Using the 7135, and a story of heat

[Curious_character]

I recently got some regulator boards from DX which have three AMC7135 regulators on them. Although the 7135 is a passive current regulator, it can still be reasonably efficient as long as the battery voltage isn't too much higher than the LED voltage. So I've been looking into using them to modify direct drive lights. Bench measurements when driving a Cree emitter showed solid regulation down to about 3.8 volts at the battery. From there down to 3.2 volts, the output current drops about 100 mA per 100 mV of supply voltage decrease. The current is about half its regulated value at 3.4 volts. This means that the current will be regulated at least within a factor of 2:1 over the discharge range of a single Li-ion cell. Because emitters become more efficient at lower currents, it also means that the light output should drop by less than a factor of 2:1 from full charge to essentially discharged. With a 4.2 volt supply -- the highest anticipated voltage from a freshly charged Li-ion cell -- the regulator chips were cool, so no heat sinking or thermal management of the regulator board are necessary.

Encouraged by this, I put one of the regulator boards into an MTE light. The one I have originally had a Luxeon emitter, but I'd replaced the Lux some time ago with a Seoul P4. (I see that they now have a P4 installed.) This is a good light for experiments, because it has plenty of room for the regulator board and is easy to get apart. The original light was intended to run from two CR123A cells and has only a fixed resistor between the battery and LED. This resistor was replaced by the regulator board, and a single unprotected 17670 cell was used for power.

But what I found isn't what I expected at all:

As you can see, the light output dropped very rapidly to a little more than half its initial value, and continued declining for some time.

I verified that the current was holding steady at least during the initial few minutes, so that wasn't the cause of the steep drop in light output. The next suspect was heat. I watched the LED voltage for the first few minutes after turn on and found that it dropped from 3.77 to 3.40 volts. At constant current, LED voltage drops about 2.2 mV per degree C of temperature rise, so this shows that the LED temperature is rising about 168 degrees C! No wonder the light output is dropping!

For that light, I had used a star with Seoul P4 pre-mounted which I had purchased from DX. Removing the LED revealed a very thick layer of thermal compound under the LED. Thermal compound is really a very poor thermal conductor compared to metal (it just happens to be a lot better than air), so this mounting would have poor thermal conductivity. I cleaned it off and applied a very thin layer of Arctic Silver epoxy in its place. After curing, the LED voltage dropped from 3.76 to 3.62 volts in the same time period after turnon as before. The LED temperature rise is now about 64 degrees C -- 2.6 times less, and more than 100 degrees less, than before. The best I've done so far with an LED star mounted on a big heat sink is about a 32 degree C rise at 1 A, so 64 isn't bad for the flashlight.

Here's the result of the LED re-mounting, on the same graph for comparison:

The regulation is about what I expected, and it's pretty decent for such a simple and inexpensive setup.

c_c

 

[Northern Lights]

CC:
I got a few of those emitters from DX. They could not tell me what the Vforward is. They published the average Vf as 3.75 and through a long drawn out email string I could not get them to understand that particular emitter should come in 4 different Voltage bins each spanning .25 volt range. DX refers to this general product data sheet that only lists Vforward as the products over all range of voltage bins: http://www.seoulsemicon.co.kr/_homepage/home_kor/product/spec/W42180.pdf
to obtain the Vf; DX did not understand the binning spec for the P4 as outlined at:
http://www.essc.co.kr/_HOMEPAGE/home_kor/product/spec/P4_BL.pdf

I also noticed the P4 in the first picture is what you get, an unmarked star. You cannot tell what it is and the remaining pictures on that page link are marked zled. It appears you are not getting the genuine Zled star and emitter. I got some genuine Zled P4s from another vender, the same U bin but then the Vf was 3.2. and the stars were marked like the first picture. That was the tip off something was wrong, at that time I would have expected the U bin P4 to have a Vf at 3.25 volts.
Since I do not have a working constant current power supply I have not determined what the Vf is yet(http://candlepowerforums.com/vb/showthread.php?t=170864). You appear to have enough equipment to overcome the problems I encountered. Having stated the star mounting was inferior and almost defective and since they are not marked like a factory P4 star it appears possible it is an after market mounting by a second party. DX could not guarantee to me after I approached them that the LED was genuine. They only would say that their supplier said it was genuine. After about 2 months they have not untangled the mess. That being said I still am not sure what the Vf really is and wonder if that could be part of the problems you experienced. Were you able to establish a test bed value to obtain the Vf? Do they output enough to be a P4 or are they fake?

