Cheap Thermal Adhesive - JB Weld?

I'm looking for some cheap thermal adhesive to attach high-wattage ceramic resistors to an aluminum plate. "Proper" thermal adhesive is really expensive! I've seen some mention of overclockers using JB Weld on their CPUs. Since JB Weld is basically epoxy with metal shavings in it, it is electrically conductive. That won't be a problem though as long as I keep it away from the leads.

Anyone have a better suggestion?

Here is a US ebay source I have used for inexpensive (at least compared to others) thermal adhesive. It has worked well for me on attaching LEDs to heatsinks. Shipping is quick also.

bdrem1 store

The goal to enhancing thermal transfer is to fill the surface voids in both sides while having the thinnest intermediate layer possible.

Without knowing more about the actual mechanical configuration, I would suggest you consider simple (or high performance) thermal paste and mechanically hold the resistors to the metal plate. I think this configuration would result in better thermal conduction because the thermal compound would (likely) be thinner and be optimized for thermal conduction without having to be an adhesive as well.

JohnRob:
The goal to enhancing thermal transfer is to fill the surface voids in both sides while having the thinnest intermediate layer possible.

Without knowing more about the actual mechanical configuration, I would suggest you consider simple (or high performance) thermal paste and mechanically hold the resistors to the metal plate. I think this configuration would result in better thermal conduction because the thermal compound would (likely) be thinner and be optimized for thermal conduction without having to be an adhesive as well.

I agree, we need to know more about your project construction.
Using an adhesive of any sort will only compromise the heat transfer.
Thin is good..and its worth spending the extra.
Tom... :slight_smile:

Arctic makes thermal epoxy that I consider to be cheap....

DrWizard:
I'm looking for some cheap thermal adhesive to attach high-wattage ceramic resistors to an aluminum plate. "Proper" thermal adhesive is really expensive! I've seen some mention of overclockers using JB Weld on their CPUs. Since JB Weld is basically epoxy with metal shavings in it, it is electrically conductive. That won't be a problem though as long as I keep it away from the leads.

Anyone have a better suggestion?

Normally there is thermally conductive grease, and some mechanical means of keeping the device
pressed to the heatsink - this allows replacement if the device fails. Electrical isolation is another
issue, usually solved with thermal pads which provide insulation as well and good thermal conductivity.

However for a resistor the risk of failure is probably a lot less and some sort of adhesive makes sense.
And resistors are usually isolated anyway.

You can just add fine Al or Cu powder to epoxy (as much as you can), giving a reasonable
performance if the joint is cured under some pressure. High temperature epoxy would be
a wise choice. I google, for instance: Thermally Conductive Epoxy Adhesives | Cotronics

I am modifying some of those inexpensive 10 watt RGB floodlights available from all the Chinese sites for around $15-$20. As sold, they run on 120V and have an RF remote to choose colors. I'm eliminating all the original electronics and controlling the LED chip directly from 12 volts (via mosfets), from an Arduino acting as a DMX receiver. I'm not doing a fancy current-limiting circuit, but I do need current limiting resistors. These 10W chips draw a lot of current, so I'm using 2 watt resistors. I had initially planned to use the ceramic kind, but for price and availability reasons ended up with the big wire-wound kind. The whole floodlight is in an aluminum housing with fins. The LED chip is mounted and heatsunk at the middle of the back. The RGB cable comes in the side of the back. There is no longer a circuit board since the incoming R, G, and B wires connect to the 3 big resistors and the 3 big resistors connect to the chip. I need to hold the resistors so they don't flop around, and also it would be good if they could dissipate some of their heat to the housing. At worst, I do not want the adhesive to act as an insulator. The housing is weathertight and thus has no air circulation. (The arduino is not to be mounted in the light housing, it is external, separate, and controls several lights.) This is a budget hack. I got a good deal on the lights and don't wanna blow my savings on expensive adhesive.

(The ceramic ones would have been much easier to glue in place because they are square, but they are a minimum of 5 watts, and many times the price)

DrWizard:
I'm looking for some cheap thermal adhesive to attach high-wattage ceramic resistors to an aluminum plate.

look up "STAR-922" on ebay.

Simple resistor is not enough, active current control is needed.
The LEDs will heat up, drop in effective resistance, and then current flow will increase and overheating will just run way and destroy the LEDs.
Active current control prevents that.

I'm not sure of the ratings you need, start here and do some filtering for a good part:
https://www.digikey.com/products/en/integrated-circuits-ics/pmic-led-drivers/745?k=led+controller&k=&pkeyword=led+controller&pv1989=0&pv87=2&FV=ffe002e9&mnonly=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25

I'm thinking thermal runaway too. On the bright side, it might allow for a snow-free yard in the middle of a blizzard. On the downside, you might have a grass fire.

