FLIR Test Data for possible new Thermal Solutions

Giveaways and Contests
1

This was a thread I was somewhat hesitant to make at first—concerned about interest it would receive, but maybe some users will see this like I do; FLIR data tells a lot. So here goes.

A few MFGRs have claimed in the past that they have used some of “the best” or “special” potting compounds which help dissipate heat out into the light host/body. For example, I owned one of the 72Kcd DEFT-EDC (a special edition), and the driver was completely potted with an off-white compound that supposedly had great thermal ability. The compound? I do not know. I do know a tiny XR-C emitter at 1.5A (what’s in the ‘special edition’ DEFT EDC) doesn’t have much thermal loading on the driver to make much heat from 1S, but this still got me thinking about potting methods and the like.

The FLIR testing:

A cousin of mine recently got me in contact with a man who has new FLIR equipment. He is about 1.5hrs drive away from me, I had some conversations with him about flashlight testing, and I would be willing to drive to him to help evaluate new cooling solutions we could possibly use in current and future light designs. These would be setups primarily for use as DIY, (possibly) cheap, effective driver cooling in demanding scenarios.

My main question to others came to me like this: What are some of the best thermal potting compounds we know of that exist out there?

Then the thought I am curious to find the answer to is, how effective will they be when used to aid driver cooling in any given practical light design?

From my experience there seems to be preference to use very good thermal potting compounds to aid consistency of output. So this is going to revolve around thermal potting, or complete (or at least top-half) submersion of a driver in potting compound, which is sealed into the host itself. (It would seem a compound with a heavy metals content could work for this, without becoming conductive, since the metals are blended in a binder that is non-conductive.)

I’ve long been on a hunt to find a potting compound which isn’t just a potting compound, but something maybe a little “extra special”, that can really help some driver heat issues we might be going up against soon.

Do we know of any potting compounds that would do this much more effectively than standard potting methods?

With his help with IR devices I can analyze just how good a method works using a FLIR (infra-red imaging) system, and I think it’s a great opportunity to put the call out to the forum to potentially find something new and unique for driver cooling solutions.

Here is what I thought I could offer. If you feel you have a solution, and others are interested in seeing this solution work, I will offer to drive to his FLIR imaging lab and test the method out, if we can come to a conclusion on proper test setup, of course also using a control light in the experiment. I would take the experiment to him, run through the paces, and bring back the full imaging and numerical data to document the effectiveness of the method. All would be allowed to see how effective the method is, I would want to make the results public on BLF. If you do not want your suggested method to become public info, but are interested in setting up a test with FLIR imaging, PM me, and I will try to work something out with you to get the test set up and data to you. I would hope that is not the case, though. I really would like to bring this data to the forum to promote effective new DIY cooling solutions. Better cooling is likely the best thing we can have in our lights, and a FLIR picture comparison says a million words. :wink:

I’m open to hearing all and any opinions surrounding this idea and testing procedures. It doesn’t have to be potting-limited testing. If any ideas have possible merit to make it worth the drive to acquire new cooling-effective data, I will make the drive and obtain as much info as I can, so long as it seems logical enough and interest is there to make the experiment happen.

Thanks for reading.

-MEM

2

Randy (PFlexPro) does alot of thermal testing…. pflexpro.com is coming soon Redirect Notice

3

Yeah, I have to second this… He is a tester through and through… I would get a hold of him. Randy is very thorough….

4

I’ve heard that fine graphite mixed with the brush on style of electrical tape is pretty effective and is nearly completely non-conductive electrically. I’ve got the ingredients here, was able to get everything at the hardware store.

5

I’ve been under the impression the JB Weld was pretty good. Others talked about mixing in some SiCarbide powder to increase the thermal conductivity while maintaining insulating properties. I don’t know where the JB stands in terms of capacitance, but I know a lot of modders here use it without many problems…

6

Yes, I too have an IR thermometer. Point the laser, pull the trigger. True accuracy? You have no idea what you are actually reading. The dot from the laser is just an area marker, not the reading point. The reading point is the reflection of standing IR waves between source and sensor. Point? An IR thermometer is still a crude means of testing. If you want to read the temp of a block or cube 2x2x2”, an IR thermometer would probably be pretty accurate up close, but you wouldn’t see an image of where the most heat is located.

Color IR pictures are generated with a FLIR. The system is a few thousand dollars vs $40 for an IR temp gun. Much more data is returned for that cost difference. The colors can be set to the temp range being worked at, so one can identify the temperature more quickly.

