Giving a Mag85 new life

[I came to the light...]

A while ago I gave my father a 9AA Mag85 as a gift. His reaction: ... . Where'd the light go?

The runtime just isn't long enough, and the batteries (Duracell 2650mAh) self discharge too quickly. Now I'm going the rectify that problem, but I'm not sure exactly which way is best. I was hoping I could just buy 3 D li-ions, but it seems like those are very hard to get at a high enough quality for this high-current mod.

In case you're wondering, no, I can't just swap in LSD AAs, because the runtime is already too low.

Alternatively, I could get an LED dropin, but then I'd have to scrap everything I've done so far. Which would be OK, but I can't find anything close to as bright as a WA1185 that will use regular D cells (4.5v total). If I were to use a higher voltage, then I could save a bit of money and keep the WA1185. The only configuration that I can see working is 9xAA eneloops and a P7 or MC-E dropin, but the only fitting dropin I've found is Malkoff's, which by itself is ~2x the price of the entire original mod.

So I guess the second solution I can see would be to find a cheaper quad-die dropin, at any voltage, and swap the duracells for eneloops if I can't get a 4.5v dropin. Any help there?

Also, do you see any other way I could keep the light at around 1000 lumens, increase the runtime, and decrease the self discharge rate, all while keeping the cost under $100? (keep in mind I'd sell unused parts)

Thanks for reading (and sorry it was a bit long)

 

[I came to the light...]

just had a thought - would 3x18650 li-ions work? The concerns I have are that it would be too long, or have too short of a runtime. I could build something to hold them centered.

 

[Alan B]

just had a thought - would 3x18650 li-ions work? The concerns I have are that it would be too long, or have too short of a runtime. I could build something to hold them centered.

3x 18650 would probably work, if it doesn't insta-flash. They are longer so the spring would have to be replaced or reworked. I have not done that but others have mentioned this combo.

If you put in a dummy cell and drop to 8xAA cells the runtime will increase a bit. The brightness is a little lower but not much. The bulb will last a lot longer as well.

You could change bulbs and drop to a slightly lower output bulb to get longer runtimes.

You could put in a PWM dimmer or regulator with the bulb. This allows selecting lower levels and getting much longer runtimes.

3 D alkaline cells will drive a P7 LED very nicely. Almost as bright and much longer runtime. Combined with a d2flex board then mulitple levels are available and much much longer runtime available.

You could put a P7 LED and a HipCC driver in and use the AAs you have, or change to LSD AAs (or Li-Ion 18650). This will regulate the P7 at up to 2.8A and give max output from this LED. The runtime will increase over the filament bulb. Also, a d2flex can be added to give multiple levels etc and really long runtimes.

 

[Gunner12]

You could do a Seoul P7 or Cree MC-E directly driven off 3 D alkaline batteries. The battery voltage should sag enough for it to be a safe for the LED.

You can also use a resistor or a driver to ensure that the current to the LED is acceptable.

 

[I came to the light...]

Thanks for all the suggestions - there are a lot there that could work.

My concern with direct driving the LED with alkalines is the very sharp discharge curve. As for the HipCC driver, it sounds great (a 2.8a driver in general), but I wonder, how much runtime will this get off of 9xAA eneloop?

 

[Alan B]

Thanks for all the suggestions - there are a lot there that could work.

My concern with direct driving the LED with alkalines is the very sharp discharge curve. As for the HipCC driver, it sounds great (a 2.8a driver in general), but I wonder, how much runtime will this get off of 9xAA eneloop?

Since the voltage is triple, the current will be about 1/3. So if you use 2.8A the current from the AA NiMH will be about 1A and they will last about 2 hours.

Remember when the current drops to half the eye sees only a slight change, and it occurs slowly, so the user will not see a big drop off until the current is lower than 1/4. This makes the batteries last quite a while, much longer than a regulated light. Incandescents amplify the dropoff of the batteries, LEDs hide it. If you use 3D cells they'll last a very long time.

 

[Benson]

3 D alkaline cells will drive a P7 LED very nicely. Almost as bright and much longer runtime. Combined with a d2flex board then mulitple levels are available and much much longer runtime available.

Have you tried a 3xAlkaline DD? I haven't, but I have run my P7 4D Mag on alkalines, and results are not impressive. I use a 2.8A AMC7135 regulator. and while it stays in regulation for 190 minutes on NiMH, it drops out after only five minutes on 4 alkalines. I expect you'd get seriously reduced brightness, even from a low Vf emitter, on 3xAlkaline, and the rolloff from there would be just as ugly as the post-regulation curve on mine. Obviously it would be somewhat improved with the d2flex, if you ran lower levels most of the time, but I can't see a Mag85 replacement being used that way.

 

[Alan B]

Have you tried a 3xAlkaline DD? I haven't, but I have run my P7 4D Mag on alkalines, and results are not impressive. I use a 2.8A AMC7135 regulator. and while it stays in regulation for 190 minutes on NiMH, it drops out after only five minutes on 4 alkalines. I expect you'd get seriously reduced brightness, even from a low Vf emitter, on 3xAlkaline, and the rolloff from there would be just as ugly as the post-regulation curve on mine. Obviously it would be somewhat improved with the d2flex, if you ran lower levels most of the time, but I can't see a Mag85 replacement being used that way.

Excellent questions and nice graphs. I don't have graphs of a 3xDD P7, but someone may have done that. You are correct, the NiMH is better for full P7 current than the Alkaline.

Since the eye's response is logarithmic, and a direct drive is not regulated, it appears somewhat different to the eyes. With DD the current drops off starting earlier and falls somewhat slower making the cells last longer and the eyes don't see the gradual dropoff for quite a long time. Plus you can always use NiMH for nearly regulated performance. If the original poster wants regulated performance (which the M@g85 didn't have) then the 9xAA solution with HipCC is a better choice, and with Eneloops it will run about 100 minutes (about five discharge cycles of the M@g85 setup).

If we replot the graph with log output scale to match the eye, and lose the regulator, it will look a little better. The AMC7135 forces the cells to run at full current all the time (turns the extra power into heat) and kills them more quickly than a DD setup would. DD trades a little less output for longer runtime. The original poster was complaining that the runtime of the M@g85, which is about 20 minutes, was too short. Note that even at 50 minutes on your graph the light falloff is nearly imperceptible to the eye. Even at 150 minutes the drop is less than half (and the M@g85 would have consumed 7 charge cycles to get there). The eye will not perceive this as a huge drop, and a DD solution will fall off a bit slower since it is using energy at a lower rate.

Using 9xAA Eneloops with HipCC and D2Flex would probably be my choice for the OP's situation. About $40 for heatsink and P7, $45 for electronics. For on-off simplicity and no levels, drop the D2Flex and use just the HipCC.

I find that when I have levels that most of the time I can use much less than full output. With the P7 and D2Flex I can use level 2 or 3 most of the time. Level 2 is about 200mA and level 3 is about 500mA. At these levels even alkaline D's will last a very long time. Especially compared to 20 minutes. Level four is about 1 amp and level 5 is a bit over two amps (direct drive).

One interesting comparison might be my 1D 3xAA P7 DD with D2Flex. With Eneloops it will last about 50 minutes on full output. This is in a 1D compared to the M@g85's 20 minutes in 3D. Of course on lower levels it will last much longer (and a 1D Li-Ion will run it for over two hours).

 

[Benson]

Excellent questions and nice graphs. I don't have graphs of a 3xDD P7, but someone may have done that. You are correct, the NiMH is better for full P7 current than the Alkaline.

Since the eye's response is logarithmic, and a direct drive is not regulated, it appears somewhat different to the eyes. With DD the current drops off starting earlier and falls somewhat slower making the cells last longer and the eyes don't see the gradual dropoff for quite a long time. Plus you can always use NiMH for nearly regulated performance. If the original poster wants regulated performance (which the M@g85 didn't have) then the 9xAA solution with HipCC is a better choice, and with Eneloops it will run about 100 minutes (about five discharge cycles of the M@g85 setup).

If we replot the graph with log output scale to match the eye, and lose the regulator, it will look a little better.

Just for kicks, I replotted ln(lux) from the same data. See below...

The AMC7135 forces the cells to run at full current all the time (turns the extra power into heat) and kills them more quickly than a DD setup would. DD trades a little less output for longer runtime.

I'm not sure you understand the AMC7135, or maybe we're just miscommunicating. The AMC7135's a low-dropout linear regulator; it can't pull any more current than the emitter would in DD (well, actually something like 2mA more, IIRC) -- what it does is drop excess voltage to keep the current at 2.8A (well, at 0.35A -- I have 8 in parallel for 2.8A total); it has a dropout voltage of <0.2V, IIRC, so when the battery voltage drops to within Vf+0.2V, the current starts falling off like a direct-drive (but with a fixed 0.2V overhead). When this switchover takes place on the NiMH, the cells have dropped to about 0.95V/cell, so they're really pretty empty. The alkalines, on the other hand, drop below 0.95V within 5 minutes of 2.8A draw, but have plenty of life left -- maybe my Duracells are just really bad, as this did seem worse than I'd expected. I just chalked it up as "Wow, alkies are trash!", and carried on, though.

A direct-drive off the same 4 NiMH cells would pull way more than 2.8A, and burn up the LED in short order. A direct-drive off the same 4 alkalines would pull a little more current the whole time because the 0.2V overhead is eliminated, but would pretty much follow the same curve.

The original poster was complaining that the runtime of the M@g85, which is about 20 minutes, was too short. Note that even at 50 minutes on your graph the light falloff is nearly imperceptible to the eye. Even at 150 minutes the drop is less than half (and the M@g85 would have consumed 7 charge cycles to get there). The eye will not perceive this as a huge drop, and a DD solution will fall off a bit slower since it is using energy at a lower rate.

Yes, I understand that an alkaline DD is not really all that bad, and does give usable light for longer -- but a DD off four cells would look pretty much like the green curve, except brighter in the first 5 minutes (no current limiting), and a little brighter across the board (no 0.2V across the AMC7135s in drop-out). I'm just saying, given that a 4-cell DD isn't bad, but also isn't substantially overdriven, a 3-cell DD would be seriously underdriven. (Unless my alkalines are just atypically bad at handling currents >2A...)


I'm pretty much in agreement with you for the best solution -- a high-voltage pack and a buck driver is the way to go. (I'd personally go for 3 Li-ions (25500, 26xx0, or D) instead of NiMH AAs, being more invested that way, but the implementation is the same; just which battery pack you choose to stuff in it.)

Oh, and for incan, there's always the IMR26700, as found in Milwaukee V18s -- three of these fit handily in a 3D Mag; I'm in the process of building a Mag85 on that setup (actually it's all built with a bi-pin/PR adapter, and works great, but I don't have the metal reflector or high-temp socket yet, so it's very-short-run only). Runtime should be in the neighborhood of 45 minutes, and negligible self discharge. I really like my P7 better, though, even though it's only something like half the output. The lower heat and huge runtime really do make it a more useful flashlight.

 

[GreyShark]

just had a thought - would 3x18650 li-ions work? The concerns I have are that it would be too long, or have too short of a runtime. I could build something to hold them centered.

I have a Mag85 that runs on 3x 18650 in a 4C host. It does not instaflash the 1185 even fresh off the charger and I get 38 minutes of runtime.

 

[I came to the light...]

Thanks for all the great responses, again!

I just remembered another issue to fix - recharging. Is there an easy and cheap way to put the socket from this light in, so I could use my FM85's charger? Otherwise, where could I find an 11v charger with leads for this?

Could you please also provide a link to some 25-26mm diameter li-ion batteries? I can't find them. So it seems like I could just buy 3 of these and adjust the spring and be done with it. But now you've got me thinking a P7 is better...

The graphs given (huge thanks to benson) were for 4 NiMh Ds, not 3. Would 1 less battery just decrease runtime, or will the HipCC put it into DD? Also, I take it the d2flex only allows max 2.1A, even with the hipCC giving it a steady 2.8A? Speaking of which, how is the d2flex operated?

So how does this sound: 3xli-ion with d2flex and no regulator. If I understand correctly, the li-ions will give semi-regulated output with a P7 as they do with a WA1185, so max output won't be too shabby in terms of output, regulation, or runtime. And the d2flex will give several more modes allowing for longer runtimes & lower output, which would also work with alkalines because of the low current draw.

 

[Bullzeyebill]

You could install AW's soft start switch with 3 levels of output, and get more runtime running at lower levels. You still can bring it up to full power when needed. This way you keep the Mag85 without converting to LED, or getting another light.

Bill

 

[Benson]

I just remembered another issue to fix - recharging. Is there an easy and cheap way to put the socket from this light in, so I could use my FM85's charger? Otherwise, where could I find an 11v charger with leads for this?

My tendency is to leave the cells loose and charge them separately or in parallel. A series charger with balancing is also good. A series charger without balancing taps like that should really only be used with protected and/or safe chemistry cells. (Although if you're careful, and check it periodically for balance, you can get away with unprotected cells, the user needs to understand what's involved here.)

Could you please also provide a link to some 25-26mm diameter li-ion batteries? I can't find them. So it seems like I could just buy 3 of these and adjust the spring and be done with it. But now you've got me thinking a P7 is better...

Yeah, you can do that, although I'd say a P7 is better. You can always change the battery now, and change to a P7 later. I apologize for the long-windedness, but there are several options here, with various implications for P7 and 1185 configs, so....

I have an 1185 hotwire running on 25500s from DX (sku.5035), which test at about 3Ah (not the rated 5Ah), and are the size of a C cell. You could actually fit 4 of these in a 3D Mag (for a different bulb, or with a hipCC/P7), but they are unprotected LiCo cells -- I wouldn't recommend these for giving to your Dad, unless you know he's willing to learn about the safety issues and treat them appropriately. You also don't want to use these in conjunction with a series charger, as indicated above, unless you're 100% certain that he will take them out and check balance frequently (every 5 charges or so) -- if they get out-of-balance, there's nothing to stop one of them from overcharging. I charge them in parallel to avoid this...

Safer choices are the safe chemistry cells, commonly used in power-tool packs -- A123 cells (and some others) are 26650 LiFePO4, and e-moli cells (and some others) are 26700 LiMn. (But the 65mm and 70mm lengths are somewhat nominal, I guess; I've heard that A123 cells actually measure 70mm, too...) I'd definitely recommend these if you go for the series charger. The LiMn cells can be charged and used the same as LiCo (the "normal" Li-ion chemistry), but the LiFePO4 have lower voltages, both operating and charging, so you'd need a special charger. The LiMn could be used to run the Mag85, where the LiFePO4 would only run it at ~9.6V (yeah, only at its nominal rating.:mecry -- the lower voltage costs you runtime on the P7/hipCC (the output's the same either way), but this is partly compensated by the LiFePO4's greater current capacity -- the LiMn still typically come out a bit ahead in total energy.

For LiMn, I got my e-moli cells from a Milwaukee V18 pack (contains 5 of them) which I disassembled myself. You can get GD 26650 LiMns from voltmanbatteries (on ebay or here). I think there's some places online selling batteries scavenged from packs, but don't know of any right off.

For LiFePO4 cells, again you can scavenge battery packs that use them, and Battery Space has a decent selection, including 26650, 32600 (D), and 32900 (F) -- note that although the D or F look like "best use of flashlight volume" in a 3D, the 26650 is cheaper, lighter, and higher capacity, but has a lower (but still plenty for flashlights) current rating.

For more info on the different Li-ion types, see LuxLuthor's excellent threads: chemistry comparison (includes some info about power-tool brand <-> cell brand correlations) and safe chemistry shoot-out -- it's for 18650, not 26xx0, so all the capacities are lower, but I think the comparative results should be similar for the same brands of 26xx0.

The graphs given (huge thanks to benson) were for 4 NiMh Ds, not 3. Would 1 less battery just decrease runtime, or will the HipCC put it into DD? Also, I take it the d2flex only allows max 2.1A, even with the hipCC giving it a steady 2.8A? Speaking of which, how is the d2flex operated?

That was 4 NiMH Ds, for about 4.8V working -- a hipCC operates from 4.5 to 25V input, so it would be fine. 3xNiMH D would be 3.6V, so the hipCC would be no good (I'm not sure whether it would DD, or just not work at all, but there's no use either way.) 3xAlkaline D is nominally 4.5V, but drops under load, so it would very quickly drop out -- again not sure how the hipCC behaves out of regulation, but don't bother. 3xLi-ion, OTOH, would be fine -- 11V or so, definitely in range.

So how does this sound: 3xli-ion with d2flex and no regulator. If I understand correctly, the li-ions will give semi-regulated output with a P7 as they do with a WA1185, so max output won't be too shabby in terms of output, regulation, or runtime. And the d2flex will give several more modes allowing for longer runtimes & lower output, which would also work with alkalines because of the low current draw.

3 Li-ion makes 11V -- you can't run that into a P7 without a driver like the hipCC. The WA1185 would be fine that way, though. (If the d2flex doesn't get too hot...)

3 NiMH or 1 Li-ion on a P7 is decent DD, so they'd go alright with just a d2flex. I really dislike DD in general, but this is one of the DD setups that usually do work out decently.

 

[ElectronGuru]

I haven't (yet) tried it personally, but word is that given the same voltage, the WA 1331 is nearly as bright as the WA 1185, with dramatically more battery life.

 

[Benson]

I haven't (yet) tried it personally, but word is that given the same voltage, the WA 1331 is nearly as bright as the WA 1185, with dramatically more battery life.

