Best Thermal Paste According to Independent Studies?

No, it's not that at all. Those strawmen sure are easier to knock down though, aren't they?

Show me a test that shows any change at all in light output between a copper MCPCB with really bad thermal paste and a copper MCPCB soldered to the pill. Show me. I would think you would want to know if these things are real too, why am I the silly one by asking for the proof that these things actually work?

I assume the laws of thermodynamics apply so increased thermal transfer rate means the heat gets from one surface to another more easily when the interfaces are more thermally conductive. How much does that affect flashlights is a legitimate question, your saying not much, if someone wants to test this and provide some data then no one will object.

As the links i posted above demonstrate don’t use cheese in place of thermal compound

Share a Coke with Bort!

In a synthetic test where you have a tiny 1cm x 1cm patch and pumping 200 watts through it, those thermal paste charts are probably perfectly true. But change that to 30 watts through a 16mm diameter MCPCB and you can't measure any difference between the worst and the best. The difference is less than the margin of error of the measuring equipment.

Now, how much time, effort, & money (this is still 'budget light forum', right?) are you willing to invest for zero gain?

This does not mean the bottom of the MCPCB & top of the pill do not need to be flat. This does not mean a hollow pill is just as good as a solid pill. This does not mean a non-direct aluminum MCPCB with a dielectric layer in the middle is as good as a direct-thermal copper MCPCB. Do the proper prep work, that's far from 'none of this stuff matters'. The stuff that makes a difference matters. The stuff that does not make more light come out does not matter.

Bort is not one to change his avatar but this photo was impossible to resist :slight_smile:

your probably right, as an unknown person said “In God we trust; all others must bring data”
and:

I've been using GC Extreme, but comfy is probably right. I'm interested in sustained reduction of temperature at the LED, and I have higher confidence in GC Extreme right now for longevity (both long runtimes and life cycle), based on various sources/reviews from real users online - of course it's all from usage on CPU heatsinks, but they measured apparent real differences in temps. Does that apply to the lower power levels of LED's? Maybe, probably not the same though.

Problem is I'm not aware of anyone that has done comparative tests on our modern LED's and MCPCB's. Even so, again, I'd be interested in how they perform over time, and also what the temps are at the critical spots. Keeping the LED cooler at high amps will probably extend the lifetime of the LED (not only brightness is at stake), but again, anything over 3A to the LED is beyond manufacturer's limit.

Best thermal paste for what purpouse? CPU cooling, Led cooling?

You see, my PC is watercooled, back in 6 months i changed the CPU to 4770K…and i didnt got the cooling results i wanted, then i have decided to change the thermal interface between the CPU core and the protective cap -voila, temperatures dropped with 10 degrees!
My great surprice came after 2 weeks, the CPU temp not only rised, but rised alot! Up to 95 deg C under prime 95( while watercooled!), and the CPU became unstable…i used Prolimetech PK 2 by that time…then i swapped it with Pk3… with Gelid….same results after 1-2 weeks of heavy usage
Then i found a topic about that matter in Extreme systems forum, it seems that under heavy load and constant high temps most CPU thermal compound are degrading pretty fast especialy when put on a naked CPu core, exept the pure sillicon based thermal paste and of course liquid metal, now my CPU`s thermal interface is Coolaboratory Pro liquid metal, and after already 5 months i can say its rock solid
The GPu core on the other hand do not degrade thermal pastes that fast! Its bigger, and usualy works cooler, under heavly load and OCed temperatures never pass beyond 55 deg C
Since LEDS are always under high temperatures, maybe results there are simmilar- liquid metal cant be used on Al headsinks , so i dont have that huge experience exept with 2x30w Bridgelux emitters mounted on huge CM copper headsinks

P.S
When i removed the CPU cap the default Intel thermal interface was a pure sillicone glue of some sort
Have pics, if someone is interested

Btw, i have never tested AS5….its 99% micronised Silver, maybe it wont degrade , but it can oxidise

I used to be on extreme and evga forums :slight_smile: Yeah been using that for years Mitko… *Just note it doesn’t work on aluminum which is like all the pills

Yeah i mentioned that, but on copper it works! Plus, AS5 issue that i mentioned earlier…have to test it though

