8x7135 3040mA + xm-l current regulation graphs

This was a quick test, not too precise (only 6 measurment points), but it helps to determine, how will work the driver with difference batteries with difference internal resistances.

Keep in mind, these are estimated values, because of the difference forward voltage of the leds, and you know, the Vf decreases when the led heats up.

So I tested the driver with an xm-l from 4.2V to 3V with difference resistors. The resistor values were 80, 130, 180, 280milliohm. These resistances are quite near to the common used 18650 batteries's internal resistance.

I did not calculate with other resistances which can affect the regulation in the light, like switch + spring (tailcap) etc. Switch + spring at least 20-30mohm but much rather higher.

So an example: if you have a battery which has 100milliohm, + a tailcap with 30milliohm, you can follow the 130mohm graph. Or, a battery with 140milliohm + a tailcap with 40milliohm, it this case you can follow the 180mohm graph.

Not an easy thing to expain in english for me, but I hope you will understand. :)

the meaning of the voltage: un-loaded battery terminal voltage

nice ty that's show how crap batteries(high res) sux with high current, same if you have bad contacts

but don't understand 1 thing, you used just 1 battery for all test adding resistors? which 1?

Thank you very much, viffer, for the measurements. They are very good to have and show the big impact a voltage sag can have on light output. I had thought of doing something like that myself.
I would like to ask if you gave the XM-L time to warm up as the forward voltage of XM-L changes by as much as 0.3V (approx. 3.33V to 3.0 V) during heat-up when it burns 9 Watt in a normal cooled flashlight. That could affect the interpretation of your curves.

I used a lab power supply, and with it I can simulate a battery with an additional resistor as you see above .

Yeah, that could affect, but do you think, that the forward voltage of the led changes 0.3V? That's unbelievable high value. I think it will be somewhere 0.1V. But I will try out this.

Can you do the test for 6x7135(2280mA)?

It certainly can. The temperature coefficient of Vf is -0.003 V/'C according to the data sheet. The temperature rise is about 108 'C @ 9 Watt (I use a spreadsheet to calculate that). That makes -0.3 V (-0.003 times 108)

As I see the datasheet, that's true . It depends on the efficiency of the cooling, how will increase the die temperature. So, I tried it out with my test configuration (a luma power clone), heated up the head to 50°C, and the voltage shifting was only 0.07-0.08V. I measured this on the driver/ battery side, not on the led.
I checked the measurment configuration (the resistances), as I could, so it should be quite accurate.

Finally, I found out, that when I turn on the power supply and the led gets the current, the led die heats up to a relative high temperature immediatly within 1-2 sec, but my DMM is too slow to catch the voltage changing. Starting from this, the 0.2-0.3 Vf changing is imaginable, but totally irrelevant in this case.

sorry, I don't have 6x7135, but 2280mA is significantly lower load for a 18650 cell than 3000mA,

so the current should be regulated in much wider voltage range

What about DMM resistance?

btw, LED die need 1-5min for total temperature stabilisation(if you have realy good cooling)

The 80mohm curve has been measured with only the DMM self resistance. I heated up to 50°C the light body with the led, you can see the result above in my post 7.

That is food for thought right there!

Goes to show how little time I have before regulation ends.

Makes me want to make a parallel wired 18650 battery body even more bad.

That would basically double the time the lights stays regulated. Right?

Thanks for doing this. I can see it takes some effort for sure.

I was curious, and I measured how decrease the Vf of an xm-l.

Led current was 3A during 4mins. The first measured value was 3.206V and it decreased to 3.131V at the end of the test. The minimum sample rate of my DMM's software is only 1sec, so dont know what happened in the first second. For this, it would be better an oscilloscope I know.

I used a not too heavy aluminium plate for cooling the led, which wasn't heavier than a flashlight body.

so from 3.206V to 3.131V

Thanks for the tests!

I didn't even know than Vf changed with temp. It is opposite what I would have thought. Similarly, internal resistance isn't actually constant either.


That tells us that the temperature of the LED junction rose 25 degree celsius ((3.206V -3.131V) div. by 0.003) from 1 sec after switch-on to 4 min. after.

From that I think you could calculate the thermal resistance of your heat sink

Good post guys

an additional graph, the pure regulation without any resistance, from 2.9V to 3.7V

the input voltage was measured directly on the driver

Thanks for the resistance tests Viffer. Ive been going through all my lights with a DMM to identify any resistance bottlenecks wherever possible. The differences in a good electrical path can be amazing. My 7135 based lights will thank you for this! :bigsmile: