Ot: Driverless cob led lights

[Willy]

Some interesting insights from Arctic Silver's website and subsequent application instructions further to what Dan and Jerry mentioned.

http://www.arcticsilver.com/as5.htm

A partial description on how the product is designed to function.

Controlled Triple-Phase Viscosity:
Arctic Silver 5 does not contain any silicone. The suspension fluid is a proprietary mixture of advanced polysynthetic oils that work together to provide three distinctive functional phases. As it comes from the syringe, Arctic Silver 5's consistency is engineered for easy application. During the CPU's initial use, the compound thins out to enhance the filling of the microscopic valleys and ensure the best physical contact between the heatsink and the CPU core. Then the compound thickens slightly over the next 50 to 200 hours of use to its final consistency designed for long-term stability.

Plus a small quote from one of their application guidelines.

Stock processors and/or heatsinks with normal surface
irregularities will require a layer 0.003" to 0.005 thick as shown below to fill the
resultant gaps. (Equal to the thickness of about 1 sheet of standard weight paper.)
Properly lapped heatsinks with mirror finishes will only require a translucent
haze.

I can achieve a pretty fair finish when milling aluminum but it won't be mirror finish so I''l have to gob on that .003-.005 layer and see how tough that cheap cob led really is.

Thanks everyone.

[J Tiers]

... Keep in mind that the thicker you put it on, the less effective it is.

And that is the issue with virtually every "paste". The appropriate amount of any paste is the thinnest possible coating that will fill all the gaps. And that depends on how smooth and flat the surfaces are, something you may not be sure of.

If the surfaces are very smooth and flat, you need only a barely visible trace. But it is difficult to put on such an amount, even with less viscous thermal compounds, and difficult to determine just what is the right amount. Many of the compounds are thick greases that do not flow well, and cannot be depended on to flow out if applied too thick.

THAT is the reason I recommend the "phase change" materials. They "self-apply", in just the right amount. You can actually bolt down the item, and put a glob of the phase change material touching one side. When the part gets over about 60C, the phase change material melts and wicks under the part, filling the gaps but not holding the part up off the surface at all.

The better way is to apply it to the underside, and let it melt and flow there. Any semiconductor should have some springiness to the clamping, so when the thin coat melts, the part is pressed down as the material spreads out. It's difficult to put on he wrong amount.

The high power IGBTs usually use a thin sheet of it, often on both sides of a piece of foil, for handling ease.

[danlb]

Good info. Perhaps the use of the term thermal paste is the wrong wording on my part.

I was of the understanding that this is exactly what the role of a thermally conductive paste/compound was, although I do realize that there are good products and then there are the rest.

I have seen some that bleed and run when subjected to normal component operating temps while others have dried out to the point they are no longer able to function as intended.

In your opinion how does the Arctic Silver 5 that I'm now using compare to those you linked to?

Arctic Silver is considered quite good in the high power flashlight groups. Keep in mind that the thicker you put it on, the less effective it is.

[lakeside53]

I have used the arctic silver on 10 watt emitters. No issues at all.

[J Tiers]

I have no idea, frankly. I have never used the "arctic silver".

The heatsinks in the computers here had had the paste dry out, so I cleaned them, and used the stuff I suggested. I do not know the temp difference for some standard condition, but the fans seem to run less. The material is in a very thin layer, and the springs in the heatsink clamping system can press it out to a very thin coating when it has changed phase to a liquid. Thinner than any grease I am aware of, and wicks into the smallest spaces.

The "silver" pastes are filled greases, with the filler intended to be a very good conductor. But the silver particles have thickness. It's a question of whether the highly conductive filler can make up with conductivity for not being everywhere, and inherently being a "spacer" between the surfaces.

The folks who use big power IGBTs use the phase change material. That was recommendation enough for me. After we swapped over to using the phase change on a particular product, the failures of power switching devices dropped off vs what had been the case with greases. In fact, on that product, I know of only one case of power device failure after the swap, put of several thousand units. (they had aluminum oxide "ceramic" insulators).

Used with 2 mil mylar, we had zero thermal failures in several hundred units.

[Willy]

If you really need low thermal resistance, which you may not actually in this case, "thermal paste" is the WRONG material to use.

Actually, IMO, thermal paste is USUALLY the wrong stuff to use.

The best material is a "phase change" type material, which will wick in and fill tiny spaces once the part gets hot. The thermal resistance of the joint can go very low, as low as 0.02 C in^2/w.

https://www.digikey.com/products/en/...=0&pageSize=25

https://www.aavid.com/sites/default/...ultrastick.pdf

Good info. Perhaps the use of the term thermal paste is the wrong wording on my part.

I was of the understanding that this is exactly what the role of a thermally conductive paste/compound was, although I do realize that there are good products and then there are the rest.

I have seen some that bleed and run when subjected to normal component operating temps while others have dried out to the point they are no longer able to function as intended.

In your opinion how does the Arctic Silver 5 that I'm now using compare to those you linked to?

[J Tiers]

If you really need low thermal resistance, which you may not actually in this case, "thermal paste" is the WRONG material to use.

Actually, IMO, thermal paste is USUALLY the wrong stuff to use.