From LuxLuthor's testing, at 11.2V the 1331 draws 63% of the current for 66% output. Given the linear perception of runtime, the logarithmic perception of brightness, and the smaller spot (the 1331 has a smaller filament), 50% more runtime for 2/3 the ioutput and similar intensity probably seems like "nearly as bright" and "dramatically more battery life", but then the P7 (a good bin also making that 2/3 brightness, and triple the runtime) seems even better, except for upfront costs. And it only takes a few $10 bulbs to break even -- you really need LEDs with your rechargeables to make your lumens "guilt-free"!

 

[I came to the light...]

Benson, you are a huge help. I think I've got it now, so I'll just run through what I'm thinking to make sure I've got it right.

I can get 3 of these batteries and put them in the mag85. I'll have to do something about the slightly off length and width, but otherwise I'm set for runtime and self discharge rates.

If my Dad wants to upgrade even further, to a P7, I'd also buy a hipCC and possibly a d2flex, if I want multiple levels. I would use the same batteries, but have to rework the head.

The benefit of the P7 would be 4/5 the output for ~2.5 hours, up from ~1 hour with the WA1185, if I just used the hipCC. If I add the d2flex, max gets lower, but I also get more modes, the lower of which will also work with alkalines.

As for the charger, I could wire the cells in parallel and charge them on my Ultrafire WF-139. Because of the wiring it is unnecessary to check for unbalanced cells, correct?

Again, thanks for all your help.

 

[Alan B]

...

The benefit of the P7 would be 4/5 the output for ~2.5 hours, up from ~1 hour with the WA1185, if I just used the hipCC. If I add the d2flex, max gets lower, but I also get more modes, the lower of which will also work with alkalines.

...

Again, thanks for all your help.

Adding d2flex to hipcc does NOT reduce max. Max is unchanged. D2flex merely turns hipcc off with PWM for lower levels, at full level hipcc is full-on as before.

 

[Benson]

I can get 3 of these batteries and put them in the mag85. I'll have to do something about the slightly off length and width, but otherwise I'm set for runtime and self discharge rates.

Check.

For the width, you need a sleeve -- IIRC 1" Schedule 40 PVC pipe is a good D-to-C size, but it might be a little tight on those cells. You can also use a rolled-up piece of corrugated cardboard, or wrap tape on the cells to bring them up to size.

For the length, the easiest fix is probably to sand the anodize off the inside of the tailcap, cut off the outermost coils of the spring so it's a tight fit down inside, and maybe shorten it at the top if it's still too long.

If my Dad wants to upgrade even further, to a P7, I'd also buy a hipCC and possibly a d2flex, if I want multiple levels. I would use the same batteries, but have to rework the head.

Check.

The benefit of the P7 would be 4/5 the output for ~2.5 hours, up from ~1 hour with the WA1185, if I just used the hipCC. If I add the d2flex, max gets lower, but I also get more modes, the lower of which will also work with alkalines.

Well, the P7 would drop you to more like 2/3 or 3/4 the output than 4/5, IMHO. Of course, it depends a lot -- you can't just compare emitter lumens vs. bulb lumens, as the spatial distributions are different, so the reflector/glass losses are different, but either way, the differences in color and beam distribution will be way more noticeable than the actual output difference.

As Alan said, you'll get the same high mode with or without the d2flex, it just adds the lower modes (with longer runtimes).

As for the charger, I could wire the cells in parallel and charge them on my Ultrafire WF-139. Because of the wiring it is unnecessary to check for unbalanced cells, correct?

I don't have the WF-139, but you should be fine; it will take a while, of course, since you're charging a 9Ah cell at (I think) 450mA, it's going to be something over 20 hours. I'd either get 6 cells (3 to run, 3 to charge) or a high-current single-cell charger -- those cells are rated to charge at 3A each, so anything short of 9A is ok if only used to charge the 3 in parallel. (The biggest cheap single-cell chargers I know of are <2A, like this.)

Since you start with your cells balanced, discharge them all the same (in series), they should be almost perfectly balanced when you put them in the charger. Any imbalance will be corrected by the parallel connection (instead of left to accumulate while charging them in series), and they'll be back fully balanced when they're charged.

That said, I would still check the balance after the first few discharges, just to verify that your cells are all equally strong -- if one is discharging substantially deeper, it will be limiting your overall runtime; you'd want to catch the defective cell, and get it replaced. Once you're satisfied they're all good cells, it's pretty much a no-checking-needed, idiot-proof setup.