Testing boost drivers

Flashlight Modding and DIY Parts
1

Hi, My name is Damian. I would like to introduce myself here providing you nice tests of two LED drivers:

1. Kaidomain H1-A boost driver for XHP50.2 LED

2. Aliexpress boost driver for XHP70.2 LED

Drivers were tested on XHP50.2 LED soldered to 20mm DTP KD board. The board was screwed tightly to massive aluminium heatsink. Two lab PSU’s were used as smaller one cannot provide more than 5A.


Conclusions:

- unexpectedly high efficiency of H1-A driver on two lowest brightness settings is remarkable. It seems it is optimized for low current modes. It will provide extreme system efficiency when used with XHP50.2 or XHP70.2 LEDs. It very likely will fall into the range of 200 lm/W on lowest mode! The driver efficiency quickly sags under higher drive currents and higher boost ratios, though.

- lowest mode with 26 mA will output 27 lm and it will last nearly 90 hours of continuous operation using 3500mAh Li-Ion cells, assuming 200 lm/W system efficiency. Wow.

- any higher mode will fall into lowest mode under 3.1V and warns about low battery voltage with intermittent quick flashes. - low mode drains the battery until it falls below 2.6V where UVLO trips in.

- considerably better efficiency on turbo mode of XHP70.2 driver. Much worse on lowest mode.

- fifth ‘Turbo-boost mode’ with huge 4.5A current is activated using 2-click. Efficiency of this mode is low and power loss is huge.

- this driver will shut-up the light when input current approaches 10A on ‘turbo-boost mode’ what happens at 3.5V input voltage. Assuming internal resistance of the battery and series resistances of typical flashlight parts render this mode useless…

- both drivers hold their constant current drive levels very well, but H1-A driver is just outstanding in this regard! Its modes are nearly perfectly spaced.

I wish you will like it. More drivers tests to come!

2

Woooow. Poland rulez :sunglasses:

3

Measurement data was collected again with more precise measuring equipment for input/output side. Newest Convoy boost driver for XHP50.2 was added. Power supply wires were soldered directly to the boards and springs. Springs were bypassed with copper wire. Input/output voltages were read directly at the inputs/outputs of the drivers.

1. Kaidomain H1-A boost driver for XHP50.2 LED
2. Convoy boost driver for XHP50.2 LED
3. Aliexpress (Mantaray) boost driver for XHP70.2 LED

Individual plots of drivers efficiences, drive currents and power losses on all available levels:

Drivers medium, high and turbo modes efficiency comparison. Drive currents and power losses comparison:

Raw measurement data

Conclusion:

These all are cheap and very solid performing drivers. Robust output and efficiency of XHP70.2 driver on high power levels but lack of efficiency on low mode (also, 80 mA is not so much the lowest mode…) Convoy’s driver is overal top performer and its price is lowest. It can regulte down low nearly into complete battery discharge. It keeps very high efficiency of low-med-high modes in 93-96% area in a broad input voltage range. H1-A driver switch into low mode below 3.0 V and continues until 2.6 V. H1-A is still a king of low-power runtime and can deliver constant-current lowest-level mode for very long time. XHP70.2 driver’s 4.5 A ability is impressive, especially that efficiency is still between 87-91%. Attention! Both Convoy’s and Mantaray’s drivers can suck-out up to 10 Amperes from the battery, especially when it is nearly depleted! H1-A driver is ‘safer’ as it tries to regulate not exceeding 6 A at the input side. All drivers switch-off the LED below 2.6 V protecting battery from further discharge. All drivers lack of really low modes, though…

4

Awesome input, much appreciated. Can you also test Convoy drivers for SST-40. These reach 5-6A I’m really curious how they perform…

5

Here we go, new Convoy SST-40 FET driver tested on TURBO and HIGH modes:

Yes, it indeed does 6 Amps on fresh batteries. Current drops to 5.8-5.7 Amps after prolonged operation due to the heat. Under 3.8 V (directly at the driver’s input) current starts to drop rapidly and the board enters direct-drive mode. Internal resistance when direct-driving is calculated at 22 mOhm. Average efficiency from two sweeps is 95.8% on Turbo and 86.6% on High. It is very important to provide low internal resistance battery and flashlight internals for this driver in order to utilize its full potential… Lower modes are to come.

