Confused by results - Astrolux S41 vs P60 quad (update)

So I just got my first quad built up to put in a P60. I’m using a copper pill from kiriba-ru, with a DTP sinkPAD soldered to it. LEDs are XP-L HI, 2pc 5A2 (4000K) and 2pc 3C (5000K). The driver is only temporary at the moment, a simple 8x7135 QLITE. I think I might put a Regulated 6A LD3 driver in as the final version.

When I fired it up, I was impressed with how cool the pill stayed at 2.8A

I took it to the ceiling bounce room and measured 96lux with not a single lux decay for about 2 minutes (as long as I was willing to sit there just holding a light). I was again impressed with the lack of drop-off since P60s aren’t known for their great thermal properties.

So on a whim, I fired up my modded (dual spring bypass, pulls about 15A) Astrolux S41 N219b version. At turn on for Turbo - an impressive 144lux, but wait… IMMEDIATELY the lux start dropping, quickly - within 10 seconds the lux had dropped from 144 to 72! Yes, only 10 seconds after turn-on, the S41 was making 25% less lumens that the P60. After making 50% more only 10seconds earlier at startup.

I guess I assumed the exposed copper pill on the S41 was really good for heat dissipation. Was I that wrong? I obviously knew the light got hot quickly, but I had no clue it was affecting output that much.

Do these results sound right to you?

Edit: Updated this post to correct the Amp draw from the S41 to 15A (I had incorrectly said 12A). Further updates in Post #13 . I have not yet had a chance to open up the head of the S41.

So I measured the tailcap draw of the S41 last night and I was incorrect when I previously said 12A. It actually pulls 15A on a fresh-charged-but-rested 30Q.
I re-ran the baseline (light in its current state, modded) test last night and got slightly better results, maybe the previous night I still had a little residual heat inside the head even though the exterior was totally cool. The (ceiling bounce) light output/time was:
Instant-on: 142lux
1second: 126lux
3seconds: 96lux
10seconds: 82lux (getting painful to hold near head)
30seconds: 68lux (head too hot to hold)

I think I’m simply overheating/overdriving the light. As djozz mentioned, his testing shows 4A to be the max for 219B, but that’s with good heatsinking (I assume djozz has the LED&MCPCB either soldered, clamped, or screwed to a good size chunk of copper for his torture tests). The little copper S41 pill just isn’t enough to keep up with 4 LEDs driven at 3.75A each. To confirm this suspicion, I put in an old laptop pull cell that was only capable of about 6A. Here are the results:
Instant-on: 96lux
1second: 90 lux
3seconds: 82lux
10seconds: 76lux
30seconds: 72lux

As expected, the maximum output at 6A (1.5A each LED) is less than at 15A (3.75A each). And also as expected, the fall-off in output is more gradual.
But interestingly (at least, to me) – the light output at 30seconds is HIGHER with the old laptop pull at 6A, compared to the healthy 30Q at 15A. Just too much heat at 15A.

My conclusion at this stage is that 15A is just way too much current/heat for this light in this state. My next steps will be to first solder the MCPCB to the pill and then see how it tolerates 15A. If it is still overheating very badly, I will remove the tail spring bypass and re-test, and then if necessary the driver spring bypass. I like a hot-rod light, but not if it is only bright for 3 seconds before overheating.

Is your S41 running an 18650 or 18350? May actually be battery sag, need to use a very good battery in that one. Also, is your P60 foil wrapped or somehow fitted tightly in the head of the light? If not it will not transfer heat well at all. It is likly that even a solid pill will need to be fitted. Also, the P60 is seeing less that 1 amp across each emitter, so it is not going to get that hot.

Matt

The S41 dropping by 50% in 10s seems like more than I would expect, but I am not familiar with how the 219b performs. What is the battery and does the driver step the output down? Like a timed stepdown?

With my 18650 direct drive triples I observe the output drop by about 10% in 30s. Some of that comes from the voltage of the cell dropping as it discharged, so the capacity of the cell should have an effect.

You need to check if there any thermal grease under mcpcb of your s41.

That does not sound right at all. My s41 doesn’t sag that much and that fast even with a 18350.

+1

Also make sure the surface is nice and flat so the mcpcb is making good contact. I had to file down a burr from one of the tapped holes that was preventing the mcpcb from sitting flat in one of my s41’s

The battery was a charged-and-rested 30Q,so I don’t think voltage sag should be an issue.

I will check for the condition (thermal paste, flatness) of the MCPCB tonight.

