LD-2, some testing of two prototypes of led4power's new driver board

Flashlight Modding and DIY Parts
1

Introduction

A month ago I received two prototypes of a new flashlight driver from led4power, named LD-2, with the request for testing them, especially the parasitic resistance. I was of course very honoured that I was to test his brand new driver (who am I to know it all about drivers? I don't), but I felt also a bit uncomfortable because testing them well involves the time and dedication more suited for someone retired with a lot of time on his/her hands than me: busy with work, family and being worried about lots of things, and in a constant state of sleep deprivation. Further, my electronics knowledge does not go much further than Ohm's law, even though in my experience in flashlight country you actually get a pretty long way with that simple equation :-) . I was somewhat relieved to hear that led4power also sent samples to HKJ, so wherever my skills stop, I'm sure he will fill that in brilliantly. Due to lack of time I almost stopped halfway, but in the end I found some time to at least gather enough numbers to do this post. So here are my findings, I did not do all the testing that led4power requested, but I also did a few extra things. I should present the numbers in nice charts. I may make them (it is quite labor intensive for me), but for time's sake I make do with the raw collected data, which I find quite readable, neat charts do not make the data that much informative IMO.

What is tested here?

As the LD-1, the LD-2 is a 17mm diameter lineair flashlight driver made by BLF-member led4power, that is current-regulated at all modes (so no PWM) by controlling the current throughput of a FET. So any overvoltage not required for obtaining a desired current is burned off by the FET, which at times will dissipate a fair amount of heat. Compared to the LD-1, (at least) the following improvements have been made:

-the circuitry is completely redesigned.

-the driver is regulated at a user definable current in high mode by adjusting one resistor (was fixed at 5A in LD-1). The prototype boards were set at 6A, a triple can be set to even 12A if needed (the light and battery must be able to deliver that of course).

-parasitic resistance significantly lowered to almost BLF17DD level (parasitic resistance of a BLF-DD-driver is ~4.8mOhm, see post#17)

-has a user adjustable physically separate moon mode by choosing a suitable resistor in series of the led (not enabled on the prototype boards yet )

-apart from a heat protection on the driver board, there is an option for a remote heat protection using a NTC-thermistor that can be glued to the pill or ledboard: an output stepdown is only happening when overtemperature is actually detected.

-the board has opposite through hole pads for the ledwires suitable for 20AWG wires.

-it has multicell support. One of the two prototype boards was set up for 2xli-ion use, able to drive 6V leds. The drain in sleep mode is mode is much lower than the usual 'zener modded' DD-drivers developed at BLF. The current of the prototype board was set to 6A which is a bit low for a MT-G2 or XHP70 (harsh on the FET with full batteries), but I tested it with them anyway :evil:

Measurements

First I did the same measurements on the 1xli-ion prototype as I did on my test of the LD-1 nine months ago. The driver was connected to a power supply and in a bench set-up a dedomed XP-L V6 2A (dedomed just because that was at hand) on a DTP copper board on a cooled copper mount was driven at various set voltages as measured at the driver 'entrance', to simulate the various voltages that flashlight batteries may deliver to the driver. The current was directly measured at the 'battery side' of the driver and also by measuring the voltage over a 0.01Ohm sense resistor that was placed in between one of the led wires. Those currents should be the same of course, it was a bit of a waste measuring the current twice I admit :-( . I also measured the light output of the led, also a bit of a waste because I had measured the current/light output relation of this very led already before.

What can be seen here is:

*on the high setting, the driver nicely tries to regulate the current at 6A but only manages that above 4.1V which even the best high drain cell fully charged will not deliver at 6A. This is not caused by the driver, but by the relatively high voltage of the XP-L, direct drive on a single cell, 6A through a XP-L will not be reached. (to compare: below is a test with this driver with a XM-L2 that is in regulation at a lower voltage). Low voltage stepdown to low mode kicks in at 2.74V

*med setting is nicely regulated at 1.2A. Low voltage stepdown to low at 2.74V

*low setting is regulated at 120mA. Shutting down of the driver occurs at 2.74V (edit: not sure about the actual voltage, but it does shut down completely)

________________

Being on a bit of a learning curve, for the rest of the measurements I skipped one of the current measurements and used the extra DMM to measure the resistance of the driver from battery-minus-in to FET-out (btw, the batt+ is directly connected to the led+ wire so no resistance in that path).

