8x 7135: Wiring MC-E in Series

The problem:
Currently have the MC-E wired in parallel, 3.7v one LIon, ~2.8a from a 8x7135 from Kaidomain. Great right? Started to notice flickering and one die putting out much more light than the others. No good. I am assumming one die is eating up all the power while the rest suffer.

Proposed Solution:

Change the 8x7135 driver. Re-wire it to run 2x7135's to each die by 4 separate circuits. 2x7135s should put out my required 3.7v and 700ma required to run each die seperately.

If I can do the soldering, can you see any error in my solution? Any comments? Has this been done before??

It will work, but I don't see that the work is worth it.

There are many torches that have the 4 dies in parallel in an MCE or SSC7 and don't have any problems. Because the dies are so close, they maintain the same temperature and therefore the same voltage drop and current.

What caused one die to flicker ?
- a defect in manufacturing
- LED damaged during mounting to carrier
- inadequate heatsinking causing a high die temperature and one die just happens to be more prone to flickering.

Hi Jimmy,

Welcome to CPF.


I dont know if what you are saying will fix the problem. Yes if you did the wiring perfectly with no shorts etc, the light would probly run, but the driver may not be the cause of this. If it was my light rather than spending all the time it would take to try that I would just get a new driver. Shiningbeam sells a 3 mode 8x7135 drive for like 7$

As Mike said there are other possibilities including a poor electrical path.

If you have any scrap drivers laying around even if they only output 1A you could hook one up to the led and see if the dies balance out.

Thanks guys.

The problem manifests itself as follows: Imediately after switch on, the light output is at maximum. It begins to dim, sometimes to about 30% of the maximum output. This takes 30-60 seconds from the time it is turned on. It is then noticable that a single die is much brighter than the other three. After dimming, it "flickers" and all the dies return to about the same brightness with respect to each other, but they do not regain the maximum brightness. This process repeats every 15-30 seconds, going from bright to dim with one die brighter. The brightest die remains the same die, regardless of how many times the light has been tested. I have pictures, I will see if I can upload them later today.

I am 98% sure in my wiring, and I do not have another driver laying around that I can fiddle with. I have not applied thermal paste yet, the star board is simply pressed on the aluminum of the flashlight with a screw-in ring. However, the light body does get sufficiently warm to indicate heat transfer is taking place.

Here is my driver:
http://kaidomain.com/ProductDetails.aspx?ProductId=1770
(prefer one mode)
And my LED:
http://www.dealextreme.com/details.dx/sku.16145
(the LED is soldered across at the chip base to bridge all the contacts to parallel wiring. Might I have heated the chip too much and damaged it during soldering?)

A star board is a poor substitute for a real heat sink. And without any thermal paste between the star and your real heat sink, you are overly depending on the thermal properties of the star to keep the MC-E cool. The fact that your flashlight body is warm may be only a weak indicator of how hot the MC-E's junction is. You are dealing with essentially a ~10w light bulb. Try putting your fingers on a 4W night light bulb. It'll get quite hot within seconds of turn-on.

In addition, it is not clear if your star's thermal properties are any good to begin with. See this link, for example.

Put a thin layer of thermal paste between your star and heat sink and then re-test.

To check the uniformity of output for the MC-E dies, direct drive the LED at a low level. Try using either 2xNiMH or 1x123A. If you have a bench power supply, use that.

To eliminate the driver as a possible source of the problem, direct drive the LED at full power using 3xEneloops or with a bench supply.

I am with Justin on this. The very first place I would start is properly sinking the LED. You would be surprised how much difference a very thin layer of arctic silver will make (be sure its thin).

If that does not cure it since you don't have another driver around the direct drive method will let you know if the driver is the issue or not.

I applied the thermal paste and the one die is still brighter. The flickering is lessened initially, but still very present during the run. Perhaps my heat sink is not good enough.

Here is a picture through sunglasses. This is what is looks like currently:


I am also uploading a video of the flickering but it will take some time. I will post it soon.

That DX star board that you linked looks like it is a series board, not parallel. Of course, with DX, there is never any guarantee that the product and image coincide. Are you sure you got a parallel board, and how does the soldering look? I've found that weak solder connections to the tiny MC-E terminals can produce uneven die output.

I will check the connections and take some pictures. Here is the video of the flickering. Occurs most noticeably at 5 seconds and 55 seconds in.

http://www.youtube.com/watch?v=vX2RScVUIRA

Yes, it is the parallel board with independent connections to each die. The chip is soldered across the connections to the star board to bridge all the positive and negative wire.

So you have this independently addressable star, and have wired it in 4P (parallel) by soldering across the MC-E's + terminals and - terminals?

How do you connect to the MC-E? Did you solder an LED+ wire to the MC-E's + terminals and an LED- wire to the MC-E's - terminals? Or did you solder LED+ and LED- wires to one of the four pairs of + and - solder pads on the star?

jimmyufl said:

I applied the thermal paste and the one die is still brighter. The flickering is lessened initially, but still very present during the run. Perhaps my heat sink is not good enough.

Click to expand...

If you applied a thin layer of the thermal grease to the bottom of the star and it is held down firmly to a chunk of aluminum in a light body this is adequate.

Provided the LED was attached to the star when it came to you and it did not raise up when you bridged the legs, your heatsinking should now be good enough to attempt the direct drive. Your LED may have a low VF so it would be a good idea to try it by using a 3 volt primary or at least a li-ion that has all the over voltage drained off. If you have a low VF direct driving it off a fresh 4.2V cell could one or more of the dies.

If it lights up correctly then you know you have a short or a bad driver. If it does not then the short may be in the star. In this case you can solder a set of wires across the legs on each side and try the low voltage direct drive again. This will tell you for sure if your led is the problem.

Resoldered the chip and the wires. I dont think the solder was even enough along the chip. It looks like that has solved the problem.

Before:



After:




And for fun, the Intova in my fish tank:

And a pic of the beam:

:thumbsup: Way to persevere. I read a sign I really liked the other day that said something to the effect of. "Experience is what you get when you don't get what you want"