12A 7135 Based SRK Replacement Driver

He put his 7135's too close to the edge, will use ATtiny84 and have a few other high end features. The cheapies will complain his price is too high, but they probably won't buy your driver either. I'd probably build a light with all three. It may be more expensive than I'm willing to try, but I wonder how well your driver with the zener mod using 18350's in 2S4P driving a trio of MT-G2's would work.

Be VERY careful putting holes/leads in that center area of the board. The cells in an SRK are held very loosely and their position can be all over the place. With flat/low top cells it is easy for them to touch things in that center area… kerblewie! Same goes for the outer ring… it needs to be fairly narrow to make sure a cell cannot touch it. Some cells have awfully wide positive contacts…

Also, be very careful depending upon the board solder mask to keep the tab under the AMC7135 chips from shorting to etches underneath the chip. Solder mask should NEVER be used to provide insulation in critical places. Plus, being able to get solder to that tab under the chip provides a much better thermal path.

Finally, I would not recommend paralleling all the AMC chips and the LEDs. At these currents imbalances in the LEDs can lead to run-away current hogging and emitter toastification. My driver breaks up the AMC chips into three groups, one for each LED, 3 amps max to each LED. Yes, you could parallel the banks (like Old Lumens did for hist MTG mod), but I would not recommend it fot things like XML’s.

These issues led to the funky layout of my driver board with 24 AMC chips in three banks.

Thanks texaspyro. I made specific mention of the board being untested for these reasons. The BATT+ contact ring is larger than on the stock SRK driver board so battery wobble should be less of an issue. Nontheless the through hole vias were simply to make soldering wires easier and to reduce the risk of ripping solder pads off. I would obviously encourage people to keep the wires as low as possible.

Points noted on the routing under AMCs etc. I know this is frowned upon, but the point of this board was more to do with big power than anything else. I have identified a way I could potentially improve the design a little bit, but I will wait a while to see if the demand really requires it.

Live and learn :)

The Qlite has routing under the chips, in the dozens (100?) that I’ve used, it’s never been an issue and so far I’ve run em up to 4 high and over 6A. Lucky maybe, don’t know.

Great stuff. I’ll keep on eye on this and see how it all pans out. I already have some SRK drivers on the way so I don’t need them just yet.


Every cheap 7135 board I’ve seen has the output trace crossing under the chips. It doesn’t mean it’s good design but does seem to work. Breaking up the output into groups could be done just by cutting the output trace in a few places and soldering an led- wire to one of the chips in each isolated group. For a high voltage Zener mod it would be better maybe to have the entire bottom of the chip available for ground trace soldering but maybe not as important for normal 4.2V usage. The natives have been getting restless waiting for an SRK board and this is a sort of stop gap measure. I like that Mattaus is willing to put things out there knowing that his first effort can be improved on and that Texaspyro takes the time to make suggestions. We all learn from them and gain by their efforts.

Sorry RBD. I failed on most of that last line. :_(

Okay, so maybe “all learn” is a bit much but we still get new designs we can use whether we understand them or not.

Comfychair wrote:

I'm pretty sure there will soon be a different version using a single direct-drive FET,in case there are others who are a bit tired of the 7135s.

(this one works great, Vishay SUD70N02-03P)

Very much looking forward to that. Ordered some and I'm off to your thread to figure out how to wire them into this board.
On TexasPyro's point above about contact pads for cells and the battery tube. If there is a problem with this design, I imagine one could reflow some copper sheet cut to the appropriate size over the positive contact ring.

I guess part of the accidental 'charm' of this (or any 7135 based driver) is the relative ease of which the boards can be 'hacked' up. Seriously - I've seen some of the boards people have created, even some I've made, and they look like they've been put through a meat grinder. And yet they still function. Boards with chips stacked 4 or 5 high (sometimes more). Traces ripped up with jumper wires running left and right. Extra components tacked on. AMC chips being driven above and beyond their factory rating. Master/slave setups. The list goes on.

All this board will really ever be is a NANJG driver sized to natively fit a very popular light. If you're concerned about a large group of AMC chips running LEDs parallel then just go to town on the board. Leave a few chips off and stack them else where if you need room. Swap out the MCU for another. Zener mod it. Hell; you could just use it as a simple contact board for an all together completely different driver for all it matters lol.

Use it how you see fit and do what people around here do best - make up for its shortcomings by ghetto engineering the crap out of it.

- Matt

The problem is not with the contact ring, it is with those pads/holes for wires at the center of the board. The positive contact for the cells can touch them… even if there are no wires protruding. The center part of the driver board really needs to be empty. If I remember correctly, anything within around 0.2” of the board’s center is a risk for shorting with some cells.

So add a copper ring.

Extra bonus, it'll work with unmodified flat top cells, no solder blobs needed.

I think the point is that reflowing a disc on the pos contact ring will shim the B+ contacts and prevent contact with anything in the middle area.

And worse comes to worse… maybe just paint a bit of liquid electrical tape over the inside of the copper + battery ring to protect those thru holes

You read me right Rufusbduck. Thanks for interpreting and clarifying.

Subscribing :)

It's worth a shot if you already have some, but it wasn't tough enough to work with my reflector and probably wouldn't here either. Fujik should work though.

So, I could make 5 or 6 zones and have each one go to one led like this?

Yellow lines is where to cut the traces and numbered areas is where to expose the traces for leads.

It could work for the 5 and 6 led SRK lights. Just stack chips and go.


Once tested, I just might try some.


Looks right to me.

WOW..was that a nod from Old-Lumens? :D