XML runs hotter than MCE ... why?

Hi all, so I have a little quandary. I'm trying to figure out why an XML gets WAY hotter than an MCE in the same application where everything is identical except for the emitter (and star thickness). I can run the MCE on high for an unlimited amount of time if I hold it in my hand. It gets hot...but not "really" hot.

• MCE mounted to a 2mm x 20mm star.
• Battery is 1x18650
• Driver is a shiningbeam 2.8A 3 mode (linear)

XML mounted to a 1.6mm x 20mm star runs about 20 degrees (F) hotter than the MCE at the same 2.8A drive current (measured at the tail).

Several people have suggested the linear driver is giving off more heat because the XML vf is lower than the MCE. Plausible. However, after running the light for 10 minutes and the head is very hot, I can remove the head and put my finger on the driver...and it's barely warm. I will try and measure actual vf's tomorrow, but even if the XML vf is lower, it doesn't seem to be related to the driver and excess voltage.

As a side note, the MCE drops about 25 lumens between turn on and 30 seconds and begins to stabilize. The XML drops nearly 60 lumens and I assume this is due to the rapid heat build up.

Does anyone else have identical MCE and XML lights to compare? Any thoughts on what is causing this? I'm wondering if it's "me" or something inherent in the XML (or the MCE). I've observed an identical effect with multiple lights so it's not just a single "quirky" setup.

I haven't been industrious enough to bench test the two LED's side by side with a power supply, mainly because I don't have an accurate way to measure temperature differences. Thoughts? Advice? Thanks!

I'm curious!

I haven't got the numbers in front of me to do the math, but which one has the greater emissive surface area?

mvyrmnd said:

I haven't got the numbers in front of me to do the math, but which one has the greater emissive surface area?

Click to expand...

I'm not sure what you mean

You mentioned that in one of your other threads. I made no sense to me either why the more efficient LED should be hotter.:shrug: If they are both drawing 2.8 amps at full power, the higher Vf of the MC-E should create more heat. Are they both drawing 2.8 amps?

Sorry if this was brought up in the other thread, but have you put a DMM with thick probe wires at the tailcap of each light to actually measure current draw? A DC clamp meter would be better but more expensive. A linear regulator basically has the same current input as output. It's the extra voltage that gets burned off as heat. I'm thinking perhaps the higher Vf of the MC-E doesn't allow it to be driven to full power after figuring in the voltage drop of the battery under load. You mentioned AW's 2,600 mAh Li-Ions, IIRC. If it were just a question of voltage, IMR's might make it behave differently.:devil:

It may actually be because the XM-L has a better thermal path than the mc-e. So if you're going by feel of the light, the XM-L may be shedding its heat better therefore the light feels hotter. Whereas the MC-E could be burning up and you'll be less likely to feel it.

archer6817j said:

I'm not sure what you mean

Click to expand...

never mind. barking up the wrong tree.

qwertyydude said:

It may actually be because the XM-L has a better thermal path than the mc-e. So if you're going by feel of the light, the XM-L may be shedding its heat better therefore the light feels hotter. Whereas the MC-E could be burning up and you'll be less likely to feel it.

Click to expand...

I suspected the same so I tested the lumen output over time in my integrating sphere. The MCE sees far less drop in lumens over time...meaning the MCE is shedding heat more efficiently while the XML is going nuclear.

I also have some data...which is a bit puzzling. I tested the vF of both the MCE and XML while they were installed in identical heads. I removed the window and reflector, put on my welding hood, fired up the light and checked the vF on both samples.

MCE=3.135V
XML=2.994V

I suspect the numbers are a little low because the battery might not have been fresh off the charger. However, it does show the XML has a lower vF (same battery was used for both tests). The MCE is surprisingly good though. Don't know if this means anything or not

As far as drive current. I've measured at least 20 different lights and I get 2.8A (+/- 0.05V) with a fresh battery with both the MCE and XML. I have a Fluke DMM. Any more theories?

Looking over HarryN's thread on MCPCB's it seems like the XML boards I have might just be of "less than ideal" design, while the MCE boards may be quite good. Tomorrow I'll look at the traces and see if I can tell any major design differences.

All of my XML lights run cooler than my MCE lights. I'm curious if you have compared different sources of XML and MCE. Its possible that you have a really good sample MCE and a bad sample XML? Also, what is the bin of each LED?

netprince said:

All of my XML lights run cooler than my MCE lights. I'm curious if you have compared different sources of XML and MCE. Its possible that you have a really good sample MCE and a bad sample XML? Also, what is the bin of each LED?

Click to expand...

Yep, possible. All of my LED's are from Cutter...so I expected them to be similar quality for the LEDs and stars, but maybe not. However, what aspect of "low" quality would cause the XML to emit so much more heat?

The MCE is "M" bin and the XML is "T6" ...both are 5700K.

