OL doesn’t say anything incorrect in the post above but I’d like to clarify one important point. Even achieving a mirror finish on both surfaces is not enough to ensure perfect metal-metal contact. The mirror finish indicates that it is smooth, not flat. We need both smooth and flat. Flat is much more important than smooth. Often flatness is measured by checking the gap under a razor.
As OL already pointed out, properly mating metal surfaces is very difficult. Stick to adding a dollop of thermal paste for now and you’ll be ahead of the game vs none.
Direct solder bond is probably the best? Copper to copper or brass.Trying to get non ferrous metals flat and e lapped micro polished, the Ra is not there, they won’t really “Ring” together like hardened tool steel, the molecular structures not the there. Now you know why a set of Lab quality Jo-Bloc’s (gauge blocks)cost thousands of green back’s $$$$$! But every little bit helps!
I’m getting a vibe here that for the average flashlight user it is best to just leave the flashlight as is and not lose any sleep trying to apply thermal compound to a $10 flashlight. I’ll probably lose the flashlight, drop in toilet, or drive over and crush flashlight first before it overheats and stops working from me not applying thermal paste. For hobbyists and those hell-bent on applying thermal paste though…go for it!
You’re getting the wrong vibe. If the flashlight does not have thermal paste between the star (MCPCB) and the rest of the light it’s likely that the emitter will be damaged after a moderate to small amount of use.
This is the case with the cheap F13 lights. IMO they should all be opened and have thermal paste added, even by the amateur. That beats having the light fail when you were depending on it.
(Some other, much less common, cheap lights use a press-fit on the star. In that scenario the thermal paste issue is dodged. A mechanical / interferance / friction fit allows for heat to be shed at an acceptable rate.)
OK…thanks wight, guess I better get some thermal compound.
Flashlight manufacturers need to kick it up a notch in quality dept. and do the little things like applying needed thermal compound. Retailers like Walmart won’t put shoddy stuff on their shelves if product is junky and customers all come back in store for refund because of poor quality…guess that’s one reason $10 li-ion powered flashlights aren’t found in most popular brick and mortar stores like retailer Walmart.
‘Stepping it up’ the requisite number of ‘notches’ is already done by a variety of manufacturers. Look at Fenix for example. Fenix is carried by retailers in the US, Fenix covers all the bases… and Fenix’s price matches that. Different markets.
If there is adhesive under the star that is most certainly thermal paste. (I’ve not seen anything else used)
I don’t know what percent better it will be, but cheap paste was used on CPU’s and is still used on cheap video cards and such. So it will extend the life of your emitter. Alot depends on the pill itself even with a copper star direct soldered if its on a POS pill then the copper can only do so much.
But as to if its worth it, YES IMO if yo have am emitter with nothing under it then it well worth effort.
There are lots of variables there. What is the condition of the original? Is it done with a thin layer of paste originally, or is it a thick glob? Is the paste actually under the star, or just under part of it. Is the light something that is only putting our low amperage, so there is little heat anyway? Is it worth the possibility of damaging things if the adhesive does not want to come off and you have to pry on the star, bending it, or damaging the pill? If in doubt of how good the thermal layer is originally, then it’s easiest, for peace of mind, to take off the old and put your own on. If you think it’s already good, then don’t mess with it.
Thermal conductivity is the property of a material to conduct heat. Thermal conductivity can be defined as
"the quantity of heat transmitted through a unit thickness of a material - in a direction normal to a surface of unit area - due to a unit temperature gradient under steady state conditions"
Thermal conductivity units are W/(m K) in the SI system and Btu/(hr ft °F) in the Imperial system.
W/(m K) 1 W/(m K) = 1 W/(m oC) = 0.85984 kcal/(h m oC) = 0.5779 Btu/(ft h oF) = 0.048 Btu/(in h oF)
If thermal paste is already there I would leave it as is.
