I have just put a XPL hi on a 20mm star and a BLF A6 driver into a Convoy M2 host, All went well with the build and I was pleased with it until it gets to Turbo mode.
It lights up fine first in moon and steps up through the first six modes as it should with the output and tint being as I would expect but when I step it up to turbo or the seventh mode the output drops and the beam tint goes blue!
Overheating? The star was advertised as DTP and I put thermal paste under it.
I have now replaced the LED and star for a xpl hi on a 16mm noctigon PCB that I had laying around and this works fine.
Can someone please tell me what is wrong with the 20mm one?
The difference may be small, but I can’t believe it would be zero. The best thermal transfer will always be metal-metal contact and lapping is intended to maximize that. Thermal paste is best used in small amounts just to “fill in the gaps”, but the best plan will be to minimize/remove those gaps to start with.
When you have no gaps (like when highly polished), the paste will be a layer between the parts, whereas it would indeed fill the gaps when there are actual gaps…
So one can question the actual benefits of two shiny surfaces over not so shiny surfaces.
When it is not shiny, you probably have more chance of actual metal on metal contact.
There are never “no gaps”, take a look at what a “polished” piece of metal looks like under an electron microscope.
The metal never makes contact either, due to the force between molecules.
This is why thermal paste is a must for any electronics that get hot. You never see a heatsink attached to a device just by pushing the metal together.
Our modern CPUs have a big heat spreat plate on it, so the contact surface between the cooler and CPU is relatevily big
But on a XPL the heat transfer area is quite small even with a 1.6mm star
What is more important is the quality of thermal paste Arctic silver gives an advantage over cheap silicone paste
The worst can happen is if the solder area on the LEDs thermal pad is very low
It looks big, but that IHS is just a piece of aluminum. The die underneath it is actually much smaller, and that’s what creates the heat.
The point of the IHS is just to relieve pressure from the die itself. You can get lower temps by delidding and removing the IHS altogether, one less piece of metal between the CPU and the heatsink.
This is true, although bad soldering or a non-DTP board is most likely the reason for it overheating so much. Even bad quality thermal paste shouldn’t make it overheat so much that it goes blue.
In the video they mention that the curve is what is important, not the lapping. If the two surfaces are touching, then there is no use of lapping. But in the example above there was a curve. Lapping would help and make the surfaces flat and touch each other. So no, not zero in all cases.
That’s also not true.
You can damage a CPU or GPU by not using thermal paste.
If you put too much it is not an issue because it just squeezes out the sides.
Read what I wrote. It’s says worse than none not worse than the correct amount. Sheesh. And it only squeezes out if someone takes the trouble to make that happen. Too much paste between the star and the shelf is a big bad nono. PERIOD!
No, you’re wrong.
It is not worse than none.
If you use no thermal paste, a CPU or GPU will overheat.
If you use too much thermal paste, it will not overheat.
In fact, it will perform the same as if you put the correct amount of paste.
And fyi, if your PCB is secured properly with a ring or screws or literally ANYTHING, even if the reflector that holds it in place, then excess thermal paste will be squeezed out.
It is simply not possible to have “too much” because if that was the case then the led MCPCB would literally just fall off the shelf since thermal paste will not hold it.
Something must be holding the PCB in your flashlight.
That video is useless. All he did was demonstrate how inexperienced he was, in what he was doing.
Watching the heatsink chatter across the sandpaper because of his hand position too high and seeing the still pics proving it certainly is Not Flat.
His direction of lapping and Not removing previous cuts as he progressed in grit only shows he accomplished nothing as the results showed.
In the end he is only 50% done. What about the other surface? I could go on about it…
To the OP,
You need to check contacting surfaces. As RBD said I bet the corners of the MCPCB are hanging up in the side and the not getting full contact. I usually use a Sharpie marker to get an idea if things are touching and how well. Paint the back of the MCPCB and set it in the light. Spin it and remove it. Where the sharpie is scratched is where the surfaces are contacting.
After the sharpie I get an idea how bad the surface contact is. Machining is by no means making a smooth surface. More like this . I use a lapping compound between both mating surfaces to make them mate, increasing the surface contact area. I use 800 grit Garnet lapping compound. The Garnet is non-embedding into the soft aluminum and copper. It need to be applied more frequently during the process but I get good results. As the surfaces start to blend together and increase contact, you will see the Grey areas increase.
This was a quick job more to see surface contact. But you can see the Peaks are cut down from the machining process and full contact is made by the swirls. I used thermal epoxy to fasten that copper spacer and clamped in a vise overnight.
What is generally more important than heat transfer from star to body is the heat transfer within the body itself, a hollow pill can be far worse than the worst thermal paste
Often we got 1.6mm copper stars, and then the body of a S3 and others is barely 2mm wide transfering the heat away from the LED
Others got a brass pill that is barely 1mm thick