________________

To compare the parasitic resistance of the LD-2 to a 7135-regulated driver I had to make a 7135-driver of comparable output current. Unfortunately that is 16 7135 chips. Stacking-fun on a 105C ! :tired:

(stereo picture)

So just for the high setting, I again measured the LD-2 prototype driver on the XP-L, but now with driver resistance measured. Also I measured the same led with the same set-up, but with the 16x 7135 driver. Finally I measured the LD-2 prototype also with a XM-L2 (older batch)(to be fair to the driver: due to the lower voltage of the XM-L2 the regulation at 6A starts between 3.7 and 3.8V already). For all measurents, the entrance voltage of the driver was set, the other values were measured. The driver resistance was simply calculated by dividing the voltage over the driver by the current.

In blue the 16x7135driver measurements, in red the LD-2 on XP-L, in green the LD-2 on XM-L2, in pink the driver resistances for the blue and red test series. The driver resitance for the LD-2-XM-L2 series is not calculated, but is almost identical to the LD-2-XP-L series.

What can be seen is:

*In getting the 6A current out, the LD-2 performs a bit better as the 12x 7135, and that is due to a lower parasitic resistance: when the driver is out of regulation the resitance of the 7135-driver is around 20 mOhm, the LD-2 is around 6 mOhm. (for perspective: a steel spring is about 50 mOhm, a copper alloy spring about 15 mOhm, a proper Omten switch 4 mOhm)

*with the XP-L led no driver is able to regulate current at 6A on a single li-ion cell, but with a (older batch) XM-L2 regulation by the LD-2 starts at 3.7/3.8V, a high drain 18650 cell can actually deliver that for a few minutes (if no other voltage losses occur in the flashlight ). BTW, this is not something that is specifically relevant for the LD-2 but for all single cell drivers, same for the BLF17DD driver.

______________________

I did a brief test to see if the remote temperature sensing worked. For that, as instructed, I glued a small thermistor (came with the prototype board) to the surface to be measured, in this case an old aluminium ledboard with an XM-L emitter, and connected that to the driver. I ran the driver on high setting on a 18650 IMR cell while pointing the led towards a luxmeter, and this is what happened:

Nicely, a stepdown to medium happened when the ledboard got too hot. Even further stepdown followed when even medium setting was too high (as this bare board has no heatsink whatsoever), then as the temperature went down again, the led output went up again. Well done! In an actual flashlight all this will be slower of course as there is actual heatsinking present then. But I'm quite impressed that this set-up actually saved the old XM-L led on a non-DTP aluminium board floating in air!

_________________________

Now on to the 2xli-ion board. This is a great new option in the LD-2. The delivered prototype was set at 6A, which is ok for say 2xXM-L2 in series, but low for a MT-G2 or XHP70 board, because on two fresh high drain cells, those leds in direct drive draw up to 9/10A, so if that has to be downregulated to 6A, i.e for the MT-G2 about 0.5V will have to be burned (at 6A) by the FET, that is 3W of heat for a component with no dedicated cooling!

Nevertheless, I checked the board with a XHP70 led . Again with current measured between driver and led with a sense resistor. The switch is a forward clicky used as electronic switch for the driver:

This is the test-mess, the wires on the bottom go to the power supply:

(left DMM measured voltage over the driver, top right DMM: voltage over the 0.01Ohm sense resistor (mV), bottom right DMM: voltage at entrance of driver. The led is under the fan, facing down into the integrating sphere.)

The results:

What can be seen:

*it works!!

*there are are several remarks written down of unstable flickering light output/FET voltage in this test at moments when the FET had to burn relatively much power. Of course voltages above 7.8V are unlikely to happen in a flashlight situation, but it also occurred in medium modes, in which case the 8+ Volt is very likely. After consulting led4power, he asked me to check if this also happened on two 18650 batteries in series as power source. I checked that with two Sony VTC4 high drain batteries, and the flickering did not occur in any mode. Then I tried to reproduce the flickering behaviour again with the power supply, and it did not happen again, also not in the flashlight mod that I have done later with this driver (see reply #1). Led4power thinks it is something coming from my power supply or from the switch or what, but not from the driver. I do not think it is likely that my power supply is the source of the trouble because as far as I know, it is a straightforward good quality lineair power supply (not a switch-mode one)(EA-PS 3016-20 B), but it could have been the switch not making good contact, or something else, I don't know.

*the driver shuts off when the temperature (measured somewhere on the board) gets too high. It only happened on high setting above 8V, a situation not likely in a flashlight. But in the bench set-up, the driver is not also receiving the heat from the led, so who knows what happens inside a flashlight...

*the parasitic resitance (driver resistance when the driver is out of regulation) of the 2xli-ion version is comparable to the 1xli-ion version, around 6.5 mOhm.

