Sofirn SP70 Alone $50, PM for AMZ US CODE(LIMITED)

As was mentioned before by others as well, I still can’t imagine that a 70.2 can make a donut hole like that, more likely the led was not focused.

1. I did not have to remove the outer spring. Here's what I did:

  • For the tail side I used my hot air station and tweezers. The hot air, like at 360C, got it softened up enough to remove the inner without effecting the outer. Basically while applying the hot air, I applied steady leverage on pulling up on the inner spring. It came off eventually.
  • For the driver side, I didn't want to risk the hot air, in case SMD's would loosen on the driver component side. I realized it really didn't matter if I damaged the traces under the inner springs, so I bent the inner spring to get it's head out between rings of the outer spring, then worked it further out, bending the inner metal so I could apply enough pressure to pull the inner off the traces. I also used a thin screwdriver type tool to pry the spring off it's soldering to the board trace. There was a little damage to the traces but not much. I've used this technique on many Q8's, but this SP70's inner spring are little more stiff and stubborn, but overall it didn't take very long.

2. Did it for greater current flow and avoiding spring melting, but was hoping it would result in fitting 700 long cells easier

3. Nope - a 20 AWG wire is wayyyy better, much lower resistance than a typical spring. If these were high grade springs (Be/Cu for example), then the wire might not make much difference but probably still better - I don't know the exact #'s, so can only guess.

4. I'm having trouble visualizing what you are saying, but the high resistance of springs is the coils. Connecting the tops or bottoms wont help. If you take a spring, stretch it fully out, it's very long, and usually is coated steel, and since steel is a poor conductor, most of the elctrcial flow is thru the coating which is a thin layer. Under compression, the resistance length is greatly reduced since the coils make contact. This is why long cells seem to do better than shorter cells - I saw and measured this with the 20700's compared to the 26650's.

I got the driver loosened but not removed. My plan was to remove the driver with using the beveled angled iron tip (Hakko BC/C) which can get a lot of heat to the tip, but the wired solder connection to the MCPCB would not budge. I had to wet the tip, and this would soften the original solder but not enough to remove the wire. I gave up for now, since I didn't need to remove the driver to add the bypass to the spring, but when I try again I'll use the bigger size BC/C shaped tip and crank up the iron temp.

I hate the lead free stock solder they use - seems to be high temp stuff, and on this big MCPCB, the heat gets pulled away from the pad quickly. I really don't understand this - some MCPCB are worse than others, like I never have trouble with Noctigons, but L4P MCPCB's are a pain to solder to. Maybe it's the coatings, not sure.

For the FET's, it seems impossible for sourcing the best FET's in China, so the Infineon Lexel (and I) use are probably about the best now, though I like the SIRA20DP as well. We know they' can't source Vishay's,since the SIR800DP's they use all seem to be fakes. Fakes may be over stating it, fakes are almost second sources on the original, like the 7135's are cloned/copied/faked. China makers can't import parts easily - they are penalized for doing that.

These springs should be Be/Cu.

How did you remove the inner spring without the outer? Can it be twisted out? I would just remove both then put the big spring back as it easy with a solder iron.

I don't know how to tell a Be/Cu spring from a regular. I thought I explained it in detail. Basically, yes, the inner spring was twisted out - on the driver side, not the tail inner spring.

The spring pad is as wide as the spring so there is no pad space outside the spring to add solder - I always struggle with an iron on this setup. I'd rather just remove the inner spring if not that difficult.

Tom E has been modding lights even longer than I have, when it comes to tricks and tips he knows as good or better than anyone here.

modding is a matter of wanting things done no matter, you think out ways and it works, or it does not work, or you screw it up.

Me too, never understand why they do this. Hard to replace spring with a better one too cause it only fits specific sizes. Tool AA I modded had this but not even all the way around just 3 slots along the circumference

@TomE, use magnet to check if these are BeCu or not.
Sofirn told me they should be using BeCu.
I don’t know which alloy they used, but it should be at least BeCu C17200 22% IACS.
Not the best, but still adequate.

If the magnet actually sticks, then oh boy.
If not, it’s at least a copper alloy.

Before the bypasses:

GOLISI 26650 4300 mAh @4.20V, one cell solder blob'ed: 14.2A

--> Lumens: 8600/7590 - 7820/6900, throw: 203 kcd (901m)

After the bypasses:

GOLISI 26650 4300 mAh @4.20V, one cell solder blob'ed: 14.65A

--> Lumens: 9620/8490 - 8740/7710, throw: 221 kcd (940m)

Roughly 10% range boost.

Parasitic drain I measured at 1.83 mA. Think that's a bit high, it includes the switch LED being ON.

Throw is surprisingly decent, in spite of a few things that induced misgivings. Sheer output can do that, a longer distance lux measurement will be interesting.

Tom, have you had an opportunity to run it long enough to get a feel for the thermal properties? The design certainly looks adequate, just wondering if it works in reality as it appears it should.

Wow that impressive wonder can it hit 10K.

Ok, makes sense. Thank you for the info!

Quite the procedure! Thank you!

Yea, I agree - does throw pretty decent. My #'s taken at 5m too, and it's a non-treated glass lens, but who knows - it still may perform well I recall the Q8 lens did well in clear, not much of a bump goin to a UCL/p. Didn't run it long, but even after the 30 secs and the throw tests, it feels a bit toasty, but really can't tell til it's really run continuously for a few mins.

Didn't mention this yet, the finning is nice but has some sharp edges. It doesn't appear to be a high qual anodizing because I can see some marks on those sharp edges. I suppose at this price and this size, I wouldn't expect much better. Kind of reminds of the old TrustFire J-18's, AK-47's, etc., but the threads are definitely nicer.

But don't get me wrong, overall I really like this light, lots of power and great UI even stock. I gotta look into changing the LED though to something more NW.

This turned out really good! I think we have to thank Jason for pursuing it with Sofirn, not sure of others were involved, but --

Thanks Jason!

Ok, some more info:

Just ran a 3 minute turbo test, temps taken on the outside:

  • 1 min: 33C
  • 2 mins: 42C
  • 3 mins 49C, and it stepped down, but dbl click brought it back up

The wires were easy to come off from using the bigger iron tip and 750C temps.

Driver pics:

Looks like a fairly simple design, might be equal to a FET+1 because that 3 legged part near the bottom appears to go out to the LED- side, same as the FET does. Looks like we could get a bump with a FET swap, perhaps.

The CTKS part is probably the LDO, but didn't trace it out yet.

The MCPCB is actually 42 mm x 1.5 mm. Back side is pretty smooth, no burrs, good amt of thermal grease, though I'd still prefer MX-4 in there. Shelf looks ok but didn't test it for flatness yet. I still would guess the wires are 18 AWG. I think the FET is the NXP MtnE sells here.

NW tint I will probably purchase!

I took some pictures of SP70 beam last evening. Do you notice the boats of the 2nd picture? It's dragon boats preparing for Duanwu festival celebration.

By the way, SP70 alone and bundle are available on amazon.com now! They are on sale as new release. For a bettery price, please contact sofirn for code!

What is the driver diameter?