originally Posted by feathers

I do have a problem with posts like yours BillA Unregistered. Even though you raise some relevant concerns - it is done solely with the intention of maintaining/elevating your status within the community).

I suggest you dig through some old threads to find out why BillA has swiftech data readily available...and why he is qualified to be in such a position? I don't think he's one to get into a pissing contest to stroke his ego. (Although it is kinda funny to see someone lecturing Bill on swiftech products... ) Personally, I am much more worried about my own objectivity and methodology than Bill's. I'm not saying one should hang on his every word and data point or some such...but he's worth paying attention to. The day he endorses the testing methodology of one of those "9/10 stars editors choice" articles full of marketing fluff is the day I give away my watercooled pc and switch to an iMac.

Feathers, you may find these links helpful...
peltier engineering data
kryotherm software
intel p4 technical specs
p4 thermal design guidelines

Bill and Ben are right...130 watt pelt on a 80-90 watt heat source is not the best plan. If they're wrong, run the numbers and show them where they messed up. If there's a discrepancy between your experimental results and theoretical results, you should be able to pin down exactly where the two diverged. These theories you dismiss so readily were not simply pulled out of some engineer's ass. "Well, I got different results testing so you must be mistaken" is just not a convincing argument without a very clear and detailed writeup of the (sound) testing methodology that produced them. You will learn a lot if you try to understand the tech docs and do the math...I've still got a lot to learn, but I know I'm glad I spent the time trying to wrap my head around that first url I linked. The forums here are also a very helpful resource.

Perhaps the thread should be moved? Whether we have a constructive discussion or not, it doesn't seem to belong in the heatsink/fan fourm.

[feathers]

BillA may have valid points but he also jumped to some wrong conclusions about me. I guess he's a 'shoot first kind of a guy'. I'm not too concerned about his background when his foreground attitude sucks.

Ok.. 130 W on an 80-90 watt heat-source works very well thank you (highest temp=ambient)

The 130W gives great performance for a 60W heat-source (ie P4 2.4ghz HT 800 and even when O/Cd to 3.2ghz) = 19c after 10 x CPU burn-in cycles on Sisoft Sandra.

The 130W gives good results on an 80W core and acceptable (sub-ambient to ambient) on a 90W core.

Where I concur with Bill & Ben is that beyond 90W the peltier struggles.

I have taken measurments for the 130W from a core power range of 60W to 120W.

I use Sisoft Sandra's CPU+Multimedia Burn-In test to establish the effectiveness of the peltier. I run the Burn-in for at least 10 cycles generally.

What's important to understand is that the Sisoft Sandra Burn-in for CPU puts the CPU through a much tougher assault course than your average 3d computer game.

I have loads of Burn-in tests on my system extending back over the past 12 months. All tests conducted on the 130W peltier with the CPU overclocked to 3.2 or higher and with core voltages from default upto 1.9 volts.

I have more recent tests recorded by a CPU logging program.. Tests which show the CPU temps during some of the latest 3d games (Chaser, UT2003 etc). CPU= 3ghz P4 800 HT O/Cd to 3.5ghz. If memory serves me correct then it's core will generate approx 70w at the default voltage? The highest temp recorded during gameplay was 14.xx celsius. Not bad for a 130W pelt, right?

If anyone would like to see all of these performance tests then I can upload them or send them out in zipped archives?

"These theories you dismiss so readily were not simply pulled out of some engineer's ass. "Well, I got different results testing so you must be mistaken" is just not a convincing argument without a very clear and detailed writeup of the (sound) testing methodology that produced them. You will learn a lot if you try to understand the tech docs and do the math.." - Agreed.. I think the wording on my site needs to be changed. Also the comments about "Watercooling being a waste of time without peltier" - was too extreme.

It's ironic that while some of you interpret my site simply as a sales site from someone with no regard for facts and figures.. I actually do follow theory, data and rules to some extent.

My wording (icetec site) suggests that I don't place any value on rules or guidelines. What you should understand is that the site isn't just a means to sell products.. It was born out of a genuine determination to cut through some of the bullshit out there.. But as you can clearly see.. I have taken this desire to break free - too far in many ways!

Your comments (the genuinely helpful, positive ones) will certainly be listened to and used to make the website more accurate in some areas.

What I won't be doing is feeding others with endless warnings about the dangers of peltier cooling (or any form of extreme cooling!).

The point about the 130 w peltier is certainly a priority issue for me (it was an issue even before I came to the forum.. Why? Because I actually DO research the latest CPUs in order to learn about their O/C capability and their wattage!).

I value constructive criticism and your words will have an impact on my website (and on my own approach as well).

If anyone wants to see the test data? Just ask.

EDIT: I do have some of last year's Sandra data on my website. The charts and CPU temps are clearly shown (there are two links on my site which show that data!)

EDIT x 2: "I use Sisoft Sandra's CPU+Multimedia Burn-In test to establish the effectiveness of the peltier. I run the Burn-in for at least 10 cycles generally." - This is the point I try to make on my site about "wishy-washy theory" (It seems
that you now have the impression that I regard all testing/data as useless).

What I actually meant was that if all you do is theorise.. You're not likely to get very far (and certainly the negativity surrounding peltier cooling would suggest that this is so). What I DIDN'T MEAN TO SUGGEST is that testing and performance data are useless. I spend MANY HOURS TESTING and recording the data. You will find some of that data on my website (I have much more on my Hdisk!).

The bottom line is that the icetec site is now too extreme and too anti-establishment (too many personal feelings and bitter resentment directed at those peltier scaremongers (they used to scare the crap out of me).

The performance fiqures I quote in this reply are not theory or make believe. They are measurements I have taken myself over a long period of time.

EDIT x 3: I have looked into Phase-Change cooling in the past. I would love to try it at some stage. Also interested in those cascade waterblocks (purely for their clever design and beautiful workpersonship).

[AntiBling]

From your website...

The TEC130W is ideally suited for CPU cooling due to it's power rating and current requirements. Even with the latest high-performance computer systems and extreme overclocking - the TEC130W is able to take the CPU to zero celsius and below.

From your posts...

The 130W gives good results on an 80W core and acceptable (sub-ambient to ambient) on a 90W core.

The point about the 130 w peltier is certainly a priority issue for me (it was an issue even before I came to the forum.. Why? Because I actually DO research the latest CPUs in order to learn about their O/C capability and their wattage!).

We sell to both AMD and Intel users (although we only use Intel ourselves). We are aware of the heat problems associated with AMD CPUs (amongst many other problems) and we are aware that a TEC130 isn't suitable for ALL CPUs and as time goes by then we will have to switch to higher power units. This is why we're making arrangements to purchase devices upto 280 Watts from a Russian manufacturer.

end quotes...



Instead of spending your time getting free advertising here, you should be updating your site to remove the BS. I don't know peltiers, but I know who NOT to learn about them from. You should be a politician feathers.

[feathers]

Your turn again is it?

(Nice to see you're still making regular visits to the site).

Thanks for the quotes.. Yes.. As u can see.. Some parts of the website do need to be changed. I have said this will be taken care of. I suggest you find something else to do other than bitch at me.

Some of you seem very nice and amiable while others show the more typical 'clique mentality'.

Is there anyone else who wants to step forward and echo the antibling's words?

[AntiBling]

Thanks but no thanks, I'm enjoying bitching at you. Do you have a garage full of underpowered peltiers you need to unload on noobs like me? Now go fix your website!

[feathers]

Special discount for blings.

What's your current CPU temp? I'll bet mine is lower than yours (therefore I must be superior to you and you will be my bitch).

[Les]

Bill.
Your graph is reminiscent of the ancient ""Radiate Calculated"CPU Load v "In Socket" DeltaTcpu - water " compilation.


However, in the light of the work by Incoherent and pHaestus , I tend to think that 100% Radiate(type) calculated Wattages maybe closer than 70%..
Not sure,but think your graph represents an "Insulated Die"(Ri=15 c/w) . Would suggest that the graph for a "Board-mounted Die"(Ri=1 ish?) would have a similar gradient to that for a CPU(using Radiate(type) Wattage).
An illustration of Kryotherm predictions:

[feathers]

Interesting.. Those graphs seem to indicate only marginally better expected performance over the actual performance data from our TEC130s run at 12 volts. Or to put it another way.. Based on the actual performance obtained from our TEC130s.. The above 172W indicates very poor performance.

