7135 based driver on LiFePo battery? (=also IR light build thread, with measurements on Oslon Black 850nm led, mod finished now)

There is almost no energy left in typical Lifepo4 below 2.8V,so you can use 7135 drivers without any modifications.

The polarity protection diode consumes .6 volts. So bypassing it will increase voltage to the MCU and, I guess, the PWM signal going out to the 7135's by that amount.

Removing the 470 ohm resistor is supposed to eliminate the low voltage protection.

I don’t know how much power output you need, but FWIW I ran a cheap chinese IR diode with just a NIMH @ 1.2-1.4V. The Vf was supposedly 1.8 as well, but I tried it for fun and it worked. More info in my post. The IR68.

Removing the diode also raises the low voltage trigger point - the MCU thinks the voltage is higher than what it's programmed for and calibrated for via the resistor.

edit: Wow, the way I put that is about as clear as mud. Low voltage function will not trip until the cell voltage is lower than it would be with the diode in place... er, that may not be much better. I think I need a nap! :Sp

I remember your thread well, I will probably put it in a sk68-clone host too . Well, I don't know either how much power I need because it is just for fun :-) (apart from my daily edc-light I do not need any of my lights, it is all hobby), it is a Osram Oslon black 850nm led 600mW output (90deg. Output angle), it might put out a bit more light than your chinese led (so I will not point it at my son ;-) ) . The aim is still is a suitable 1A driver, but if all else fails I will go for direct drive on AA's, thanks for the suggestion :-)

Cool, good luck with the build djozz. Be sure to show off what you do with it :slight_smile:

hmm, I did some testing on the led on copper tonight, I may at least want to go to 2A :evil:

Here's the underside of the led, the solder pads footprint is smaller than an xpg but I reckoned it would fit a xpg Sinkpad. I just reflowed it:

And yes, it worked (50mA from led-tester), a very faint red glow in reality but my phone camera (even though it will have an IR-filter) picks up a bit more of the IR radiation:

The central heat sink pad of this led is unfortunately not electrically neutral (it is connected to led+) so it will be an extra challenge to fit this board into a flashlight while keeping a well enough heat path from led to the outside.

The test set-up, with the luxmeter clamped-in close to the top of the led to pick up enough radiation to measure anything at this insensitive wavelength. Ledboard on huge chunk of alu, directly connected to power supply.

2.8 A :evil: :

And the data neatly put into a graph. The measurements are quick and dirty, as usual because of lack of time and motivation to do it well, but it all seems to make sense:

Of course the led had less than a minute (about 30 seconds) on the highest currents, but I reckon it should at least survive 2A for a while, even inside a flashlight :-) . The current really runs away with minime voltage rise, so I really want a good current regulated driver now.

Whats your plan for this IR emitter btw?

Member GMARSH made some custom Oslon PC boards for his high-powered TV-B-GONE flashlight… he might be able to supply one… The TV-B-Gone flashlight build

Couple notes, based on my experience with the TV-B-Gone flashlights:

- The IR Oslon’s have two dies in series, so their forward voltage is double a typical IR LED. At 2.8A, the 940nm SFH4725S that I’m using has a ~3.7V forward voltage. This means a single LiFePO4 might not cut it.

- The center thermal pad of the Oslon is connected to the anode, so make sure the sinkpad is electrically isolated from the pill when you build the light. I’d suggest using thermal epoxy (arctic alumina, etc) instead of fujik, so there’s a hard insulating layer between sinkpad/pill. Measure with a DMM to ensure the pill/sinkpad are isolated when you’re done, otherwise you’ll short the cell when you turn the flashlight on.

Thanks for the notes! The IR Oslon that I used is actually a different one, it has a 90 degrees beam angle (very round dome) instead of 150 degrees, and it apparently has its dies in parallel because the Vf is between 1.5 and 1.9V. It is part nr. SFH 4715S , I guess that your led is SFH 4725S. It makes my driver solution different from yours.

I will use arctic alumina adhesive to mount the led board :-)

I should have got the in series version instead of the parallel version, it would have been much easier to drive. But I did not know there was one until now :-(

I see in the thread that texaspyro linked that you put 2800mA through the (series) led. That would compare to 5.6A in my test, quite rough on the led I'd say :evil:

Ha, I didn’t know SFH4715 and SFH4715S were two different parts - S being the two-series-die version. If you’re using the non-S version, you’re fine with a LiFePO4 cell.

Both parts are rated at the same forward currents, interestingly (1A RMS, 5A pulse) - implying you can get twice the light out of the same package with the S version, even though the S version would presumably generate twice as much heat in the process.

Just caught texaspyro’s post - I mount my LEDs on a FR4 board, and I do have a couple of spares I could give up, but if your LED will be powered on continuously I’d say just stick with the sinkpad.

Oh; on a Nanjg 105C driver, remove R2 to disable low battery shutdown.

Instead of isolating the sinkpad how about switching the build to be host positive?

Battery positive to the tail. Isolate the ground ring on the driver from the host, bring driver ground to a battery contact plate, connect driver positive to led.

Did you have a host in mind?

A reference image might help with some of the mod suggestions.

I did not have enough time today for real flashlight modding, but I did test out a nanjg 105C on the Oslon, with succes :-) (after my output test and reading about gmarch's mods, I got greedy and wanted more than a ak47, but still thanks for the image Helios )

I zapped out the R2 resistor, as suggested by gmarch (edit: and ImA4Wheelr) (thanks for that!). Then wired the driver to led and 14500 LiFePo battery, with the current measured at the led wire (the wire colours look funny: I accidentally swapped red and black at the led board):

I used the third star and got 0.13A on low, 0.85A on med, 2.6A on high (close enough for me). The driver gets hot fast, as expected, I need to pay attention to that in the mod.

djozz wrote:

I zapped out the R2 resistor, as suggested by gmarch (thanks for that!).

You're welcome

Oh oh, apparantly I used a wrong word again ;-) , nothing personal.. (I lookedup "zap out" , it is nothing I intended to do to that poor resistor)

It sure is not easy for us non-native english speakers to just find correct words, and I am getting ever more careless about my language nowadays. (and I'm not even mentioning humour, we sound stiff and humourless in english)

There is something wrong with “zap out”? I’m a native english speaker & it seems fine.
Perhaps I’m not cool enough to know.

Google brings up urbandictionary for “zap out” but often people add junk to urbandictionary.

I figured you meant you removed the resister from the driver, which is what I first suggested in Post 13. I was just childishly feeling unrecognized as I did spend a little bit of time looking into the driver to get that information. But now, I see I left a zero off the resistance in my above post. I should have said 4700 ohm resister. The resister you removed from Bay R2 should have been marked "4701" which is 4700 ohms.