We have some suggestions in the mix here also. A couple of pills could be made/used, potted precisely with similar amounts of different materials, then each pill screwed into a bare UF-1504, for example. I have a bare 1504 host, and when I say bare I mean coating removed, as I sandblast them before Cerakote. This will probably effect very little having the anodizing removed, but will at least remove one very tiny variable. So a perfect example here already would be, how well does the graphite mix work against the more expensive 3M mix?

A color IR picture basically shows exactly what you want to see; where the heat is accumulating.

He appears to have good potting methods in the link posted above, but I just want you to know the method I’m talking about is not his method of testing or even close to it.

Do we have photos at this forum of lights that have been FLIR analyzed before/after a treatment?

For a FLIR image example:

7

I think I’ve seen a few thermal images here, mostly HKJ charger tests though.

8

I generally prefer non-permanent potting so I have the option to modify things. The combination of duct seal and silicon carbide suggested in the potting compound thread has worked well for me. The duct seal is sticky enough to hold a lot of grit before falling apart and with a mix of 600/220/120 grits I think it packs well in the matrix. I would be worried about using anything with graphite fillers because if you add enough it can become conductive. And JB weld has a steel filler so not great thermal conductivity wise. My money is on silicon carbide, non-conductive, cheap, easy to get and of course with a great thermal conductivity (some crystal alignments are higher than copper, others are about equal). I’ve also used it mixed in epoxy to make thermal glue which could be used for potting, trick there is to use thin casting grade epoxy so you can add a ton of filler before it gets too thick. That’s my vote of what I want tested.

I almost think a less real world test would make the results more obvious. Like if you stuck equal size blocks of all the test materials to an aluminum plate and then evenly heated it from behind, ideally with a resistance heater mounted to the back. Then pointing the camera at it we would see a nice visual side by side comparison. Also of course aim the camera down with the plate parallel to the ground to avoid convection errors.

9

Yes, excellent thinking. Indeed, I would be doing a host of tests with the drive being so far. He works at Argonne Lab in IL, I guess he can do some pretty high-end tests from what he was talking about. There is a laser lab he said with optical characteristic generation. It reads a laser after it passes the “lab-rat” lens mounted in place and provides feedback data on various wavelengths. It’s metaphorically the FLIR of lens quality. That really spiked my interest, because THOR has those analyzer types and they approach $20,000+. But we have a “B” in our name, better not get too scientific and accidentally learn something. :bigsmile:

10

IR temperature measurement is not really useful on small parts, you need a thermal imager.

But as some say, it is rather expensive. The price is going down, but there are a lot of patents and the companies behind the patents like the high prices.

11

I take it your looking for the effect the lens transmission has on final tint then? Or are you just checking to verify coatings? I could see using it to compare lenses with the “same” coating from different suppliers to compare their actual quality.

While it’s interesting data, I think that simply testing a lens for output and image quality with the LED you plan to use provides more useful data when it comes to a humble flashlight.

12

Ah…well, no. I mean I see why you would think that as I’ve talked a storm up about it lately. But from what I understand there’s many different approaches to these. In one setup type, a laser grate (like an XBOX Kinnect uses to place laser dots all over the room and “see” in 3D) is used. Like a bundle of beams; which are red and blue for instance. Essentially the beams are used to see what the lens is doing everywhere. I’m sure camera manufacturers test lenses like this vigorously in the lab. If you go to a lens manufacturer and browse their aspheric section, you might find 20 lenses of different sizes, possibly believing that what varies is lens diameter and focal length ratio. But the variance is far greater. Not just due to tolerance, but due to application. Two lenses that have the same FL and diameter and even that look good and nearly the same, may have entirely different aspherical formulas used to create their profiles and define their application. Of course, test with a lux meter and look at it. That’s how we do it now around forums. But it was just a perk involved. I wasn’t trying to focus on the lens analyzing part, but the thermal imaging—yes thermal imaging is something I know is important.

Making lights for yourself or for a few friends in small quantity, is one thing. Producing a light system for many end-users is something a little more nerve wracking. Some of the best light builders have fallen because they got burned playing with the fire of production. Some of us might know who but I’m not stating names.