My point here is that PK1-2-3 or even Gelid wont do any good here, vise-vesra, they are worst that a cheap pure sillicon paste, the emitter in our case is always working at its max( most of the times at least), temps are above 85 deg, those CPU pastes arent designed for that

Yet…AS5….unfortunately my CPU is permanently metal- glued, and the LED die dont have build in thermal sensors

Have to ask someone from Xtreme if he tested with AS5, they do experiment alot

Ive always had great luck with arctic silver 5

There is very little difference among all of them. Even the best thermal paste will only give you a 3-4 Celcius better than the cheap one, which doesn’t make meaningful difference IMHO.
It’s more about the peace of mind, knowing you have used the best.
I’m not suggesting that you go with the cheapest one, but I’m also not seeing the point why anyone would spend $12 for a 4-gram syringe.
I’ve tried possibly all the best thermal paste in the market for comparison, from Shinetsu, Gelid, Prolimatech, Zalman, and my conclusion is that they are not worth the money you spend.
Even the infamous Shinetsu 7783D ($8-9 3 gram) only lowers my CPU 3C better than the cheap Servisol thermal paste ($2 25 gram tube) at full CPU load.
I bought Servisol 5 years ago, and I still have half tube even now. It performs the same like 5 years ago.
If you want to save your money, go aliexpress & search for Halnziye HY-810 or GD900, which I suspect to be the same thing.
It’s so much cheaper than those “fancy” pastes from Shinetsu, Tunix, Gelid, etc for a much more reasonable price.
You can get GD900 for $7.70 for 3 syringes @15gram, which you can use for a lifetime. They are non-bleed pastes. Viscocity is medium.

I have used Halnziye HY-810 & GD900 in the last 2 years. I bought HY-810 for only $10 (20 syringes 1gram) , and $ 4.5 for GD900 (30 gram tube). They give the same performance as the stupidly expensive Shinetsu, Prolimatech PK3, you name it (but I actually use Servisol for my flashlight). Unfortunately HY-810 is a little more difficult to find in retail package now, but you can still find it in Ebay for $1 a tube 1 gram.
Or simply buy GD900 in aliexpress, which you can buy much cheaper in a tube of 30gr ($5).
Both HY-810 & GD900 performs exactly the same in my overclocked i7, and I actually suspect they are the same thermal paste produced by Halnziye, only re-packaged.
Even Tunix has considered selling HY-810 under its own name as a successor of TX-4, and has asked puroverclock.com to evaluate their investment.

Here is the result:
http://www.pureoverclock.com/Review-detail/hy810-and-hy880-compound-review/

Excuse me, but i wont agree with you: reasons

- Firstly, the temp differences arent 2-3 deg but they can reash >10

- Every deg lower DO MATTER when we speak of extreme OC

- When the LED or the CPU are well OCed that thermal paste difference is even bigger

  • The main problem::: most of them DO DEGRADE pretty fast in time, and by ” in time” i mean 1-2 weeks when we are talking about extreme OS, believe me, i have a huge headaches about that matter before we cleared that issue out in Xtreme sys forum

Well, a causal user wont bennefit from an ultimate thermal interface, its true

Fujik sold by DX is the worst I’ve ever used, usually use Arctic Allumina.

In VERY high temperature and small footprint die’s the thermal coefficient/conductivity really matters, especially when it’s attached to an air cooled (or water cooled) radiation device, the faster it can move the heat from a small surface area to the larger surface area heatsink yes…those very detailed charts do come into play

With lower power direct soldered emitters and large footprint copper stars, the temperature coefficient might not be that important, due to the fact that thermal saturation occurs because the heat really doesn’t have anywhere to go once it bleeds off from the emitter to the star, to the pill, to the body of the flashlight

Cost effectiveness vs thermal effectiveness comes into play in this area

Computers vs flashlights = apples vs oranges

I use silicon and epoxy (and sometimes super cheap artic silver on the threads of the pill)…and the thermal management seems to do just fine, dropping $20-$40 on a tiny tube to put on a <$20 flashlight…well you see the conundrum?

