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

Epic, just epic.

LinusHofmann, while you are at it and have lionheart_2281`s blank check. Just double your amount and send one light to me as well.

Thanks man, appreciate it! :)

:beer:

Haha, reminds me of this quote, “First rule in government (or blank check) spending: why build one when you can have two at twice the price?” ( Zemeckis’ Contact ) lol

If you’re serious lionheart_2281 , once I have things finalized I’ll have a little think about what would be feasible.
DavidEF there’s just so much work required on every corner of this light that I don’t think a group buy would do much to the overall cost, but I dunno maybe if a few people are interested.

And because it’s always hilariously awesome, here’s the C8 piggybacked onto the BTU again :slight_smile:

…hell just because it’s Christmas… why not a Flashlight on a Flashlight on a Flashlight :bigsmile:

Haha it’s a silly light, my wrist is starting to complain already! :stuck_out_tongue: Even without all the extra passengers the BTU itself now weighs 1.4kg, not exactly the most ergonomic feature for a handle light haha.
Need to get one of these steadycam rigs for it! :slight_smile:


So much win, on so many levels!

:D

Insert Ice Cube meme here.
If you make more of these I’m VERY interested mate, big time.
So much work went into this one, and it’s been a pleasure following you every step of the way, I’m sure the next one you’d have done in a quarter of the time.

Alrighty, first test results are in.

For now it’s just a full output test versus heatsink temperature to see at what point I should be programming the turbo mode to ramp down.
It’s interesting to see, over the initial 30s there’s the expected turn-on heat sag. Then it’s relatively stable for the next 1.5mins as the hefty mass of the light absorbs the heat. Finally things start gettting a bit too hot overall and the output starts taking a nose dive. This is also the point; right about the 2minute mark, with a heatsink temperature of 50degrees, where the hot potato would be hitting the deck :wink:

The light was already slightly warm at 30degs before beginning the test, and ambient temperature was ~22degs C. I would expect better performance outside obviously.
I’ll set turbo timeout to a maximum of 255 (255*0.5/60 = 2.125mins so that’s just about bang on)

What I can’t determine precisely from this test is whether the drop in output after 2 mins is as a result of thermal perfomance drop at the leds or if the 7135s are throttling back because of heat. I’m struggling to get a sturdy enough setup going for measuring current during these tests. Ran out of deans connectors to make a harness so that I don’t drop any extra voltage while measuring.

Interesting to see.

I get the impression that output seems to be dropping a lot quicker compared to what the heat sag is responsible for after the 2 minute mark.

Yeah I think you’re right, believe it’s most likely the 7135s ramping down because of heat. But it’s a gradual ramp down rather than the silly flickering and stuttering I was seeing after only a minute with the first iteration driver, so I’m pretty ok with that. In any case I know where to set the turbo step down now :slight_smile:

I was also taking note of pack voltage during the test (not on the graph) and that only dropped gradually and linearly from 8.25v (under load) at the start to 8.08v by the end so there should still have been enough voltage overhead to maintain max regulation current. Hard to be sure of that with the light fully assembled though, I could be getting some extra resistance loses that I’m not calculating for.
Should know more when I get a harness made up to splice my ammeter into the situation correctly.

Leaving the light to cool a little (just a minute or so, it’s still >50degs on the outside) and turning it back on returns the output to near the maximum.
Seems to me that’s another indication that the biggest decline in output is related to driver heat.

I cant find where you said you current draw was. In djozz's graph I can confirm my measurements were close to this. Post 16.

If you take away a conservative 10% in losses could it still be a voltage issue? Can you stick the probes on a single led without the reflector installed without blinding yourself?

This was and is a very enjoyable read. Job extremely well done. Looking forward to beam shots!

Well done! Bravo!

Thanks guys :slight_smile:

Max drive current should be 5.84A x 3 = 17.52Amps

I was also using djozz’s great graph and it seemed to measure out correctly for me when I verified voltage across the emitters at 6Amps.
I was using ~7v as a minimum target voltage to feed to the drivers while still maintaining regulation (think I verified 6.9v at the leds @ 6Amps). So I have faith that the basic calculations are near enough correct.

I know I drop around 0.6v by the time the cable plugs into the back of the light including connector losses and I was using a 0.2v drop across the 7135s. What I didn’t get round to measuring was how much I drop across my twisty contact interface going into the driver contact board. That’s the last place where I would imagine larger losses are to be found, unfortunately I didn’t have access to the resistor array dummy load and 1000w psu that I used to test for resistance losses within the rest of the system.

So, starting with an 8.25v battery voltage (under full load) I should have a vF overhead of 8.25 - 0.6 - 0.2 - 6.9vF = 0.55v


Here’s the same graph with pack voltage included. It’s still well above 8V at the end when output has collapsed down to 50klux. And again, turning the light off for a bit then on again the pack voltage will return to that same level but output will be up near 70klux again. Can’t explain why that would happen if the pack voltage was inherently too low to maintain that level of output right?

Urgh I dunno, I’d be rather surprised if the voltage was limiting output this drastically but I may still have some extra resistance losses eating into my overhead.

