I present you with Giggles, the most hardcore BLF thrower in the World.

Ok, my decision to slice the dome off was kind of made for me. I was trying different focus points and at some point something got on the LED and started the dome burning.

So I went ahead and sliced the dome off to get rid of the burn. Not my best slice by any means, my Soldering iron died today (yay, now I have to find a replacement that will not cost a ton and get here this year) so I could not remove the mcpcb from the light and it was hard to get the razor in the right position.

Still even with a less then perfect slice the results are way different then I expected. I thought the days of 50% gains from de-doming were behind us?

Turns out they are not.

With the dome on I got 450 Kcd for 1337m of throw :wink: :wink: :wink:

With the sliced dome I got a stinking 700 Kcd for 1680m of throw!

So the GT now matches the best throwers on the market with 2x the lumens and an XHP70.2 LED!

Now for the results.

I am not sure what the lumen loss was as I can’t test the GT all put together but you can figure there is some loss, still over 6k if I had to guess.

The darkspot in the hotspot has been virtually eliminated now, only if you really look close will you notice it.

The only con to this is the tint shift, the tint shift around the corona grew significantly in both size and harshness with the slice.

Not unbearable but noticeable now compared to before when you could ignore it. White wall hunting is not it’s strong point like this.

Now it is possible some refocusing could improve this, it did with the domed emitter. Or using a cooler LED for the slice, a 5000 or even 5700k could work better but I don’t remember seeing any 5000k high bin XHP70.2’s in my searching.

I will play with the focus later and see what it does.

Wow, I am still amazed at the results, I didn’t even expect this much out of it. Goes to show what kind of performance you can expect when modders get ahold of this thing.

It worked yesterday, but it’s an unusual url. It does not end with a file extension like .jgp, .png, etc… Anyway, I posted a link to the post DEL made that he put the picture in. It seems to be working. If not, I’ll host it myself. :slight_smile:

Edit: Oh snap! The url changed. Maybe his host Dropbox changes the url every now and then.

Dropbox is funny like that. You can still ‘steal’ the link by quoting the post and copying the link text:

Note that the picture is a rev00 driver (and the only built sample is with me). Production units have rev04, not much functional difference and identical BOM though.

6 A is about as high as the stock components can take. Bottlenecks would be the inductor, FET and that sense resistor. At 6 A the FET and Rsense see about 1.4 W of dissipation each. Inductor is good for 6.5 A on paper, 5.5-6 A in that application.

For Rsense, the LED current is approximately:

I = (0.25/Rsense) - 0.15

Rsense is 0.091 ohm stock for ~2.5 A. Just stack a 1+W 0.1 ohm on top of it to get ~5 A. Garden-variety 1206s are typically only good for 1/4 W, so not up to the task.

Yeah, the IR gun was showing those components getting up around ~105-110c external temps at 6A, so figure higher at the junction. Still within reason but I would not use it in turbo for long periods like that.

The nice thing is that the ramp stops at 4.8A so you can use that most of the time and just use turbo, for well, turbo.

Nice, it’s working again. I also added a link to the full size picture if anyone wants more detail.

Del, since your here I wanted to ask something.

Can you briefly explain in layman’s terms how this driver works?

It has a FET, but it’s not a direct drive light. It seems to maintain a set maximum current (2.5A) as long as the voltage stays above 13.5v? This is called “in regulation”, right? Then below that voltage “out of regulation” the max amperage starts to drop. Correct?

Sorry for sounding like a newb. I missed all the early development of the GT.

Its a proper ‘buck’ driver.

The FET is pulsing the inductor and the inductor, by nature, outputs a DC current proportional to the duty cycle of the FET. (Not to be confused by the PWM output of the tiny85.) The inductor is the big black cube on the board, and is basically copper wire coiled around a ferrite core.

The brains of the buck operation is that small chip sitting next to the sense resistor. The tiny85 simply generates a 0-1.24 V ‘set-point’ for the buck chip. The buck chip compares this set-point with what it measures across the sense resistor and then does its thing to put out the requested amps.

Some things to note:

  • There is some PWM happening in the driver, but for most output levels the LED current is almost pure DC
  • To get to the very low modes, it does output PWM to LED (modes below 10%, or 0.25 A). This is under the control of the tiny85 when we go PWM and when we go DC, so there is flexibility here for people doing their own firmware.
  • Battery current is not same as LED current. Like for a transformer, the ratio depends on input vs. output voltage: Iled = Ibat * (Vbat/Vled) * Driver efficiency
  • Driver efficiency is around 92%, depending again on voltages
  • The FET is a p-type, not our typical n-type. It is being pulsed at up to ~600 kHz, so it also needs to be much faster than our typical ‘direct-drive’ FETs.

110C, ouch , that is getting up there. I guess the driver was not screwed down at the time? At 80 W output the driver has about 7 W of heat to shed. Screwed down into the head with the retaining ring will make all the difference.

Highest I personally saw was 105C but I am not sure how much I trust my harbor freight IR sensor.

Correct, it was turned sideways in the driver cavity so I could look at the components. Properly screwed down the backside of the PCB did stay cooler but impossible to tell what the components were doing.

I think you were spot on with 5A being a good max current. For those that are willing to take some risks, 6A appears to work ok.

Here is a copy paste from the GT thread about the XHP70.2 death.

After seeing how well the GT took to the XHP70.2 with a sliced dome (over 1600m of throw!) I am sure that there are BIG gains to be had with a flat black or other LED in the GT.

That said I finally got some proper centering rings printed up for the XHP70.2 and did some testing. By defocusing it slightly I was able to eliminate the dark spot! Only lost a little throw as well, it was still over 1600m IIRC. The beam looked better as well.

I took these quick wall shots during my G35 review with the sliced dome and a 3.5mm tall centering ring (vs the factory 2.5mm, I think that 3mm tall could be the sweet spot for a sliced dome XHP70.2).

Closer to what the eye sees (although much brighter, literally painful to look at the reflection):

With the exposure turned down to see the beam a bit better:

Sadly shortly after this I was messing with it some more and something must have landed on the die and caused it to start burning. While trying to fix it I scraped off the Phosphor layer so I now have a mega powered UV light saber (it is super painful to look at, I am scared to turn it on).

So I will be switching to the 80CRI LED next and most likely leaving the dome on it.

Once I get the V1 prototype back I might try another shot as a sliced dome, maybe even an xhp50.2. Or maybe I will try a flat black….

Ok, so today I stole a few minutes to install the XHP70.2 80CRI 4000k in the V2 prototype. HM440H model for anyone interested.

I LOVE THIS LED!

This LED looks SO much better then the 4000k 70CRI that was in it before. It is a lower bin and it lost some output but not as much as I would have thought. Still easily as bright as an XHP70 except without as much of a dark spot and with 80cri.

I tried a few different sized centering rings. Throw readings showed a peak of ~440kcd (vs ~450kcd for the 70cri) with a 2mm centering ring.

Although I am going to print up a new centering ring that is ~4mm tall since manual focusing shows that around there throw only drops to ~430kcd yet the dark spot virtually disappears. WELL worth the trade off.

So what do I end up with?

A ~6000 lumen (possibly more since the throw readings are so close to the 70CRI @ 7000 lumens) 80CRI 4000k 1300m+ throw GT with corona tint shift hardly worse then the stock XHP35.

This light will be keeping this setup for sure, it is fantastic! I would never have guessed that the 80cri version would have this kind of improvement in the tint shift issues. I still don’t know why but it has very little tint shift (not perfect but only slightly worse then the XHP35 that comes stock once focused).

I expect that slicing the dome would have similar results to the 70CRI version with ~1600m+ of throw but I am not willing to risk another $10 LED to find out right now, maybe if I get board or feel lucky later.

Or if someone wants to donate an LED to the cause I can put it on the sphere to test the output and try slicing the dome to see those results as well.

^

The hell is this?!

spam…

“It is designed to provide the highest level of detail”

Re-post: How to deal with trolls :wink:

Hey TA, since you’ve got your production light now, do you think the tint is still rosy? It does seem rosy to me, but everyone perceives things differently.

Is it still putting out 2600 lumen?

It is the same LED so it should be exactly the same as the test LED +/- the binning tolerances (aka, 7% either direction)

The reflector is improved so the cutoff seems sharper, which is nice.

Tint is fantastic, I love it more each time I see it. It has a hint of rose but mostly just looks like a small sun.

It should be around ~2500-2600 lumens at the emitter. People have to remember that getting a reading in the light is very hard, if the entire reflector is not in the sphere, the reading will be much lower as quite a bit of the beam is along the outside edge. Even a ceiling bounce can be skewed as they are not usually setup for mega throwers like this.

After the reflector and lens losses I would expect the OTF lumens to be around ~2200 if I had to guess.

I’m really happy with this tint as well. I’m glad you settled on it. I’ve been comparing my other modded throwers with my new GT. The thing that stands out the most for me is the horrible tint they have because they are all dedomed. Everything seems monotone and flat compared to the GT. Tint was a sacrifice I made for maximum throw. But with giggles I get the farthest throw and a great tint to boot. You guys really hit this one out of the park.

Thanks TA.

Glad to hear that, a lot worried about 4000k being too warm but I had a feeling most would like it once they saw it in person. Those that dont, are welcome to change it naturally.

Indeed, there was no way this light was going to use a de-domed emitter for just that reason. You can swap one in and get close to 2mcd by some estimates but it is not worth the loss in tint for me.

Although I have been working with DEL to put together a parts list that would allow driving a 3V LED like the Flat black with up to ~5A by swapping a few components on the stock driver.

So the 91mOhmF shown on this driver is .091Ohm resistance?

Looks good, mine showed 2.45A at the emitter tested through an 18GA loop at the MCPCB with a UniT clamp meter.

Edit: Looks like the large MOSFET is a different value, if one were going to the 12V XHP-70.2 would a change here be beneficial?

I have been discussing upgraded components with DEL. I ordered them today actually. We will post them once tested.

Although in stock form the driver is good for 5A at 12V (aka, same as 10A at 6v). I have my xhp70.2 setup for 6A right now (.040 ohm 2512 resistor) and it is working good so far but the driver runs hot.

The FET that is on it is hard to beat, DEL did find something that is possibly a bit better when combined with some other changes though. The goal is to allow ~6A reliably with 8A bursts. More then this would need an FET driver.