Thermal Test Wrap Materials and Resulting Junction Temperature

21 posts / 0 new
Last post
pflexpro
Offline
Last seen: 8 months 2 weeks ago
Joined: 12/26/2013 - 10:49
Posts: 171
Location: Georgia
Thermal Test Wrap Materials and Resulting Junction Temperature

Disclaimer: These test were performed for my personal knowledge and use. They resulted in the forming of my personal opinions (listed later in this posting)…there is enough information contained in the resulting data for you to form your own opinion…and you should.

I have been working on enhancing the thermal efficiency of a P60 drop in for some time and the one thing I found is ‘no matter what modifications are made to the drop in, the heat transfer ultimately depends on the efficiency of the thermal wrap material’. Search the forums and you will find everyone has an opinion as to the best wrap material, but no one seems to have real numbers or real answers.

Also, I couldn’t find the typical range for a P60’s junction temperature –do you know your junction temperature? For all I knew, with the best thermal modifications (at higher drive powers), I may be pushing the 150° envelope.

It is a little challenging to measure the Tj of a P60 drop in in a host, but not impossible.

My setup:
I mounted a emitter on star (XM-L2 U2 1C) in a brass pill, then dilled a tiny hole adjacent to the thermal solder joint through the MCPCB and pill platform. I ran a k-type probe through the back of the pill, through the star even with the solder joint and set it with a thermally conductive silicone. I had to bore out the small opening in the reflector an additional 2mm to accommodate the silicone and probe tip.

These tests were preformed without a driver installed in the pill, instead, I used a lab power supply to drive the emitter at 3.00A. I used 16 AWG wire for the leads since they needed to run through the battery tube to the power supply.

I wrapped the drop in with various materials and tested each and plotted the first 15 minutes while monitoring the Tsp and light output simultaneously. (exceptions –when I tested with no wrap material, my Tsp reached 112°C at 8 minutes so I terminated the test. Even though my calculation to go from Tsp to Tj only added 25°C to the Tsp, this only works if the Tsp is stable…if the Tsp is climbing –the Tj will be more than 25° higher than the Tsp…if the Tsp is climbing steeply, the Tj will be much higher than the calculated 25°.)

I ran the test setup with various combinations of MCPCB materials and wrap materials. I have listed a few of the most common combinations.

The test setup was mounted in an integrating device with contact only at the mounting plate on the head of the host. Ambient temperature was 25° C with a +/- .2°. The Tsp and light output was recorded simultaneously.

Your setup:
Can you look at my graph and determine what your Tj or Tsp is…yes and no. These tests were performed in a controlled environment with a controlled power source and the host did not have the benefit of air flow or most importantly your cold hand wicking heat from it. If your P60 has a 3A driver and a fully charged battery, you can look on the graph and find your MCPCB material, find your wrap material, run it, standing on it’s tail cap, at room temperature for some time up to 15 minutes, then look on the chart and determine your approximate Tsp or Tj…but the real reason for all of this is just to compare thermal wrap efficiency.

Now the Results:

Some Clarifications on the above chart legend:
Al star = Aluminum MCPCB 16mm x 1.5mm
Cu star = Copper MCPCB 16mm x 1.5mm(Noctigon)
Cu foil wrap = .002” copper foil 5/8” wide
Al foil wrap = Reynolds® Heavy Duty
Cu tape = Copper foil tape w/ conductive adhesive 1/2” wide

As promised, my opinions:
1: wrap it or don’t bother going with p60 (at least anything over 1A)
2: forget the copper foil tape –the thermal resistance of the adhesive layer is too high to make it an effective wrap…aluminum foil performs much better and since you probably have some in your kitchen…it’s basically free.
3: if you run 2A or above…use copper foil
4: not list on the chart but a poor performer is aluminum foil tape (same issues as copper foil tape)

Why did Copper foil tape do so badly? (another one of my many opinions)
Our friends at 3M® have very good resources for their products, they consider the adhesive on thermally conductive tape to be a good conductor at 0.60 W/m-K…for an adhesive, it is good. Copper has a nominal conductivity of 385 W/m-K. Without using a calculator, you can see that copper has a thermal conductivity several hundred times greater than the adhesive (641.667 times greater if you use a calculator).