I think it’s good to know designs and concepts well in certain fields you plan to cross, and the possible misfortunes, possible future problems, and how one might solve them on their own, before unleashing something of nuclear proportion into the “humble flashlight” world. :stuck_out_tongue: :bigsmile: A light you can buy at Menards is humble, for a flashlight. Thor’s Hammer is not humble, for a hammer. If you are satisfied with a light at Menards, there is nothing wrong with you. That’s what people know a common flashlight to be. Many do not have jobs/hobbies that require a target to be continuously lit at 1000, 1500, or even 2000 meters. They may just need to see the steps, or the sidewalk to carry the trash out, and 500-1000 lumens is mind boggling to them (no matter the CRI! :bigsmile: ). Some people however, do need this other ability to keep light targeted very far away and not have batteries die in 30 minutes. Those people are not on the flashlight forums, for the most part. They just buy what they need once it exists. A carpenter usually isn’t a member of a DeWalt power tools forum, even though he is their biggest form of customer and supporter. He doesn’t know someone could easily pull the motor out, pull the batteries out, replace them, and have twice the power in said tool very easily. But when DeWalt does just that and lets him know about it, it’s under his Christmas tree. (I realize I’m talking about business strategy now.) Realistically a MFGR is not going to design something intended only for a small forum to be happy with, if it appeals to far more users of a different application. They are going to design it so that the people who need it, get the best platform they can buy for the money, and it won’t blow up when they plug it in the wrong way—if they are a good manufacturer who stands by what they offer. Know your customers before trying to create them.

As I do my best to create new designs for a different consumer type and application than I see around here, I offer hints I learn that can help some people here in their quests with flashlights, but I am still well aware most of the things I am doing or speaking of are not typical to do or are not completely justifiable designs to own for the majority of flashlight users around here. People like to keep things simple and I get that part, part of the time. :slight_smile:

13

I have developed several thermal solutions, many involve potting and managing potting so it doesn’t work against the thermal design of the system. Most of my methods for heat management were specifically developed for the P60 system, because P60 in it standard form needs the most help. Below is a graphic cutaway of my P60 drop in:

!!

I had mentioned that potting can work against the thermal design. This is because, in a flashlight, there are two heat sources: the emitter and the driver. If the space between the two is filled with a thermally conductive potting material, then the higher heat source (emitter) will feed heat into the other, more sensitive component (driver). What I have done is created a thermal barrier between the two. In my drop in with interchangeable contacts, I’ve moved the driver closer to the emitter platform and this thermal isolation is very important…no longer can the heat from the emitter feed directly into the core of the driver. Each component has an independent thermal path.
The potting material does a great job diffusing any hot spots on the driver, but it doesn’t perform very well moving the heat over longer distances…like the 8mm it must travel from the center of the driver to the pill wall. To overcome this problem, I install a copper heat spreader just below the thermal isolation pad. This copper disc is shielded from the emitter heat and sits just 1mm above the heat sensitive components on the driver. The heat from the driver components travel through 1mm of thermally conductive potting compound to the copper spreader and then efficiently moves to the pill wall.

A few of the improvements I’ve made to the emitter portion: I’ve installed a thermal pad between the MCPCB and the reflector. Most systems rely on heat traveling from the bottom of the MCPCB to the typically thin pill platform. By installing the thermally conductive isolation pad, I am removing heat from the top side of the MCPCB as well -while the thermal pad is not as effective as direct metal contact, if you look at the cutaway, you will see it’s a much more direct path to the reflector shaft. The majority of the emitter heat will still pass to the pill platform, but with the isolation pad, I was able to reduce the Tj by 12°C at 3.8A…not too shabby. I use a non-hardning thermal paste between the MCPCB and the pill platform. This paste requires constant pressure -I don’t like the idea of applying too much pressure onto a small isolation disk onto the MCPCB. With thermal cycling, it can cause a short in the MCPCB emitter trace. If properly sized, the rubbery isolation pad will extrude itself as the reflector is tightened and fill every void between the reflector and the top side of the MCPCB. This exerts and maintains even pressure on the MCPCB pressing it into the thermal paste while providing a top side thermal path.

When I stack regulators, I always separate them with a heatsink. While the 7135 regulator does quite well coping with heat, it’s output is lowered as the temperature rises. When you stack 2 heat producing components on top of each other, the internal heat can become a problem, the copper heatsink between the stacked regulators carries the heat efficiently to the pill.

I also polish the reflector shaft -in a p60, the reflector shaft has very small grooves around the circumference -by removing these groves by sanding and polishing, you can double the thermal contact area.

Some of these thermal enhancements are specific to the P60, but I employ many of them into all of the lights I build.

14

Nice idea. Is there any way to apply that approach to something like the qlite, where stacking chips stick up so high that you can’t use the threaded retaining ring to hold the driver in place?

15

If I understanding what your asking: If you’re installing in a system that uses a driver retaining ring, you simply stack the regulators on the MCU side instead of the spring side…you can still install the heat sink between the regulators.

16

pflexpro, thanks for all your help.