Using good thermal compound is always good…but in flashlights will it really make that much of a difference?

just to add some info

I used Gelid GC extreme on GPUs, it has excellent performance in beginning, but after few months, it degraded to have absolutely terrible performance, worse than using no paste at all, I think it actually became a pretty good thermal isolator

The thing with Coolabs is it doesn’t degrade. That’s the key. Everything I tried before that- well maybe not AS5 that always lasted but it had a awful long cure time. But other pastes fail. They work for a time but dry up and sure enough over a couple months you would see the temps basically skyrocket and everything would fall out. In flashlights it may not be critical, but no thermal transfer over time is something I’d rather not deal with, it can only come back to blow up in your face :slight_smile:

What you see now is not what you get down the road… If the emitter is screwed down it may not suffer too much, and if the light is cheap then I use the bucket tim… But I’d definitely use it in lights you are going to push if possible (I know the aluminum thing is something we never had to deal with- always copper or nickel)

Personally I’ve never seen any thermal paste (even the cheapest one that I got free from a $2 heatsink) that causes >10C difference, unless when compared to tooth paste (as shown in Tom’s Hardware graph above ). The only exception might be Coollaboratory liquid metal, which also has pros & cons (I wouldn’t explain this here).
It’s true some super cheap paste degrade pretty fast, perhaps several months. If it degrades within 1-2 weeks, then I think it must be the lowest of the lowest quality (which is why I said I don’t suggest to buy the cheapest paste, but at least a decent one).

Even in extreme OC, when a couple of degree makes the difference betwen stability vs instability, then I believe it has been pushed too far beyond the limit because it can crash anytime. The room temperature itself can go up & down several degrees throughout the day. I wouldn’t call it a good extreme OC. I also believe when we are talking “extreme OC”, then it is already beyond normal thermal paste (watercooling, liquid metal pro, etc).

I agree with what Warhawk said, that is why I only use the cheap Servisol for my flashlight, which has been proven to perform almost as good as the expensive paste. This cheap Servisol also has the right bleed consistency/density to be applied to the flashlight, better than my other expensive thermal paste.
I’ve been using this Servisol for more than 12 years. I used it mostly for friend’s computers, and it doesn’t degrade for years. I have many friends still using the same paste which I applied 5 years back, and their CPU temperature is still relatively normal (apart from dust cleaning on the heatsink itself). And it is only $2 for 25 gram tube.
I have applied HY-810 & GD900 to my overclocked i7 desktop @4.8GHz & all my family’s notebooks since 2 years ago, and all seems to be working very well till today.

This! Read it carefully

I think you havent read my whole post, pls did it

Is Prolimatech Pk1-2-3 lowest quality? 15 euros per 5 grams? OR maybe Gelid? 21 euros per 5 grams!
Its not a matter of months dude, its a matter of weeks, especialy on a naked CPU die! Believe me, i passed through this, and i am just sharing my experience, for the others NOT TO SUFFER the way i did, and wonder wft is going on
There is a Xtreme forum topic about this matter

Arctic Silver Arctic Ceramique 2 on the other hand is extremely cheap, and lasts for years, as user “”fellfromtree”” explaind it well, it doenst dry that fast…maybe ina year or too yes, havent tested it much
I use this one widely on my LEDs

I ordered AS5, and will torture my wifes PC :bigsmile:
WIll post results eventualy, on theory it shoudlnt degrade that much, but again, the silver can oxidise, who knows, have to see it

P.S
On my PC setup, the difference between Gelid( tested it second) and Arctic Silver Arctic Ceramique 2 on the WC setup and under 5 gigls load was exacly 10 degrees, compared with CL liquid metal its 15-17 degrees
All this with the Intel default thermal glue removed

I assume that's the same stuff I have, they've changed the packaging since mine was bought ~10 years ago. The tube mine came in split open and died long ago, it's since been scooped out and now lives in a repurposed Tylenol bottle. Still works like the day I bought it (whenever that was). There is no paste that will work any better in any conceivable flashlight application, the heat loading just isn't high enough. It also doesn't degrade over time either in use or in storage.

I use this same stuff on my 4.4GHz i5 3570K, been assembled for over a year and still runs the same full-load temps as when new.

Interesting… Well you would know on the flashlight end for sure Comfy- I’ll bookmark that for later