If no thermal paste/grease, then try to lift the emitter star upwards. There is usually enough wire from driver to star that you don’t have to de-solder the wires. Then simply apply grease underneath and press star back into place.
Okay, this brings a question to my mind. Using a Sipik SK68 as an example, if nothing were done to the light at all, would using a 14500 reduce the life of the LED and by default a regular AA maximize the life of the LED? I’ve been thinking about this with the Ti3 I just purchased. I received an Efest 10440 the same day the Ti3 arrived. I also have Eneloop AAA’s that just arrived. Now of course the Efest 10440 makes the Ti3 a lot brighter. But would the light last me longer, all things being equal if I used the Eneloop AAA’s rather than the 10440’s?
And again, I appreciate everyone’s input. It is answering questions and bringing other questions up. Quite an education.
In complete generalities with electronics heat shortens there life. The problem is there isn’t a direct correlation to X heat will shorten by this amount of light.
So say you have two 50,000 hour LEDs A and B. LED A is ran 2x hotter than LED B so in theory LED A would fail in half the time, well nothing I have ever seen proves that theory. It is almost certain LED A would fail first but who knows exactly when.
I would be amazed if a SK68 on a 14500 didn’t fail before the same SK68 being ran on AA. IMO it is worth it, it’s a cheap light with much better output. Now if I was in a SHTF situation and I had to pick between 14500 and AA for the SK68 then trying to extend the life I would use AA, but in real life I’m going to burn that baby up and just buy another.
It depends on if you’ve got a press-fit star in there. If the star has been shoved into place such that it is a very tight fit around the edges, this should be sufficient IMO. (press fit aka an ‘interference’ fit)
Also in your guestimations we might want to consider that the 14500 is probably easier on the driver than a NiMH or Lithium AA, which tax the boost circuit. The boost circuit was intended for that purpose, but it’s also bottom of the barrel.
I’d wager that more SK68 drivers fail than emitters?
As a general hint I would say that it is important to check the situation on every budget light at first before even putting cells in it. I had to learn this the hard way and it costs me a XML-Emitter. It was a cheap Ultrafire SH-98 from Fasttech (18650 SK-68 clone). After some short times on high the emitter was flickering, very dim and bluish. Only the half of the surface gives light, it had to be replaced. After inspection of the cause I found that the disc which should hold the emitter down was not tightend at all and to make things worse there were spills of metal under the star (residues of the wire holes).
This video is indeed awkward…
—-
Vaseline can harm rubber.
This is an interesting source to check compatibility of some materials to chemicals:
I checked petroleum and rubber as long as petroleom and aluminium. Seems not suitable so I never use it.
—-
I am also one of the too much paste guys, I often apply paste put star on pill twist a bit and after that I remove some of the paste again with a tissue messing around the hostsknurling and fins…
For screws I am often too lazy especially if the reflector is a tight fit I don’t think that it improves the result in a seeable way, it’s more the:“I have done it the best possible way” thinking that sometimes let me do this.
I also have filled up more than once a center gap in a pill just with a bit of thermal paste, but especially on small lights they get so hot so fast that I think it doesn’t matters in a real world way.
Not to argue over brands or anything but for longevity, you want a non-solvent type compound or just use something like Fujik. Pull off an old CPU and you might find the paste has dried out and conducts heat worse than no paste at all. Depending on use, solvent types should be replaced every few years. This what I’ve been using: http://www.microsi.com/user/document/DS%20X23-7921-5.pdf
If anyone knows of another reputable brand that makes a non-solvent type, please share.
If you’re going to spread it, put a little more in the middle and gradually spread it thinner outward like a pyramid. If you don’t, air in areas where it’s thinly spread gets trapped and as you push down on the star, the compound is forced from between the surfaces. You want the opposite, compound surrounded by air, not air surrounded by compound. I just put a little drop in the middle and slowly wiggle the star down so all the air is displaced by compound, then compress it. After that, don’t pull it off and put it back on without reapplying more compound to the center. Just a little tit is all you need.