219b’s sag quite a lot, don’t they? Their vF isn’t very favorable. My 219b S41S showed the same behavior and I wasn’t terribly surprised. I was okay with it though, because I had plans all along to gut it and rebuild anyway. Woulda just bought an empty host if they’d offered it.

I did not realize that. Is this less of an issue with single emitter lights?

Well, maybe it’s a good thing I’ve got a bunch of 219C in the pipeline via Clemence’s groupbuy!

The 219B V1, on a copper DTP-board and well cooled, I measured to have its maximum output at 4A, so a quad at 12A, 3A per led, seems about perfect to me. A 50% drop in 10 seconds sounds like heatsinking problems.

XP-L HI is more efficient and runs cooler than 219B.

The s41 copper head is mostly hollow. Looks great tho.

Do you have a beamshot of the mixed XP-L HI? I’m interested in the same combo.

I have soldered several MCPCB directly in the S41 head, it helps. Too hot to hold in about a minute is what you should be getting if the thermals are good.

Update: Editted OP to correct the Amp draw from the S41 to 15A (I had incorrectly said 12A). I have not yet had a chance to open up the head of the S41.

So I measured the tailcap draw of the S41 last night and I was incorrect when I previously said 12A. It actually pulls 15A on a fresh-charged-but-rested 30Q.
I re-ran the baseline (light in its current state, modded) test last night and got slightly better results, maybe the previous night I still had a little residual heat inside the head even though the exterior was totally cool. The (ceiling bounce) light output/time was:
Instant-on: 142lux
1second: 126lux
3seconds: 96lux
10seconds: 82lux (getting painful to hold near head)
30seconds: 68lux (head too hot to hold)

I think I’m simply overheating/overdriving the light. As djozz mentioned, his testing shows 4A to be the max for 219B, but that’s with good heatsinking (I assume djozz has the LED&MCPCB either soldered, clamped, or screwed to a good size chunk of copper for his torture tests). The little copper S41 pill just isn’t enough to keep up with 4 LEDs driven at 3.75A each. To confirm this suspicion, I put in an old laptop pull cell that was only capable of about 6A. Here are the results:
Instant-on: 96lux
1second: 90 lux
3seconds: 82lux
10seconds: 76lux
30seconds: 72lux

As expected, the maximum output at 6A (1.5A each LED) is less than at 15A (3.75A each). And also as expected, the fall-off in output is more gradual.
But interestingly (at least, to me) – the light output at 30seconds is HIGHER with the old laptop pull at 6A, compared to the healthy 30Q at 15A. Just too much heat at 15A.

My conclusion at this stage is that 15A is just way too much current/heat for this light in this state. My next steps will be to first solder the MCPCB to the pill and then see how it tolerates 15A. If it is still overheating very badly, I will remove the tail spring bypass and re-test, and then if necessary the driver spring bypass. I like a hot-rod light, but not if it is only bright for 3 seconds before overheating.

Edness, I do not have a beamshot at the moment. I do have this pic:

With the 10621 Carclo quad spot optic, the diferent tints were visible in the beam when white wall hunting - not sure it would be noticiable in use. But I also didn’t care for the beam pattern anyway.

With the 10623 frosted medium spot optic, the beam is much more what I wanted (very smooth) and you cannot see any tint variations in the beam pattern. I was expecting the frosted optic to suffer slightly lower output in my ceiling bounce, but I only measured 1lux difference, which could be measurement variation so I consider the output to be basically the same.

:+1:

My S41S pulls 12A stock and 14A with the driver spring bypassed with VTC6. It gets hot after 15secs.

Nailed it. It’s tempting and fun to see how much current you can dump through LEDs and have them survive but sooner or later cooler heads will prevail. :wink: In situations like this you also have to consider the size of the heat sink relative to the amount of heat generated. DTP copper makes a truly amazing difference at the primary thermal interface but you can’t afford to ignore the next stages.
It should be interesting to compare the results of singles/triples/quads vs current to see how have multiple LEDs on one star and sink affect the results.
It’s a judgement call whether the extra lumens up front are worth the loss of output and run time at the back end but that’s always been the case. DD drivers are just the latest way of demonstrating the quandry, in spades.

IMO a non bypassed S41 with XP-L HI is the perfect setup for this host. It’s more efficient and runs much cooler. I have a second S41S and that’s what I plan on doing to it. I also have an E14 with 219C D320 that draws 18.5A and it’s a instant hand warmer. I use that only in short bursts, never on for more than 10 sec. The E14 also has a bypassed forward clicky so it’s perfect for self defense.