[satanicoo]

If i have learned one thing about extreme cooling, is that temperature that you see:
-first is not accurate;
-second it ONLY means the thermal diode is being cooled better, not the CPU.

Why did Cathar's wc system beated an vapochill?
I would say because it cools better hotspots then the vapo.

So, i ask pros 1 question: does this happen with peltiers also?
Or is it just another 'vapochill' ?

[feathers]

"-second it ONLY means the thermal diode is being cooled better, not the CPU." - That's right.. However.. My own system places the peltier over the CPU itself and bypasses the Thermal diode. What this means is that although the thermal diode reads a temp of 68 celsius at idle.. The working CPU is actually cooled to -15c by the 40W peltier.

I really can't comment on Vapor cooling but I've heard it's a complete waste of time (if the cold vapor leaks then you will die within 60 seconds because it tends to suck the air from the room).

[satanicoo]

No no no, i mean the on-die thermal diode, the one that is inside the CPU, and that one is never as hot as the part of the cpu that makes the calculations (Hot-Spots)

What i meant by vapo is Vapochill, a phase-change unit weaker than Prometeia, but that everibothy thought it was powerfuller then any water system.

Cathar proved them wrong, i wounder if he can do the same about peltiers?

[feathers]

:-)

I was joking...

Who is this Cathar? Is there a link to his piece on phase change vs water?

Regular water or chilled water?

I can't really imagine ordinary water being more effective than phase-change (but then I've never owned such a system and so my doubts would merely be speculative doubts based on rumours and theory without any absolute grounding in reality).

[Riscoo]

Of course you where...

[feathers]

Ok, I was serious - but not about the 40W peltier.

I see.. I looked at the cascade last night. It's a nice looking waterblock!

Thanks for the link.

[satanicoo]

http://forums.overclockers.com.au/sh...hreadid=202666

Edit:
You have to register, but those are also good forums.
Cathar is the developer of cascade, the best wb ATM.

What about a cascade + chilled water?

[BillA]

Originally Posted by feathers

The grumpy bastard was correct on a few points but wrong on many others.
. . . . . .
It's interesting that UNREGISTERED (aka BillA) has chosen a SWIFTECH chart with which to make his point.

Swiftech went to some considerable expense to develop their air-cooled peltier heatsink (based on 220W pelt) only to discover that it's performance vs noise vs cooling was unacceptable (I know this from personal contact with Swiftech). No doubt BillA would have launched a similar attack on Swiftech for wasting time on such a flawed design?

"your site is a SALES site" - In your opinion. It's a sales site because it has a shop, right?

Do you seek to elevate your status within this community with that particular observation, Bill?

Or are you simply the next in line trying to establish dominance?

I suggest you learn to ease off and don't make too many assumptions.................

I am not going to apologise for owning a site which dares to sell computer hardware.

you are flying blind kiddo, talk with someone, you haven't a clue as to the who, what, and why

[feathers]

Yes, of course Unreg.

Thanks for your kind input.

Would you care to share your own experience of peltier systems with me?

(I am generally interested to hear what people have to say).

[bigben2k]

If it's any help... Bill is THE pioneer of testing. Dismiss him all you want, but he's more than qualified to talk about what he does. That aside, yes, he'll occasionally jump to conclusions, but there's always something wise in his words, if you can put the negativity aside, as all people who run a Forum have done (hence the warning: "Beware, he's grumpy!").

Bill's left his testing days behind, as he was invited to join the Swiftech team last year (and now tests for them). Whatever Swiftech did before Bill joined, is irrelevant here.


The other reason that referred you to the testing Forum (you have to dig up the older threads) is because you can quickly find out that motherboard temperature readings are grossly innacurate. Testing is a LOT more complicated than it looks like.

In this particular case, you could use a heat die (and temp probe) as well as a temp probe within the cold plate, and a controlled cooling solution. You can compare the results to your TEC's expected response curve, and see for yourself the real results: the actual efficiency, then compare it to a water cooled solution.


I haven't even come close to approaching TEC testing, but I started a website for water cooled testing (in my sig).


Let me put everyone's comments into a perspective that would give you a better idea of why ProCooling is around: Would you be able to control the TEC so that it maintains a CPU temp of say, 5 deg C, no matter what the CPU is doing?