 

[Curious_character]

CC:
I got a few of those emitters from DX. They could not tell me what the Vforward is. They published the average Vf as 3.75 and through a long drawn out email string I could not get them to understand that particular emitter should come in 4 different Voltage bins each spanning .25 volt range. DX refers to this general product data sheet that only lists Vforward as the products over all range of voltage bins: http://www.seoulsemicon.co.kr/_homepage/home_kor/product/spec/W42180.pdf
to obtain the Vf; DX did not understand the binning spec for the P4 as outlined at:
http://www.essc.co.kr/_HOMEPAGE/home_kor/product/spec/P4_BL.pdf

I've bought Seoul parts from a number of different vendors, and haven't seen a complete part number from any of them. Although it would be useful to know Vf ahead of time (for example, low Vf parts would stay regulated to lower battery voltage with a 7135), it doesn't seem common practice for vendors to furnish this information. It's probably less common yet for customers to be able to choose on the basis of Vf.

I also noticed the P4 in the first picture is what you get, an unmarked star. You cannot tell what it is and the remaining pictures on that page link are marked zled. It appears you are not getting the genuine Zled star and emitter. I got some genuine Zled P4s from another vender, the same U bin but then the Vf was 3.2. and the stars were marked like the first picture. That was the tip off something was wrong, at that time I would have expected the U bin P4 to have a Vf at 3.25 volts.
Since I do not have a working constant current power supply I have not determined what the Vf is yet(http://candlepowerforums.com/vb/showthread.php?t=170864). You appear to have enough equipment to overcome the problems I encountered. Having stated the star mounting was inferior and almost defective and since they are not marked like a factory P4 star it appears possible it is an after market mounting by a second party. DX could not guarantee to me after I approached them that the LED was genuine. They only would say that their supplier said it was genuine. After about 2 months they have not untangled the mess. That being said I still am not sure what the Vf really is and wonder if that could be part of the problems you experienced. Were you able to establish a test bed value to obtain the Vf? Do they output enough to be a P4 or are they fake?

As you say, the star isn't from SSC, but in tests on a good heat sink it appears to be adequate thermally. I'm convinced that the emitter is genuine. I've measured the light output of quite a few, and don't see any consistent difference between the DX parts and any others -- or any discernible physical difference, either. The mounting is definitely poorly done and certainly not done by the LED manufacturer.

"U bin" is a reference to the light output, not the forward voltage. Any U bin or any other light output bin part can have a Vf anywhere from 2.9 - 3.8 v. So there's no way to tell from Vf what the light output bin is or whether a part is genuine (unless Vf falls outside that range, which is unlikely).

I mentioned in my earlier posting that I had measured Vf, but at the test current of about 1A. The SSC Vf spec is at 350 mA, which of course would be lower. If you're interested in measuring Vf for some reason, you can make an adequately constant current source by simply using a higher voltage (say, 10 volts or so) and a series resistor. But it's also easy now for almost anyone to make truly constant current measurements at multiples of 350 mA, by using the DX boards with 7135 chips. Just make sure the supply voltage is at least 0.2 volt or so higher than the load (LED) voltage, and not much greater than about 5 volts. Above that (when the load is an LED), at least with the three-chip version, the chips can overheat and reduce current. With the 7135 the input current from the battery is very nearly the same as the LED current, since the chip supply current is very small. So you can measure at either the battery or LED, whichever will disturb your system the least.

c_c

 

[Northern Lights]

"U bin" is a reference to the light output, not the forward voltage. Any U bin or any other light output bin part can have a Vf anywhere from 2.9 - 3.8 v. So there's no way to tell from Vf what the light output bin is or whether a part is genuine (unless Vf falls outside that range, which is unlikely).
c_c

Yeah, I know I meant type out the reference to the other stars I bought and not to indicate any other connection to that, but it was 4 am, and it definately does not make sense what I said but you figured out what I meant thanks, hey I am still up its 5 am:sick2:
Thank you for the tip on the 7135. I have been holding back on a mod of a Nuwai Tm-800x3 because I was not sure of the Vf with these. After a nap here, I will get on it today. Your information was very helpful

 

[scott.cr]

Fascinating. Thanks for the post!