The best material is a "phase change" type material, which will wick in and fill tiny spaces once the part gets hot. The thermal resistance of the joint can go very low, as low as 0.02 C in^2/w.

https://www.digikey.com/products/en/...=0&pageSize=25

https://www.aavid.com/sites/default/...ultrastick.pdf

[Willy]

Check out this website to see if you have similar package

http://www.cob-led.com/COB-LED-array-standard-CRI.html

Most of these devices have 3*C per watt thermal resistance, and have to be de-rated above 80*C

Max voltage 60 @ 800mA

The ones I'm dealing with are like those shown in my first post and linked to. From what I've been able to gather most do seem to need derating beyond 80°C. Who knows if these ultra cheap units don't get close to the advertised 50 watts it may not be an issue.

Willy wrote: I fully intend to use thermal paste as I do have lots on hand. I'm thinking of machining the housing flat where I intend to mount the led as it will certainly aid in the heat transfer,

You probably already know this, but I'll say it anyway. After machining you need to lap the housing to get it as flat as possible. Then use as little thermal paste as possible, just like with a CPU heat sink. That's the way to do it if you need to maximize the heat transfer.

Dan

Thanks Dan, yes well aware of the proper technique and for a more worthy component this is what I'd do also. However for as little as these things cost I'm not going to get too anal on their install. A milled seat and a good quality thermal paste is all they'll see.

In practice I found the COBs to be thin and flexible, not rigid like a CPU, so just piled the paste on.

These have an aluminum backing plate. Not sure how thick or rigid but I suspect flexible is probably going to be an apt choice of terms. We'll see.
I will monitor temperatures though as well as current draw and a subjective appraisal of light quality/quantity. I have light meters but I don't need to turn the house upside down in order to find them.

[Davidhcnc]

Willy wrote: I fully intend to use thermal paste as I do have lots on hand. I'm thinking of machining the housing flat where I intend to mount the led as it will certainly aid in the heat transfer,

You probably already know this, but I'll say it anyway. After machining you need to lap the housing to get it as flat as possible. Then use as little thermal paste as possible, just like with a CPU heat sink. That's the way to do it if you need to maximize the heat transfer.

Dan

In practice I found the COBs to be thin and flexible, not rigid like a CPU, so just piled the paste on.

[danlb]

Willy wrote: I fully intend to use thermal paste as I do have lots on hand. I'm thinking of machining the housing flat where I intend to mount the led as it will certainly aid in the heat transfer,

You probably already know this, but I'll say it anyway. After machining you need to lap the housing to get it as flat as possible. Then use as little thermal paste as possible, just like with a CPU heat sink. That's the way to do it if you need to maximize the heat transfer.

Dan

[bob_s]

Check out this website to see if you have similar package

http://www.cob-led.com/COB-LED-array-standard-CRI.html

Most of these devices have 3*C per watt thermal resistance, and have to be de-rated above 80*C

Max voltage 60 @ 800mA

[Willy]

I appreciate the additional input and the need to get rid of the heat. I fully intend to use thermal paste as I do have lots on hand. I'm thinking of machining the housing flat where I intend to mount the led as it will certainly aid in the heat transfer, I also have a number of cpu heat sinks so if necessary I'll try to utilize them as well. I'm hoping to keep the cooling process passive as adding fans will only complicate the installation and add noise as Darryl mentioned.

I should probably have clarified the term "cob" when referring to these as cob leds. It does not mean corn cob style lights but instead cob is simply an acronym for "chip on board". The technology on these types of products is progressing so fast these days that we don't always have time to catch up on all of it as it transpires. Only a few short years ago none of these terms would have been recognized by any of us.

I do have several led flood lights already as mentioned with an integrated power source mounted externally and I've been very impress with them, even the 10 watt units are very impressive given the power required to drive them.
Here's hoping that the cheapest chip on board type of light that I could find will be only half as good as it'e 50 watt claim says it it is. This is kind of an experiment to see how much light I can get out of a buck and a half and a bit of experimentation. Nothing ventured nothing gained.

[Davidhcnc]

Yes, I have recently used these for the job you describe, Mine did draw the 50watt so are true to claim. If you buy a new light that is what is in it.

Longevity, I don't know yet but they work. They need flat metal contact and silver paste for the heat sink.

This seller https://www.ebay.co.uk/sch/baoshengh...p2047675.l2562

[MrFluffy]

I use those led chips for machine worklights, but the versions with the seperate ac->dc transformer. 20w gives a good light with low voltage outside of the transformer. I've migraged all of my ex lovo machine lamps to them as 80v bulbs are getting hard to find and as I get older, I like a lot of light. In fact the first one I tried, I used a 100w led unit, and it was too much, blinding almost. So I swapped the led itself for a 20w unit.
But, you do have to add heatsinks, and heatsink to the metal lamp casing also as they do run hot. I tuck the transformer away safe in the machine base with the other electrics. For heatsinks I have access to some dead computers and I go swipe a cpu heatsink to mount them onto directly with some compound.
I have 4 seperate 10w chip style lights pointed at my motorcycle work bench to give additional lighting to improve the work area too, theyre great and not too bright so theyre probably fake 5w units :-) but they were just bought as 10w leds.
When you said led corn cob type bulbs, I thought you meant the led corncob bulbs that fit the ex halogen floodlight fittings. They are good too, lots of light for significantly less power budget and a lot less hassle to fit, but theyre not quite as good as converting to the chip style ones and probably a shorter life as they dont have great heat transmission to the case unlike mounting a flat one right on the casting.

[grzdomagala]

I don't have experience with this particular model - i tried other cheap driverless led module. The damn thing blinked with mains frequency like a stroboscope - barely usable as reading lamp, absolutely unusable around rotating machinery.
It was a few years ago so your cob may be better suitted (some sort of filtering, longer decay time on phosphorus?).

Wysłane z mojego GT-N7100 przy użyciu Tapatalka