OK, so this is changing the subject, but in most cases, I fail to see the need for fancy current control for single, relatively low powered LEDs. I can't count how many bazillions of LEDs I have connected to 5, 6, 12 or whatever volts and just used a limiting resistor. Keeping in mind that:

  1. LEDs, especially cheap Chinese ones, often do not operate within the specified range, much less are they uniform.
  2. Using a light meter, I have found almost no difference between running an LED (rated for 20ma max) at 20ma or running it about 12-13ma. After about 12-13ma, the light meter registers almost no increase in brightness. Therefore, I give myself a nice safety margin and calculate my resistor to target around 12-13ma.

Now, these 10watt RGB chips are of course a step above the little 5 and 3mm. According the datasheet, the red should have a voltage drop of 6-7 volts, and the green and blue a drop of 9-10 volts, and all 3 have a max current of 350ma. I have tested 9 of these chips so far, and have found the red has a voltage drop of 6.22-6.35 volts, green 10.1-10.8, and blue 9.5-10.2. So none of them are really in spec. I have targeted my resistor values to drive them at around 260-280ma, and my measurements have confirmed my values to be in the desired range.

I hacked 4 of these 10watt RGB floods as described 3 years ago, and they survived 2 whole seasons of a Christmas light show so far without problems. Now I am hacking 5 more.

And now, all that being said--- Some of these LED current driver chips are much cheaper than I realized. With a cost similar to the high wattage resistors. So now you've got me rethinking my approach if I build more of these in the future.

If I didn't make this clear before, this is a budget build. I scored a deal on the lights (usually $15 ea) for $8 ea and I don't wanna spend a fortune or a lot of time modifying them. As it is, the cable and connectors cost me as much as the lights.

And getting back to the original subject---

Out of curiosity I have conducted a non-scientific test, and JB weld seems to be a very good thermal adhesive!

I took a small chunk of scrap aluminum, placed a 6V incandescent flashlight bulb at one end, and a temperature sensor at the other end. To keep the bulb from directly affecting the temp sensor, I wrapped it with a piece of foam. I then turned on the light and watched the sensor reading go up as the temperature rose.

I then formed a chunk of JB weld about the same size and shape as the scrap aluminum chunk, and repeated the test.

The results were nearly identical. If anything, the JB weld seemed to conduct the heat faster than the aluminum!

The JB weld is semi-conductive, so that needs to be considered with any potential design.

And JB weld is available almost everywhere and considerably cheaper than commercial 'thermal adhesives'.

Hi,
Does JB weld set hard? Like a cement?

Tom... :slight_smile:

TomGeorge:
Hi,
Does JB weld set hard? Like a cement?

Tom... :slight_smile:

Yes, unless you don't add in enough catalyst. Then it's a gooey mess. It's epoxy.

tinman13kup:
Yes, unless you don't add in enough catalyst. Then it's a gooey mess. It's epoxy.

JB Weld, fancy name for one of the 2Part Epoxy concoctions you can get now.

As its only resistors, it would probably be okay, though thermal expansion may eventually cause it to crack.

Usually you make a bracket to clamp the ceramic resistors down.
Or bolt a heatsink over them to clam them down.

Or in hindsight;

Tom... :slight_smile:

Actually, JB weld sets harder than cement, it sets like steel. It's often used by mechanics to fix engine parts including cracked heads and blocks. Once cured it can be cut, milled, drilled, and tapped just like steel (well, close). I've never heard of thermal expansion causing it to crack.

Of course, gluing the resistors in with this stuff means there is no way they can be replaced, they're there for good! It makes a good thermal adhesive with the emphasis definatly on adhesive. If it was a component subject to failure, this would not be a viable option. The housing is big enough though, that if one did fail, I've got extra room to glue another one in.

Hi,

I've never heard of thermal expansion causing it to crack.

Not the JB but the ceramic resistor packaging if you are using these resistors.

The ceramic exterior can be very brittle.

Tom... :slight_smile:

DrWizard:
I'm not doing a fancy current-limiting circuit, but I do need current limiting resistors. These 10W chips draw a lot of current, so I'm using 2 watt resistors. I had initially planned to use the ceramic kind, but for price anavailability reasons ended up with the big wire-wound kindd . The whole floodlight is in an aluminum housing with fins. The LED chip is mounted and heatsunk at the middle of the back.

It would be good to see pictures and a hand drawn circuit.

I would suggest that the current limiting resistors are mounted outside the unit.

I have done similar using 96 IR LED in the past inside a floodlight housing, the limiting resistors heated the housing so much the LED's blew.
Do not know much about the current 10 W designs.
Have you measured the internal temp of the housing in origional configuration ?

DrWizard:
I hacked 4 of these 10watt RGB floods as described 3 years ago, and they survived 2 whole seasons of a Christmas light show so far without problems. Now I am hacking 5 more.

It was the heat that got me.

JB Weld is good for up to 300deg C. Officially you can't buy it in the UK due to EU packaging laws but there are sellers on ebay. I just got a pack to glue the bearing shafts onto the ball screws for my CNC mill