Copper foil –the best, but still has a drawback: every time the heat cross a layer (or wrap), no matter how tight, you loose some of copper’s thermal conductivity (you may have from 10 to 40 wraps)…but there are ways to minimum this problem, which I have tested and will address in a later posting including copper corrugated and dual gauge wraps.

MCPCB Wrap Tj Final C
Copper Copper Tape 152
Aluminum Aluminum Foil 143
Copper Aluminum Foil 137
Aluminum Copper Foil 135
Copper Copper Foil 130
PflexPro Copper Foil 112

Now, Just for Fun:

During one test, at 32 minutes (after I had reached a stable Tsp) I decided to cool off the host before stopping the test. I read the host temperature with an infrared thermometer and decided I wouldn’t be cooling it with my hand. I used a freezer ice pack (the kind that never freezes hard) and wrapped it around the battery tube of the host. I continued plotting the resulting temperature and light output listed them on the same chart. Remember, the only change, at 32 minutes the ice pack was applied –notice how the light output increases as the temperature drops…just another reason to take thermal management seriously.

Edited by: pflexpro on 12/26/2013 - 11:44
djozz
djozz's picture
Offline
Last seen: 4 hours 27 min ago
Joined: 09/07/2012 - 17:04
Posts: 18078
Location: Amsterdam

Wow, you are clearly still in the middle of the write-up, no graph yet, but it sounds like a very thorough en informative set of tests. Eagerly awaiting the results Smile

Edit: now with the graphs: awesome results, thanks for doing this and posting! I would not have expected that wrapping would help a lot because of the many material/air transitions with those layers, that the heat has to go through, but you have proved different, in a very convincing set-up! 

 

Oh, and welcome to BLF!  Smile

dchomak
Offline
Last seen: 3 years 5 months ago
Joined: 03/17/2012 - 20:10
Posts: 4122
Location: Connecticut

Welcome to BLF. A lot of good stuff here, looking forward to reading more.

I did a simple experiment awhile back, where I supplied a constant voltage to a P60 light and measured the current. It rose as the temperature of the pill rose. What would be interesting to me is what was happening, lumens wise, as that current rose. Supplying a constant 3V DD initially pushed 1.6A thru, but as the temperature increased, it rose to 2.7. That is when I shut it down. What would be interesting is seeing a graph of lumens vs current under those conditions.

pflexpro
Offline
Last seen: 8 months 2 weeks ago
Joined: 12/26/2013 - 10:49
Posts: 171
Location: Georgia

You’re right about having thermal loss each time the heat must travel from one layer to the next…I have tested 10 layers of .002 copper foil vs solid .02 copper sheet. I haven’t compiled the info to make it presentable yet, but the solid sheet performs much better. Copper is so good, even with the wrap layers, with good pressure, it still seems to be the best for heat transfer.

nikanon
Offline
Last seen: 6 years 4 weeks ago
Joined: 08/25/2012 - 08:50
Posts: 439

Welcome to BLF pflexpro ! Party

You’ve proved something I already assumed, the widely available 3M copper adhesive foil (which is also quite expensive) performs not as good as al-foil everyone got in his/her household.

Thank you for your test and presentation !

lightme
lightme's picture
Offline
Last seen: 5 years 8 months ago
Joined: 11/12/2012 - 19:51
Posts: 615
Location: WV

I read that the surfaces of objects can’t touch each other because of all the negative electrons repelling them apart. The closer the surfaces are, the more they repel. A test was done with annealed copper and it was found that the rate of improvement in thermal transfer diminished with increased pressure beyond a certain point.

Chloe
Chloe's picture
Offline
Last seen: 5 years 4 months ago
Joined: 01/30/2013 - 14:14
Posts: 4595
Location: 联合王国

Welcome to BLF! Party Are you selling these PflexPro MCPCBs/drop-ins? Thanks for the tests! ^^

FlashPilot
FlashPilot's picture
Offline
Last seen: 5 hours 4 min ago
Joined: 05/10/2010 - 16:07
Posts: 4814
Location: USA

Welcome to the forum! Its great to have you here.

Somewhere online can be found a graph that shows thermal conductance ratings of Cu alloys. Pure copper is obviously the best, whereas any common CU alloy becomes far less efficient as the alloy percentages increase. It was enough to make me scrap the use of copper pipe fittings in building high powered LED flashlights. Im still looking for the graph. If anyone finds it, please feel free to post it here.