I’ll need to get some accurate current measurements to figure out what is going on. I did do a test with my crappy harness but that was dropping 0.36v before it even left the battery pack and ammeter so that’s a bad test setup. Max drive current I saw in that test was still 16.7A though and it held it up for 1.5minutes fairly steady before again starting to drop off more quickly.


At the end of the day it’s a bit of a catch 22 with these linear drivers. Especially with a battery pack like this which doesn’t really sag.

If I improve resistance losses and get a higher voltage to the drivers

  • Theoretical longer time in regulation
  • More energy for the 7135s to burn off
  • Possibly getting 7135 overheat, throttling and flicker earlier negating the extra regulation time. :stuck_out_tongue:

Well I’ve been having fun with graphs again.

To check whether my thoughts in my last post where correct I did another test. This time eliminating the battery pack and cable entirely, plugging the two packs across a reinforced harness and directly into the light. I also rigged up my hefty power meter to measure current.

Here’s my test setup

And the balance board reinforcements I needed to make to reduce resistance losses to a minimum

This way (by eliminating the resistance loses across the spiral cable and relays in the pack) I am verifiably gaining 0.6v in voltage overhead compared to the fully assembled light. That should eliminate any concerns that the voltage is dropping too low and causing the sag in output/regulation.

The test was carried out in two parts.

1. I ran the light on high from cold taking Battery Pack voltage, Battery Current, Lux, and Heatsink Temperature readings every 30 seconds.
2. After 6mins I turned the light off and let it cool down for around 5 mins.
3. Then I ran the light again for another 3 minutes taking all the measurements again.

So here’s the graph.

It’s quite interesting, and I think it verifies quite nicely my thoughts that driver heat is the primary cause of the drastic output drop overall, but especially after about the 2minute mark.
In fact with the higher voltage reaching the driver this time, I see a less stable regulation phase initially that drops more quickly and more steadily than with the fully assembled light. Last time between 1:00 and 1:30 I saw a flat spot in output (indicates decent regulation), this time it’s all downhill from the start.
After the 2 minute mark again output takes a nose dive. And this time it took only 4 minutes before the output had collapsed to 50klux while on the fully assembled light it took a minute longer to drop to the same level.

The voltage gradually drops as expected during the test but the crucial bit regarding this aspect of the situation happens after the cooldown period.
At the 6minute mark battery voltage under load was 8.08v, during the cool-off period with the light off it recovers up to 8.15v however crucially upon turning the light back on for the next part of the test it is bang on again at 8.08v.

As a result the output and current readings taken right after the light is turned back on can be compared directly to the ones taken before the cooloff period. With battery voltage being bang on identical between these two points the only variable to blame for the massive difference in output/current draw has to be heat. Aha!

You can also see that with the lower battery voltage at this point, the driver actually has more of a chance to try and regulate current and there is even an increase in driver current and output a minute after the restart. Unfortunately it doesn’t last long before the sharp drop once again kicks in.

Finally, having the current readings and seeing them track pretty neatly with the output measurements rules out thermal deterioration at the leds being the primary cause of output sag.

My conclusion is that the driver assembly is choking on it’s own heat and can’t shift it away quick enough, as a result it drops output prematurely.
It’s probably just keeping up while the temperature of the pill is below 50degress, but once it’s environment has reached that temperature and beyond it’s a losing battle and things start getting really toasty quickly.

I will try to do something to maximize thermal contact between the aluminium driver assembly and the walls of the pill. Maybe by stuffing some aluminium foil in there are and applying a bit of thermal grease. In any case I’ll redo this test to see if it has made any noticeable difference at all, suspect it won’t do all that much but it’s worth a try.
There’s just too much heat involved here, liquid nitrogen might be the only solution :slight_smile:

Next one of these will have a copper slug sinking those 7135s! Or a bunch of ventilation holes and a blower fan keeping things in check :bigsmile:

If anyone has any thoughts on this dataset it’d be appreciated, hopefully I’m interpreting the data correctly and not barking up the wrong tree here.
Cheers

Nice testing LH. I wonder how a small fan blowing on it would go to simulate the light being taken for a walk?

Not a bad idea, how about two? :wink:

Does BTU actually stand for “Bi-Turbo Unit”?? :bigsmile:

This light just keeps looking better and better!
I know it’s just a concept, but hot damn that’s awesome

i was pretty sure heat generated from amc was the problem. they have some sort of self protecting thing regarding temperature . when u push hard them and they cant radiate enough heat, u could have than all sorts of exotic problems in output. in one of my projects one stack with amc start to periodically change output for example. apart from this little problem everything looks cool on your project and i am enjoining the investigative part of your project with graph and so, even more.
seeing all this new projects and cool works that is being done around here, is starting to awake my sleeping appetite for projects.

usually its hard to realize how complex similar projects can be and how much work and effort they require. thank you for your effort and for sharing this.

make it liquid cooled version :wink:

I do like the potential or these new integrated water cooling solutions don’t see a way of making it work here though. Maybe someone else can build/modify a light around something like that, be cool to see.

i always ” shoooh! ” my thoughts away when a liquid cooled Light comes into my mind.

(un)fortunately i don´t have access to a lathe and a mill…