Sofirn SP33 V3 thermal regulation issue

Review from 1lumen

Very disappointing to see how the regulation constantly goes up and down.

I then did my own test and this V3 can only achieve 10 minutes at 950 lumens before it drops to 506 lumens, as soon as temp drops below 45°C it starts increasing with an immediate drop. The V2 could go 2hr 15m at 920 lumens without an issue.

This 3V XHP50.2 must produce more heat than the 6/12V?

I got

2 min at Turbo - 3400 lumens
4 min at High - 1500
10 min ramped to 950

Owen, who wrote the review, must of had a bad unit to get so low output. My stock gets the above at 30 sec without a spring bypass.

Still a good light that should sell imo. Put regulation in perspective to its price and the fact that it stays quite cool. Btw., I also measured around 3500 lm without bypassed springs.

Only the CW XHP50.2 is an imposition. Whoever likes that will have no problem with everything else, I’d assume :smiling_imp:

I’m still convinced it was a bad spring/contact. The thermal regulation itself seems standard across most/all units, but the definitively lower output seems to look directly at that.

I hadn’t checked before, but I just ran a resistance check across the spring to the “flat” part of the tailcap (where the tube connects) - bypassed, I get 0.1 ohms (expected), but unbypassed, I initially got 2.2 ohms - but now I’ve bumped it somehow and I’m getting 0.2 ohms. Just ran a test and got ~2700lm (driver spring still bypassed)… So maybe the spring wasn’t in the “notch” enough. I’m going to clean out the notch, put it back together unbypassed including removing the driver bypass, and run another test.

If I end up getting considerably higher results, I’ll update it, and with a change to “if you can, remove the spring and clean it all out before testing”.

(Doing my best to be 100% transparent)

Removed driver bypass, cleaned spring up, re-settled, 2550-2560 lumens. Removed tailcap and put a literal bar of copper across tube to cell, 2810.

So there’s still a lot of resistance in the spring/tailcap, even after cleaning and smoothing to give the best contact I can.

I know the primary issue is the thermal rollercoaster, but there’s an itch in my brain that can’t let go of this output issue.

Yeah, that’s it. I wondered why I measured so little resistance (even under 0.2 ohms), but decided not to bypass it due to the figures.

Okay, I’m completely done with this, the itch in my brain has faded. 16AWG bypass on both driver and tailcap, and I get 3071 lumens with the stock Sofirn cell.

Yes, the thermal regulation definitely needs some attention. I observed the same behavior as oweban.

From my testing

Note: I mentioned in my review that I measured 3800 lumens. Afterwards, I recalibrated my lumen tube and got considerably lower than that. As it turns out, it was treating throwy and floody lights differently, so no matter how I calibrated it, it was never quite right. I’m currently making a new setup that should be less susceptible to beam pattern. That aside, the regulation pattern still holds true.

At 10Amps I would guess 3100-3200 lm based on tests of 6V LED but that 10A is measured with tail bypassed. With such large springs they should just switch to dual springs so they dont have to worry about bypass reliability.

What was the final result with a 40T?

I’m really looking forward to Toy Keepers version 2 of Anduril, sounds like what I’ve wanted since it released. With the simple UI and optional Advanced UI, thermal regulation it’ll be best the option out there

Will have to test that tonight if I can fit it in :frowning:

Finally made time (had surgery on Thursday, so a bit doped up) - 3201 lm with 40T, 16AWG bypasses.

Thanks mate.

Hopefully nothing too serious

Just some vascular stuff that needed fixing :slight_smile: All good, nice scar, and mending nicely.

Just a quick measurement on high. Bad, really bad thermal “regulation”. And you can clearly see that there’s no real output regulation, no current sensing, just slowly fading output with decreasing battery voltage. Don’t know how this got into production.

I just started using that app. Very handy for the models I stock with no reviews. Saved me from getting a $500 nzd light meter that has data logging straight to CSV format for graphs.

Check the app against a known flashlight whenever you try it on a new device. One of my newer phones has a light sensor that only has relative values or something and I didn’t notice before I got rid of my old device. The new sensor doesn’t give enough precision to be of any use or compare to the old device.

I’m writing up a build document at the moment, but it’s basically a Raspberry Pi with Ubuntu on it, and an adafruit sensor (TSL2591), along with bmengineer’s RuTiTe. Using it for lumen checks, as well as runtimes.

Adafruit VEML7700 Lux Sensor - I2C Light Sensor [STEMMA QT / Qwiic] : ID 4162 : $4.95 : Adafruit Industries, Unique & fun DIY electronics and kits - this looks good too, and cheaper.

Oh damn! Thanks, will try and pick one of those up and give it a shot.