Before getting rid of my entire P60 collection, I used a heavy gauge specialty roof flashing (with low alloy content) that I cut into precise strips, then wound so tightly around the pill that I had to turn the head with a large strap wrench to fully seat the coils inside. I even cut a thin coil to contact only the pill and then a larger coil to fill the gap so there was no air space inside. All Al to Cu contact had a very thin layer of AA, mostly to prevent galling while turning the assembly. That is when I discovered that the P60 concept should have been discontinued the moment they made the jump from incan to LED. Sure, the design can be improved with copper wrap, but it completely circumvents its only attraction, which was to quickly swap dropins from one to another.

Probably any cheap similarly sized LED host will greatly outperform the best copper wrapped P60.

FlashPilot
FlashPilot's picture
Offline
Last seen: 5 hours 4 min ago
Joined: 05/10/2010 - 16:07
Posts: 4814
Location: USA
Chloe wrote:
Welcome to BLF! Party Are you selling these PflexPro MCPCBs/drop-ins? Thanks for the tests! ^^

Gosh I hope not, but it is starting to smell like the makings of a sales thread (complete with seller derived performance evaluations). IMO, completely avoid the P60 design for builds in access of 2 amp. There are far better options available for the price.

Don
Don's picture
Offline
Last seen: 1 year 1 month ago
Joined: 01/12/2010 - 16:32
Posts: 6617
Location: Scotland

FlashPilot wrote:
That is when I discovered that the P60 concept should have been discontinued the moment they made the jump from incan to LED. Sure, the design can be improved with copper wrap, but it completely circumvents its only attraction, which was to quickly swap dropins from one to another. Probably any cheap similarly sized LED host will greatly outperform the best copper wrapped P60.

 

Probably would, but that wouldn't be as much fun. They were probably fine till LEDs capable of being driven past 2A became available. At least as long as you don't use a stainless steel host. 

 

But these days there is no point to the P60 as you say.

 

The numbers from my light tests are always to be found here.

https://spreadsheets.google.com/ccc?key=0ApkFM37n_QnRdDU5MDNzOURjYllmZHI...

pflexpro
Offline
Last seen: 8 months 2 weeks ago
Joined: 12/26/2013 - 10:49
Posts: 171
Location: Georgia

I have made some of the PflexPro drop ins to sell -it has other thermal enhancements that puts it lowest on the temp chart. The wrap testing was a necessary final step in developing the pflexpro.

Hikelite
Offline
Last seen: 8 months 1 week ago
Joined: 07/13/2011 - 16:18
Posts: 3888
Location: Honeyland

Very nice chart. 110C vs  about 72C at 7.5minutes it's a huge difference.

NightCrawl
Offline
Last seen: 5 years 8 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

How can the pflexpro run so much cooler?

Chloe
Chloe's picture
Offline
Last seen: 5 years 4 months ago
Joined: 01/30/2013 - 14:14
Posts: 4595
Location: 联合王国
pflexpro wrote:
I have made some of the PflexPro drop ins to sell -it has other thermal enhancements that puts it lowest on the temp chart. The wrap testing was a necessary final step in developing the pflexpro.

Ah, thank you! Can you share any of the differences in the MCPCB compared to Noctigon? Is it thicker? Nevermind, there is this list from the auction:

Quote:
Thermal Enhancements:
  • The lead wires are silver plated, oxygen free copper with Teflon Insulation -(the thin Teflon insulation along with flat solder joints allow the use of a thermally conductive pad under the reflector)
  • The thermally conductive pad is used to give heat from the emitter an alternate and more direct path to the reflector shaft (the reflector shaft passes the majority of the heat to the flashlight body)
  • The reflector shaft is smoothed to 2000 grit and sealed with thermal compound (this step lowers the Tj about 4 degrees C)
  • The emitter is mounted on a copper Noctigon MCPCB (this moves the heat away from the emitter more efficiently )
  • A copper foil thread wrap is used when installing the reflector to the pill (this lowers the thermal resistance of the threads)
  • A thermal Isolation pad is installed between the emitter platform and the driver…this step is important for a potted drop in: the bottom of the platform on a drop in running at 3 amps with good
    thermal management can easily reach 100C -the driver has a maximum temperature of 60C. The potting material creates a thermal bridge from the bottom of the pill’s emitter platform to the
    driver components. This isolation pad blocks the emitter’s heat from feeding directly to the driver.
  • Driver components are fully potted -this protects the driver components from thermal shock, mechanical shock, and physical damage (potting material: Stycast 2850ft with 12lv catalyst -widely
    used in the aerospace industry)
  • The potting material is set back from the edge of the pill to ensure compatibility with hosts that do not use the large outer spring
  • The battery contact spring is copper alloy for low electrical resistance
  • The P60Flex has signal contacts running through the potting material and with electrically conductive paint you can ‘connect the dots’ and in just a few minutes have a different mode group
    (Typically, if a driver has mode group options and is completely potted (both sides) you must choose the mode group before potting and and it can never be changed.

