BTU Shocker Triple MT-G2 with a twist -- Aiming for >100Watt ~9000Lumens -- With external 2S power pack, handle etc...

I looked into it a bit more and running the pill separately from the driver I noted that a few of the 7135s were getting too hot and started throttling back within a minute.
This was running the driver direct from a single pack without the long cable so the vF difference that the chips need to burn off was exaggerated (I drop ~0.6v across the cable at 18A so max vBatt the 7135s should ever see would be around 7.8v or rather 7.6v with sag) but it resulted in the same issues I was noting before.

Mostly the chips on the bottom of the slave boards without a direct heatsink link where hitting the limit rather quickly but all of the chips were inadequately heatsinked (-sunk? -sank? -sunked?). I underestimated how much heat these chips produce and expected a single solid solder point from the top ground ring to also sink the bottom of the driver adequately. Should have known really.

So anyway I spent the evening soldering copper tabs between the ground rings of the three slave boards top and bottom and reinforcing the connections to the copper posts. I’ll probably also pot the bottom of the boards to the aluminium assembly with fujik once I’m happy things are working well enough to be more permanently mounted.

Urgh it’s a mess tbh, don’t look too closely. It’s more flux residue than driver at this stage… :wink:
Was tempted to just start the thing from scratch and use those new Nanj 105d drivers instead. They have a great layout for this type of thing and much more generous grounding ring, not to mention I’m pretty sure the copper is thicker on those than on these crappy KD drivers…might still end up going that way if I can’t get this to work well…

It’s not a solution to the mental heat the Mt-g2s put out, actually they’ll just be running at high power for longer now…doh …but at least the driver doesn’t start going mental before it’s own heatsink has even had a chance to heat up to match the rest of the pill. After that it’s just a matter of tweaking the turbo timeout for sensible operation.

This is why I focus on making things easy to disassemble, I certainly don’t trust myself to get things right first time even with all the testing I do along the way. Generally it’s time well spent. :stuck_out_tongue:

Are you telling us Kim Kardashian succeeded?

j/k, the light looks great

Not unless she was the one working in our garden and snagging the cable with a shovel. :stuck_out_tongue:

Well think I’ve hit a bit of a speed bump here. Even with the thermal path reinforcements some of the 7135s are still getting too hot. Honestly wasn’t expecting this to be an issue, never really had any overheating issues with these chips before as long as some basic heatsink considerations where undertaken.

Especially the stack of chips on the outside of the puck, where I had to file down the solder blob and ground ring for clearance issues, are getting very toasty. Hard to get extra copper connections on there too.

Will probably have to rebuild this thing from scratch to maximize the thermal paths… :frowning:
Last thing I’ll try before going that route though is potting the whole thing in fujik…but urgh that’s really a last resort since I wouldn’t be too happy with such a messy solution…

Welcome to the MTG-2 on heat club.

Haha, let me get the hose!

Ok, let’s see if this will work. Hacked up Nanjg 105d with extra copper thermal pathways.

Basically I used one of those new layout Nanjgs and instead of stacking an extra layer top and bottom I triple stacked the top because that allowed me to solder a copper ground band all around to maximize heatsinking. I noticed before that the 7135 chips on top of the stack and only connected to the heatsink through a thin solder bridge where the ones that couldn’t dump the heat quick enough. This should no longer be the case, all 7135 ground tabs (at least on top of the board) are now solidly soldered to the copper band which in turn is soldered down to the ground ring. Finally I will solder two copper “ears” to the band which will be bolted and possibly also soldered to the copper posts on the assembly to complete the heatsink/grounding path.

I guess I just like pain… haha…but I’m not going to jump in and make three of these things before I’m absolutely sure they’ll work! :stuck_out_tongue:

Time to test this thing and see if it can dump the heat fast enough! Fingers crossed! :nerd_face:
Cheers

+1 !

Nice looking solution. I hope it works as good as it looks.

Fingers crossed...

Thanks, looks like it works.

I just ran the new slave in free air with no additional heatsinking and it didn’t throttle back at all even after 2mins @ ~6A! Yay! Pack voltage was around 7.6v in this test which should represent the kind of voltage the slave would have to deal with in the light. Obviously the whole thing got quite hot (over 60degs C ) because the heat had nowhere to go, but no drop in power during that duration so all the chips seem to be getting evenly hot, awesome. Same conditions on the old slaves would see instability and step down after about a minute.

With a fresh pack at 8.4v the slave stepped down after about 20s! Haha lots of energy being burned off there but that’s obvious (I think that’s around 6W of heat being dumped into the poor little slave)! :open_mouth:

Ok feeling optimistic that this will work well in the light, let’s get to building the other two :slight_smile:

Urgh, 2 down 1 to go. :exmark:

Nearly two hours of squinting through solder fumes to build the second one, is it all worth it?
I certainly hope so…Hah

Probably time to learn how to draw up custom pcbs :stuck_out_tongue:

Will to live…gone… :stuck_out_tongue:

But these 7135s are very well sinked now and soldering thermal pathways to the outsides is a piece of cake so getting the heat away from the slaves quickly will be much more easier than before.

Looks nice! You’ve got the same amount of .380 chips per emitter as I have in my M6 build. How noticeable was the throttling back before your latest changes? A sudden drop or graduate?

I didn’t notice anything on mine during the hot potato test, but maybe I wasn’t looking close enough (not all that comfortable holding on to a light with this amount of juice for over a minute… yet). I’m not running stacked chips though, and there is a decent bit of space between my driver and the LED shelf.

Have you done the hot potato test?

Well tbh I don’t think you’d see the initial gradual throttling behavior unless your measuring current. I believe the 7135s ramp down output as they approach their ciritical temperature range so it’s hard to tell by eye if the temperatures are at the threshold. If the chips get very hot (120deg C iirc) it seems they’ll try to shut off completely and you’ll notice some flickering and larger instability in the output.

In my experience seeing as you’re running your light off 18650 cells, the voltage sag associated with that power source and the high current draw is probably enough to bring down the amount of energy that needs to be burned off by the 7135s pretty quickly after turn on. You may well be dropping out of regulation well before the 7135s start having a hard time.

I have a fairly hot Apex 5t6 running 20Amps and I don’t think the 7135s are ever taxed too hard even with realtively poor heatsinking, simply because the batteries can’t maintain a high enough voltage for very long.

It’s really a matter of looking at battery specs, taking note of their voltage sag and the vF of the Mt-g2 at that current and hooking up an Ammeter to check your current draw during the test.
If you notice your current dropping out of regulation well before you expect your cells to drop below the vF required to maintain that drive level, then it’s likely the 7135s are getting too hot.

Hot potato isn’t too relevant for me because of the handle, as long as I can keep the 7135s below critical and stop the Mt-g2s from desoldering themselves I should be alright :wink:
But tbh I’ll need to really tune in the Turbo timeout to avoid causing damage, this thing gets HOT!

Yeah, I know you have the handle, but would be interesting to know how long it took without using it :slight_smile:

Will do when I have the light assembled again. :slight_smile:

Boom! Driver puck V2.0


Two copper tabs soldered to each slave provide a solid ground and heatsink connection to the rest of the assembly.

Very pleased with how this came together, took a while but it’s a much much better design than the first attempt. :bigsmile:
Now back to assembling this thing properly!

I like it! J)

I love it!

:smiley:

Thanks guys :slight_smile: