Luminus SBT-90 gen2 5700K tested (SBT-90-WDS-F72-SA600 SA-H4, from Neal)

Looks like a single 30T then maybe max's at about 23 amps, but with multi-cell setups like in the GT90, could go higher I'd suspect. I've been modding GT carriers to 4P and getting good results in a converted GT->GT90. I should figure out a way to measure amps in one of these.

Weren’t there K75 measurements somewhere that showed turbo at like 28A?

edit

28A might be beyond the peak, maybe not. Dunno what kind of driver that is in the K75, or if the measured #'s were stock or not. Guess multiple batts helps a lot.

Did VOB measure the GT90 amps?

I think the cells are in series in the k75 and its a serious buck driver. Because of the effort it would take to design that driver, I’d be shocked if acebeam overdid the current.

The Vf was only 3.5v at that current level though, so there must be some series resistance we are missing in multi 21700 configurations. Parallel 30Ts are going to hold wayy more than 3.5V at 28A (14ea). Even a single 30T should overwhelm this emitter in a 0 ohm circuit.

But then again, it only takes 15mOhm @20A to drop 0.3V

That thread is sketchy - missing pics, missing details, so not sure what he's really measuring, or how for that matter.

djozz could not get that high in amps to test the SBT90.2, but even at 20 amps, it was starting to flatten out - I'd guess roughly 25 amps would be flat in his test. So maybe there's a little more life at 28A when the LED is cold, dunno. I got a few pretty serious low resistance SBT90.2 builds and I'm peaking at 23 amps on a 30T at best, so maybe doubling up 18 AWG wires, and maybe improving over a SIRA20DP FET, dunno, but seems like ther'es not much there for single cells.

I’m going crazy these days not being able to fit the light hobby into my budget. The only thing I bought in the last 6m months was a skilhunt e3a. But if I get a chance to buy the 4x18a 90.2 version this year that’ll be my Christmas gift to myself. I have a 5V 100A power supply I can test this led with. If anyone has an emitter they would send me I can set up perform a test, but with this emitter I doubt that’ll happen. I don’t yet have a calibrated lumen sphere/tube, but can easily get relative numbers.

Contact resistance, bypassing traces, and parallel FETs are the heavy hitters (i know you know this). I wonder how much can be gained by increasing contact area with the cells and higher pressure. What is RTSon for that FET? 20cm of 18awg is about 4mOhm. So 2mOhm is about all you could save there in practicality.

Wondering bout parallel FET's - thought that was a bust? I thought maybe TA tried it and got no advantage? I even thought it could do worse, maybe a timing issue? This is very fuzzy with me, can't recall where/when/who. The HK90 has places for 3 in parallel, populated 2, but I replaced the 2 with one SIRA20DP. Think they had older Infinions in there.

Don't assume too much with me - yeah, some things I know, but not everything for sure

Issues with parallel FETs can arise at low duty cycle and high frequency AFAIK. But at WOT in turbo, parallel FETs can only reduce resistance. But with the FET you are using has an RDSon of about 0.8 milliohms. This is 5x less than your wires.

So between the wires and FET you have about 5mohm. That accounts for about a third of what I’m spitballing to be your total. Don’t know how much the traces on the driver, mcpcb, and tailboard are contributing. Bypassed blue springs are about as low as you’ll practically get. That about it. Short of bypassing the driver and getting a comparative lumen value idk what to suggest that isn’t radical.

The battery itself is a factor as well, I would think. Sounds like I should have tripled up the SIRA20DP in the HK90. The K1 SBT90.2 actually has 3 springs because of the USB-C charging - I bypassed all 3 but dunno what the path is through the charging circuit.

I gotta check my other SBT90.2 lights, specially the 2 zoomies 1504 and B158. Dunno but maybe the switches are a high resistance source, but taking an amp reading with the tailcap off.

OMG, I am really thinking about the copper FT03 w/SBT90.2 - that's a fairly big light, big chunk of copper - https://budgetlightforum.com/t/-/31589/5418

Yes, the cell contributes a whole lot, but here I’m starting with the voltage under load of a 30T based on HKJs data and reported Vf values from djozz and other’s. It’s seems safe to me to estimate a 300mV drop is occurring at 20A elsewhere in the circuit. The FET doesn’t seem like a meaningful contributor according to the spec sheet.

Djozz and MRsDNF conducted switch resistance tests years ago and found between 1 and 10 mOhm values across different models and current levels. switchtesting , (june13th2015:additional test by MRsDNF in post#44)

If you can, check output of mechanical switch lights with your bypass and compare with tailcap installed. These could show a large difference at times. Those that do though don’t usually last that long.

Ok - looks like I didn't recall the tests I did correctly.

SBT90.2 mods from notes:

Brinyte B158: 25 amps on a 25S

UF 1504: 24.5 amps on a GOLISI 26650

Just measured:

K1: 24 amps on a T50

K1: 23.6 amps on a 40T @4.21V

The new Vapcell T50 5000 mAh 21700 is for real, at or close to a 40T.

T50 sounds great. VTC6 of the 21700 world.

And those amp draws look more in line with what I’d expect. 5-10 milliohm is typical for a good FET light.

I really think it might be a Samsung 50T, which I don’t know if it actually exists, but we historically were slow to learn of these new cells. One thing I’m hoping for is the ‘s’ line of Samsung 18650s (or even an ‘r’) to make it to the 30 series. I think by now there is room to improve power density and maintain 3Ah

Hhmm, 50T sure bout fits it's performance.

But how fast does the T50 drop off compared to a 40T

I'm not really set up for doing long term testing. I did get in another pair though of the T50's

I don’t understand how to read the chart. For example, to achieve 4000 Lumens the lamp will consume 59.4W of power, right

I know this ^ was a while ago, but I want to put one in a light that came with an xhp50, what did or do you use for “centering ring”, or more specifically so the reflector doesn’t short out in the mcpcb? I have been thinking about cutting a thin piece of plastic just so the reflector has something to sit on, would that work?

In what light? I had a few lights that had the Gen1’s 90/70’s in them that I salvaged from older lights/builds… In the FT03 I didn’t use a centering ring… I used Kapton Tape to cover the mcpcb from shorts…I was using the MTG2 26mm boards until Hank came out with the 32mm….

His tests were not done in a lamp/flashlight. They were done in a sphere measuring the raw led with a power supply.

In a battery powered flashlight with it’s added losses (let’s use a conservative 20% loss) you would need 5000 total lumen to get 4000 out the front of the lamp. That would be about 18 amps. Then you need to consider the voltage sag. A decent 21700 might sag to about 3.8 or so. 3.8 x 18 = 68.4. So I would estimate about 70 watt is what a flashlight would consume.

If your talking about a raw led powered from a power supply, you can see that 4000 lumen draws 12 amps at 3.15 volt. Thats only about 38 watts. These not very practical numbers for real world use.

Personally I’d estimate a 30% loss and say you need to push it to 20 amps and you might be lucky to get 4000 lumen out of a flashlight.

These are all my own estimates based on djozz’s test results, as I haven’t ever used or measured this led. Maybe someone who has tested it in a flashlight can give some more detail.