lightme
lightme's picture
Offline
Last seen: 5 years 8 months ago
Joined: 11/12/2012 - 19:51
Posts: 615
Location: WV

NightCrawl wrote:

How can the pflexpro run so much cooler?

Wrapping the threads isn’t specifically mentioned above but it is in the ad. All the other mods will have minimum effect without bridging the gap between the threads since that’s the main path from the pill to the outside world!
Tom E
Tom E's picture
Offline
Last seen: 56 min 16 sec ago
Joined: 08/19/2012 - 08:23
Posts: 14934
Location: LI NY

This is all great information! I'm glad you joined up and posted this very thorough write-up on the testing method and results. Sure, there are some negatives with the P60 design but it's still very popular and one of the rare true form-factor standards in the LED flashlight industry. Every few days, it seems like, there is someone asking about availability of higher end P60 drop-ins.

Your published test results, along with the details is all too rare to find on our forum. Please just ignore the negative posts -- most of us will appreciate the insight, and it really doesn't matter you sell a P60 drop-in. This has never been a factor before and personally, I wish more sellers would post as much detail technical info on their products as you do.

I myself, have read a lot about wrapping and other P60 techniques, but without the data, was never sure what was a truly better method. Of course I've heard about the copper vs. aluminum wraps, but never actually 'knew' copper was any better. I don't have that much practical experience with P60 lights, though I've built up a few P60 pills.

skyrider1
Offline
Last seen: 2 years 5 months ago
Joined: 11/18/2012 - 21:54
Posts: 762
Location: USA

Great information. Always love threads about p60 lights. If you search around there are a few good threads about heatsinking. I think what was concluded was that after added a copper star, the next junction to take care of is the star—>pill—> reflector. Someone bypassed the pill portion with copper on top of the star mated to the reflector so the heat flows directly into the reflector. This is really good because wrapping a drop-in pulls heat away from the reflector. In a typical set up the heat has to first get to the pill and then the reflector through the threads which is quite inefficient.

skyrider1
Offline
Last seen: 2 years 5 months ago
Joined: 11/18/2012 - 21:54
Posts: 762
Location: USA

Found the thread by lilkevin. Great information in there too.

http://budgetlightforum.com/node/22761#node-22761

I think the ultimate set up for heat would be:
- copper board
- copper mod in linked thread
- replace spring with copper ring (also mentioned in aboved thread + a link to lilkevin’s thread on CPF with data)
- copper foil wrap drop in
- thermal grease on threads
- lapping/filing flat the brass pill as the surface is usually quite uneven
- copper rounds on underside of pill that touches the walls of the pill (more thermal mass + improve the heat transfer of the original thin brass shelf)

texaspyro
Offline
Last seen: 2 years 4 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593
skyrider1 wrote:
I think the ultimate set up for heat would be

Also try thermal paste on the copper wrap… thermal conductivity sucks, but it is a LOT better than any air trapped between the layers of wrap.

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 day ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley

lightme wrote:
I read that the surfaces of objects can’t touch each other because of all the negative electrons repelling them apart. The closer the surfaces are, the more they repel. A test was done with annealed copper and it was found that the rate of improvement in thermal transfer diminished with increased pressure beyond a certain point.

That is definitely not right, as you sate it. The layers are electrically connected, so all the net charge goes to the outside. Also, the voltages are too small to cause measurable electrostatic forces. Clean similar metals can be welded together, by pressure only, at room temperature, because attractive forces dominate at very small distance. Of course there is repulsion between the atoms that keeps the material from being compressed to much more than its normal density.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 day ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley

The title would be more specific if it had “P60” in it.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。