I noticed that your website calls out

“P60 Drop In Fully Potted with Nichia emitter mounted on Noctigon Copper MCPB. If you’re looking for exceptional color, this is the emitter to go with. It’s light output at 1.7A is about 330 lumens for the B10 (4500K) emitter. Unlike the High CRI emitters that are very yellow/orange for 85CRI, these emitters produce a very pleasing neutral color with 92CRI. In addition to high power lights, everyone should have a Nichia for lower power but incredible color”

For the Nicha option. With the new 219C emitters will you be revising or releasing a new option with (better) specs? Yes I know “better”can be interpreted lots of different ways.

And as your an expert (and I have never owned a p60 i am about as ingnorant as is possible.)

Is 1.7 amps as far as a p60 can be pushed without limiting actual use?

17

For a 219B, 1.7A is not the limit, but my flashlights fall into the hands of the novice and enthusiasts alike, so what I show on the website is not pushing the limit.
’How hard you can push it’ is different from ’how hard you should push it’. Recently, I’ve started building the 219B at 1.9A…I think this is still in the safe range.
Soon, I will have some 219C emitters and will have to burn a few to decide where I will run them. With a new emitter, I build a test light -direct drive, then power it with a lab power supply and run the light at a given power level with no cooling for 10 minutes. If the light survives and the output looks consistent in my sphere, I increase the power for another 10 minutes until it fails or the output becomes unstable. Once I’ve determined a ‘worse case failure (no cooling)’, I build one for myself at the highest power level, then step the power down (usually about 20%) and thats the level I offer on my website.

I do push harder…often people contact me looking for something ‘pushed to the limit’, I try to talk them out of it for about 5 seconds…then I build it for them…it’s always fun.

I’ve tested the ‘thermal decay’ on many flashlights and have been able to match or exceed their thermal characteristics with a P60. I’ve posted my P60’s thermal enhancements to share my ideas on thermal management. Hopefully, MEM’s thread will inspire some new thermal ideas.

18

I understand that a P60 with good heat flow must have a good niche. Your solution is certainly innovative and thorough, flexpro. A good fit of the pill to the body is still important though. I don’t see how to help test it with FLIR, but inserting thermometer thermal probes must tell a lot.
I am starting to be convinced to pot drivers. I have connected ground tabs of stacked 7135s with copper braid and left extra braid sticking out past the top one.
I wonder whether drivers in the usual configuration cool mostly by air convection or by conduction to the pill. I have one light in which the driver floats, held only by the wires. That one suddenly dims after running for a while. So maybe conduction to the pill is important. Air flow is harder to predict than conduction. The more heat and the more space the more air flows.
I wonder why I haven’t seen drivers with MCPCBs. led4power uses multi-layer boards.

19

I think one of the biggest benefits for potting is driver cooling. I have tested the light output of a P60 before and after potting. Both potted and unpotted will start with the same output, but after a few minutes, the output from the unpotted driver will fall much more than the output after potting. The driver’s PCB is the main thermal conductor for the components, but to be effective, it must be soldered to the pill 360°…on both sides if possible. The potting compound I use has a higher thermal conductivity than the PCB material, and has the advantage of contacting the components (regulators, MCU, caps…) on 5 sides instead of 1 side as in the case of the PCB. The potting compound also has a relatively large contact area to the pill. In addition, the potting compound ‘bridges the gap’ on the edges of the PCB to the pill. When a few copper heat spreaders/heatsinks are added and potted in, it can make a real difference in the driver temperature.

I really don’t think convection adds much cooling to the driver. The available air currents are just too small and the temperature differential is too small, so cooling is really left up to conduction.

My regulator heatsinks are merely strips of 2mil copper soldered to the bottom of the added regulator. These strips are left a little long and make a 90° turn, then are soldered to the negative ring on the driver. When inserted into a pill, these copper strips press against the pill wall. If the added regulators are on the spring side, I solder the strips directly to the pill. In this case, the stacked regulators are kept cooler than the single layer regulators.

In the P60 drop ins, all of the heat in the system must pass through the reflector shaft and the wrapping material before going to the host. I remove the small grooves that are machined in the reflector shaft. These grooves limit metal to metal contact by 50%. Then I wrap the drop in with copper. I have found, with any wrapping material, each additional layer adds thermal resistance. Although aluminum foil is a ‘good’ conductor, it’s typically less than 1 mil thick and takes about 15 layers to fill the gap. I use copper (which has 2X better thermal conductivity), starting with 4 mil for about 3.5 wraps and do the final fitting with 2 mil copper foil.

20

Is the wrapping between the conical bottom of the pill, in the drawing above, and the flashlight head?