*low voltage step-down and eventual shut-off occurs at 5.85V, which is quite perfect for a 2xli-ion setup.

___________________________

Conclusion

A weak point inherent to this type of driver is the amount of heat produced by the FET in certain situations (not always, the driver gets away with it most of the time, and there are heatsink solutions for the FET thinkable). That said, the LD-2 has a whole load of improvements and extra features compared to the LD-1, and in these tests they seem to work. From what I can measure, I think this is a great successor to the LD-1, and one of the best flashlight drivers around!

My own conclusion: driver testing is more challenging for me than I like, and not completely my cup of tea. But I must admit that I have said that about emitter testing earlier..

In the next post will be the build and observations of a mod that I did of my Shadow TC300 flashlight, using the 2xli-ion LD-2 prototype and a 4000K MT-G2 led. It works together nicely! When time allows I will do the post, I hope within a few days.

EDIT May 11th, 2015: I measured the parasitic resistance of a direct drive FET-driver as well, in post#17

2

Almost two years ago, when the MT-G2 just hit the market, I did an output test of that emitter on the brand new MT-Noctigon board that suggested that it would survive being directly driven by two 18650 cells in series. Of course I had to try that and modded a Shadow TC300 flashlight that I got from a cnqualitygoods sale with a 5000K MT-G2, without driver ( https://budgetlightforum.com/t/-/19600#comment-420408 ). It is still one of my favorite lights, kind of a building site lighting at 2500 lumen. But without lower modes and protection circuitry it was kind of unfinished, so it was a good candidate to check out the 2xli-ion LD-2 prototype. I also wanted a 4000K MT-G2 instead of 5000K and I had one waiting so that could go in as well :-) .

First, the driver board size for the TC300 is 20mm, so the LD-2 was soldered onto a brass ring. I had to ream the ring with a knife to make it clear the driver components. A steel spring was soldered on the driver, and then it was bypassed by 18 AWG wire. This is my new favorite way of bypassing springs at the moment.

The heat detection thermistor was glued to the backside of the MT-Noctigon board with Arctic Alumina adhesive (no electrical contact was made with the board this way).

One of the led wire holes in the pill was made bigger with a file to make room for the thermistor. Then the tricky business of soldering the thermistor wires, led wires, closing the pill and screwing down the reflector without shearing off the thermistor began. I failed and ruined the thermistor in the process, but a second try with a new thermistor worked well. (I made the mistake not to screw down the ledboard, even though there were screw holes in the pill. Well, it worked in the end :-) )

I got the light with the new driver working, and the performance of the driver is a total succes :-) . The only thing that went wrong is that to fit the led behind the reflector I had to slice the silicone off the side of the emitter, and now after lighting up the dome is in the process of detaching from the emitter, with all usual consequences: tint change and output loss. That never happened to the 5000K led that was in this light before and was tortured way more than this new 4000K one. This was the led (pictured through a cd) yesterday, the yellowish part is where the dome comes loose:

Now after some more playing around with the light it has even come more loose. I may have a go at removing the entire dome, at the risk of ruining it further.

But never mind that, the beam is still fine, the output is a just a bit less and warmer tinted, kind of like that anyway :-) . What matters is how the driver performs.

First this evening's output: on high(6A) 2010 lumen (30 seconds after switch-on, yesterday it was a bit more ), on medium(1.2A) 605 lumen, on low(120mA)63 lumen. I like this mode spacing!

Then I did a runtime test on the Sony VTC4 cells. I started with fresly loaded Sony's, this is how it went:

-switch on on high setting: driver board thermal shutdown occurred after 52 seconds.

-I switched the flashlight off and let it cool down for 8 minutes

-switch on on high again: remote thermal protection occurred after 50 seconds: stepdown to medium mode

-I switched the flashlight off and let it cool for 8 minutes

-switch on on high again: remote thermal protection occurred after 40 seconds: stepdown to medium mode

-I switched the flashlight off and let it cool for 8 minutes

-switch on on high again: remote thermal protection occurred after 43 seconds: stepdown to medium mode

-I left the flashlight running at medium mode: after 90 minutes a stepdown to low occurred (low voltage protection). 7 minutes later the driver switched the led off.


Well, that went well, all protections in this driver work inside a flashlight as well, and I think in the correct way :-) .

One perhaps minor point: I know from two years ago (see the link at the beginning of this post) that this flashlight can handle more heat than the current thermal protection setting of this LD-2 prototype allows, in fact I know that with the DTP copper boards that most of us use nowadays, the flashlight body temperature can be allowed to rise way above 'too hot to hold' without much output-drop or led failure. So perhaps the treshold for the remote sense thermal stepdown could be set a bit higher. Or, probably easier, the thermistor can be mounted further away from the led, i.e. somewhere on the pill.