I just got in some "N" bin MCEs and "U2" XMLs so maybe I'll throw a couple of lights together for comparison.

I'm sure you already know this .....but if this phenomenon were related to the led itself ANSI would show higher readings from mc-e than xm-l right?

archer6817j said:

As far as drive current. I've measured at least 20 different lights and I get 2.8A (+/- 0.05V) with a fresh battery with both the MCE and XML. I have a Fluke DMM. Any more theories?

Click to expand...

The only thing surprising about all this is that there is any difference in heat level between the two lights. Not which light gets hotter. The fact that the XM-L is more efficient shouldn't make any difference. Because whatever power isn't used by the LED is burned off as heat by the linear regulator. The XM-L and the MC-E should be equally hot. Here is a discharge graph of the new 2,900 mAh batt's that AW is selling. Fresh off the charger you get ≈ 4.0 volts @ 2.8 amps or 11.2 watts, for both lights.

In the case qwertyydude mentioned, eventually the heat is dissipated at the surface of the light. So apart from the LED being seriously overheated, it still has to reach the same state of equilibrium (conservation of energy). If you made the copper star from your other thread and soldered the LED directly to it, the outside of the light gets hotter faster while the LED stays cooler. You can drive it harder, the LED lasts longer and is slightly brighter. Harder in this case meaning less likely to turn 'angry blue'. I have seen a few threads that call into question the thermal efficiency of the stars Cutter is selling. Assuming for a moment that's true, the heat still ends up at the outside surface of the light. It just takes a little longer.

One other possibility might be how does heat affect the battery. The XM-L's regulator being hotter and right next to the battery might have an effect. There is a discharge graph at various temperatures in the pdf file for the Panasonic NCR 18650. There is a slight increase in voltage of < 0.1 volt as the temperature rises from 25°C to 60°C. If the battery were 140°F (too hot to hold), it would cause about 2% power increase to the LED/regulator combo. So apart from personal safety issues, battery temperature is a non-issue.

Perhaps like you mentioned it's time to hook both lights up to a bench supply one at a time for a long term test. Put a muffin fan next to the head of the light to take the place of your hand. Since the MC-E has a higher Vf, turn the voltage up until the light is just drawing 2.8 amps. Check both lights at that voltage for an hour with a thermistor stuck on the head of the light with a blob of AS5. An 18mm wooden dowel or some paper towel in the battery tube would help to keep the heat inside to simulate actual use conditions. If the temp of the head gets hotter than 50°-60°C, get a bigger fan.

As previously mentioned, if your running an AMC 7135 based driver, the lower vF of the XM-L means that the driver needs to burn off more excess voltage and dissipate more heat.

It's possible that the way the heatsinking was designed, and the fact that the driver is running much hotter with the XM-L means that part of the light (that is maybe not heatsinked so well) feels hotter.

I'm not sure how / where you are measuring this 20F difference in temperature, but try measuring different parts of the flashlight body, not just one area. Take an average of several measurements then let us know your results.

Also try measuring the temperature of the heat coming out the front of the light at 1 or 2 inches. It could be that a more optimized reflector is somehow reflecting more heat out the front of the light, leaving the light running cooler.

It is also very important to do the test with batteries that provide identical performance and SOC. If one cell is older and sags a bit more, this could potentially mean better driver efficiency, vs a cell with less resistance and a higher voltage leading to lower driver efficiency.

Lastly, don't just measure current at the start of the test, but measure current at the end of the test. Keep in mind that if you do unscrew the tailcap, the voltage of the cell will rebound so it will draw more current for the first few seconds than before you pulled off the tail cap. I think there is a good chance that you will find after running the light for 10 minutes the current on the MC-E will be less than the XM-L, because vBat.

There are a LOT of factors involved that could contribute to your results.

Jason, are you using IMR cells or AW black label protected 18650? When I was in your shop the other night for the GTG I only saw a 1-2 orange 18650-IMR, and they were the ones brought by Craig and Jose (IIRC).

Question to everyone, is it possible the black cells are capable of driving the XML harder and closer to its thermal limit? Is it possible the black non-IMR cells are not pushing the MCE close to its thermal capacity?

The 7135 driver in Jasons design "floats" inside a copper collar inside the head/bezel. The board itself will generate some heat when Vin>Vf, but still there's not much of a thermal path between the bezel outside wall and the 1735 chips. I would be surprised if THAT much of the heat you feel on the outside is originating from the 7135.

I spent a lot of time with your MCE-neutral tint (great light BTW with the ledil boom), it did get warm but not alarmingly hot at all.

As a side note, my MCE P60 module draws 3.2A with an IMR 18650 and runs noticeably hotter. With purple Panasonic CGR cells it draws ~2.35A. I switch cell types depending on how much light I really need and how I intend to use the light.