6

Not sure how I missed these. Very nice testing from a power/efficiency standpoint.

7

Nice post. Too bad you cant get your hands on a MTN-BST3

8

Great reviews, thanks for sharing.

9

Aweome reviews mate thanks for all the hard work!

10

Thanks for the provided information, I'm a fan of fully regulated drivers.

Jensen567 provided related testing on an early layout version of the H1-A plus H2-C and KX70 drivers here: Buck and Boost Drivers, Testing, Modding, and Discussion (Pic Heavy). His H1-A measured efficiency figures were somewhat better. I have an H1-A modified for close to 4A at the output (≈3.95A) with an R100 plus 2x R330s stacked over its stock R025 sense resistor, driving a dedomed XHP50A inside an SK98. The tailcap switch was replaced with an Omten 1288 and its spring bypassed, of course. The pill was also copper plated (you can check some pictures here).

That ∅22mm “Manta Ray” 2S emitter boost driver looks very nice. I'm not in a hurry for a new build, albeit it definitively goes in my wishlist. They have an interesting C8.2 host in that store. For those of you who like aspherics, this 26650 Fishing Flashlight Host may be quite nice (personally untested yet).

Cheers :-)

11

In that measurements is important to have 4 wire measurement setup. I don’t know was setup has been used by Windforce. If you use measurements for input voltage and current from Power Supply that not be very accurate because you have voltage drop across suplly wires and that will perform errors. Same as measurements at load. Another solution is to be used short as possible wires with large AWG for lower resistance.
something interesting about that topic https://e2e.ti.com/blogs\_/b/powerhouse/archive/2018/03/12/measuring-efficiency-the-right-way

12

Thanks for the input. Yes I used 4-wire setup for in/out voltages readouts directly at the points of load using two DMM’s. Input current was set at the PSU and LED current was read with third DMM. I checked whether PSU current and ammeter’s readings were close to each other. All wires were soldered, no alligator clips etc. were used, springs were bypassed with copper wire. Output voltage was measured taking ammeter’s and its leads resistance into an account, it resulted in bare converters efficiencies. Highest modes were measured with short bursts allowing the components to cool down. At the end, thermal effects caused LED’s Vf to drop and drivers internal resistance to rise so it was altering efficiency calculations slightly.

13

Indeed, measurements from my opening post contain some errors. Ignore it and please take a look on 3rd post, where the measurements were properly conducted and efficiences on higher currents are much better. What is more, probably there are parameters spread between individual converter boards so even the same converter type can have different measured efficiency in the end. But yes the difference in our results is high...

Finally I mounted this Mantaray driver with dedomed and diced XHP70.2 in KDLITKER host and now it is my fav flashlight. It throws nearly as far as XHP35 HI variant (on natural background throw is virtually the same) It provides wall-of-light effect with great electrical efficiency at 4500 mA drive current. The lowest 80 mA mode is too bright for indoor use due to crazy efficiency of the XHP70.2 on low drive currents.

14

Left: old H1-A. Right: newer version.

The H1-A driver has been downgraded, lower efficiency reported. The loss may severely affect its ability to be set at its maximum (≈4+A, due to 10A input limit on TPS61088 boost IC when close to 3V at the input).

15

Sad news. I tested ‘2R2’ old version and I still possess several pieces. Thankfully, H1-A isn’t the only single one cheap&good boost driver in the world. ‘Mantaray’ or Convoy’s driver are still very good in certain areas.

16

Yep but H1-A is ∅20mm, it really doesn't plays in the same league as ∅22mm units. I have a 2.2µH inductor recycled from a deceased H1-A plus a few Murata FDA1254-H-4R7M=P3 (11.2A/10.2A max rated inductor). The latter are very beefy and may not fit in some builds. Anyway, this adds more work and expense for a very small purchase price difference. It clearly is a net loss sorry to say.

Is there any small gadget which can measure inductance? I have a couple recycled toroidal inductors (and even Amidon cores LoL), but no idea of their values.

17

Here are the best sale ads for the Manta Ray drivers as far as my consciousness went:

Enjoy.

18

Can these be modified for 2s operation to drive a 6v xhp70.2 with 2x26650’s, like a zenor mod or similar? Thanks for the testing results