I can not find my two unprotected NCR18650B batteries at the moment (protected ones are too long for this flashlight), but I think that using less high drain cells will prevent the initial driver board shutdown with fresh batteries from happening.

I think I keep the flashlight like it is now, I have to sort out the emitter problem but in all other aspects this has become a fine flashlight :party:

3

You are still up late Jos! It's like 3:30AM!

4

You continue to amaze djozz. You have done an amazing amount of testing here on what looks like a promising driver. Thanks.

5

good catch! And I'm up again for an hour already now, slept for four hours, family life is ruthless

And thanks Steve, I'm not completely happy about these tests, so I'm glad that HKJ will do some testing too.

6

Seems like this is first post about the LD2. Could you provide some more pictures of it?

7

Here are the front and back of the 2xli-ion prototype next to the LD1.

I forgot to take pictures of the 1xli-ion prototype version before soldering the #@* out of it, but most of it is the same, a few small components are different (i.e. no components on the battery side).

8

Thanks for all the hard work Djozz!
Your testing will help the designer to perfect the driver which in turn will be a benefit to all of us flashies!

Grtz
Nico

9

Very nice test and pictures.

Thanks

10

Nice work. Your handwriting is as bad as mine lol

11

I really appreciate your commitment Djozz! Thanks for all the hard work. And thanks for explaining the results in words! Because I'm not very good in understanding the results in numbers only!

Thank you

12

Looks like the LD-2 is a more professional looking, more “polished” product! It also looks like it has thicker copper and larger vias than the LD-1. Wow! :heart_eyes:

13

Thank you for the review!! :smiley:

14

I modded a flashlight with the 2xli-ion LD-2 prototype. Building and testing in reply#1

15

nice … good to see that there will be a linear driver for 2s lights

16

There was already the LD-1 from led4power , and the zener modded qlite (mod a qlite yourself or get one from Mountain Electronics).

edit: sorry, the LD-1 was not suitable for 2S, my wrong.

17

Thank you for the review! Nice work! :wink:

18

Led4power asked me to do one more test-series, on a BLF17DD driver. I did not have one on hand, but I could easily make one out of an AK-47 driver, by removing the three 7135 chips and adding a FET (for how it is done, see: https://budgetlightforum.com/t/-/32231 ). I used the same FET as is commonly used for wight's LFPAK56 FET drivers, like the A17DD-S08 driver, and the effective lay-out is the same, and on this converted AK-47 driver there are no traces in the main current path that add extra resistance compared to the A17DD-S08. The specs of this FET tell that the source-drain on-state resistance should be between 3.2 and 4 mOhm, if I read that correctly.

I used this driver as representative of the various direct drive FET-drivers around BLF. Like I did for the LD-2 prototype and the 16x7135 driver I wired it up for driving a XM-L2 T6 4D (old batch, so lowish Vf and high current capable) and measuring input voltage, led current, voltage over driver and led light output.

Results:

(the striked-through numbers were measured with a loose-ish clamp on the plus-entrance of the driver, so I had to redo them)

If you compare this to the LD-2 prototype driver driving the exact same led, you will see that the current that it provides to the led is basically the same except that above 3.72V, the LD-2 starts regulating at 6 amperes, while in this direct driver without regulation the amperage just keeps going up (while not adding much to the light output btw above 6A). The same output current from the LD-2 and this direct driver is also reflected in the parasitic resistance, that is very comparable: at 6A it is 5.6mOhm for the LD-2 and 4.8mOhm for the direct driver.

There are even lower resistance FET's, like the big Vishay FET that started the direct driver development at BLF, so the resistance of a direct driver could be a bit lower even, but we are already talking about resistances that are almost negligable compared to other resistances present in your flashlight, going lower will hardly have influence on the performance anymore.

19

Wow! I don’t have time now to compare your notes from each test, but I’ll take your word that the LD-2 performs about the same as a DD driver up to 6A, where it begins to regulate the current. That’s awesome news! Thanks for doing this testing! :bigsmile:

20

Thanks!. I'm not 100% sure of the performance of my DD-driver btw, I might want to see it another mOhm lower, according to the specs of the FET. I would like to see what the MCU (programmed for 7135 chips) does with the FET. I have inherited an analog oscilloscope, I have to arrange a coupling for it and then look at the driver output if the 100% setting is 100% all the time (I do not know squat about that stuff but I've seen funny things happen in comfychair's threads, I can at least look at it).