Hey Kramer, I'm using the regular black cells but when I measure the tailcap current it's always very close to 2.8V with fresh batteries. I'm going to double check the driver temp in the near future. In the past, I'd run the light for 10 minutes till it was alarmingly hot, screw off the head real fast and stick my finger on the driver...barely warm. The driver is pressed into the copper collar so there is electrical contact, but not a "ton" of thermal contact. Initially I was putting a bead of thermal paste on the back, but I stopped doing that because it was messy and didn't actually seem to do anything. So, given your comments and the construction, I agree...even "if" the driver was getting super hot (which it's not) it wouldn't be delivering 20F worth of heat to the head. Also, the difference in vF is only about .15V. I don't know how to do the math, but it doesn't seem like that would be proportional to the difference in temperature.

Oh also, I've been measuring the head temp with an IR thermometer so it's not super precise, but I can confirm that the XML head is significantly hotter than the MCE head...as in it's not just in my mind I also discovered I have two types of XML with different MCPCBs so I'm going to make up two identical heads with the different LEDs and see if there is any measurable difference.

jasonck08 said:

Also try measuring the temperature of the heat coming out the front of the light at 1 or 2 inches. It could be that a more optimized reflector is somehow reflecting more heat out the front of the light, leaving the light running cooler.

Lastly, don't just measure current at the start of the test, but measure current at the end of the test. Keep in mind that if you do unscrew the tailcap, the voltage of the cell will rebound so it will draw more current for the first few seconds than before you pulled off the tail cap. I think there is a good chance that you will find after running the light for 10 minutes the current on the MC-E will be less than the XM-L, because vBat.

Click to expand...

Good points! I'm using new batteries that I top off on my hobby charger before each test, then I check the voltage. Thy are quite consistent. So to your first point...the reflector is clearly more efficient with the MCE. With no reflector (just an open head) the XML puts out 100 lumens more than the MCE. Once I install the reflector, they produce the same lumens, but the XML still throws farther. I'm assuming just because of the die size. Actually, I'll do a heat test with an open head to eliminate the reflector as a factor...thanks for the idea!

I'll also check the current at the end and compare...should have thought of that. I hope the XML doesn't melt my sphere The ambient is really high here today.

**EDIT** I should also mention, when I realized I was "loosing" lumens I thought those photons had to be going "somewhere" and doing "something" as a result. I thought maybe they were leaking to the side of the reflector and hitting the inside of the head, and not all going out the front. I took several reflectors and machined them down so the LED would sit deeper and deeper into the reflector. I expected the output would go up and I could reclaim those lost lumens. Turned out lowering the reflector had only very minor effect on the output. Mystery continues.

It sure is an interesting question about why this is happening. I guess you could always call Cree's customer service group and see what they have to say about your results.

It's interesting that you are getting 100 lumens more from the XM-L vs the MC-E. This would indicate that although the die itself is more efficient, the phos is working a lot harder. Since the phos is absorbing at 460nm and re-emitting at longer wavelengths, there is heat associated with the wavelength change, roughly:

Planks constant x the wavelength shift, if IIRC.

As a first order approximation, if you assumed that 1/2 the light was shifted from 460nm to 560nm (averaged out), then it is possible to do this calculation relatively easily.

Not sure if that is significant or not, but who knows. I certainly have been impressed with the heat that appears to be coming from the phos on high CRI Rebels.

The difference in Vf is very small for the two packages, so that does not seem like enough to matter.

BTW - Power to the LED package = Vf x current


Harry

archer6817j said:

Several people have suggested the linear driver is giving off more heat because the XML vf is lower than the MCE. Plausible. However, after running the light for 10 minutes and the head is very hot, I can remove the head and put my finger on the driver...and it's barely warm. I will try and measure actual vf's tomorrow, but even if the XML vf is lower, it doesn't seem to be related to the driver and excess voltage.
!

Click to expand...

With 2.8 for a linear driver and the driver is barely warm that tells me that the dissipation is not in the driver, but in the LED (obviously) but also in the batteries. 3V at 2.8 for the LED, 65% of that as heat (likely more at those currents) = 3 W. Fresh battery - (3.7-3.0) * 2.8A = 2W ... between the driver and the battery but it seems like mainly from the battery. How warm are your batteries getting?

Semiman

Didn't make it to the shop yet but I just test ran an XML light at home. I let it run on high for 10 minutes while tailstanding. It was almost too hot to touch while I unscrewed the head. I immediately dropped the battery out and held it in my hand. It was warm but not hot. The actual body of the light was far hotter. Then I stuck my finger on the driver and made sure to contact the copper post as well...both barely warm.

The heat is clearly localized in the head and then conducts down the body through the threads...that is to say...it doesn't "seem" like the battery is really contributing much to the heat buildup.

Is there much difference in the light spectrum for the two LEDs you are using ?