Sold/Expired FS: empty LED tower for Surefire M3T/M4/M6/KT1/KT2/KT4 head(accept 17mm driver)

[Justin Case]

If you wire two Sanwich Shoppe drivers in parallel (e.g., 2xGD or 2xSOB), you can double the drive current. Thus, if the tower can fit a double stack, you can hit 2A or 3A drive.

I would use a set screw to fix the stem, not a nut, to give max space for the driver(s).

 

[blackbalsam]

If these become available i would be interested in a complete LED tower.

 

[precisionworks]

since this is an LED unit (and probably fully potted for most of us), shock compensation is all but irrelevant-- would it be advisable to eliminate the outer spring and use the space to increase mass?

That sounds like a great idea. The increased thermal mass should help, as will the added area that's near (or in contact with) the inside of the head.

i would be interested in a complete LED tower.

I'll be happy to do the machine work & mechanical development, but one of the electronic modders will be best qualified to assemble the package. Justin Case, are you available :devil:

if the tower can fit a double stack, you can hit 2A or 3A drive.

The sweet spot for the XM-L seems to be around 2500 mA, and I'm guessing that most users will want the highest setting at that level. The data suggests 1000 bulb lumens, meaning about 700 lumens OTF. That will put some spark in the old M4.

 

[notrefined]

I'll be happy to do the machine work & mechanical development, but one of the electronic modders will be best qualified to assemble the package. Justin Case, are you available :devil:

that was my plan 🙂

The sweet spot for the XM-L seems to be around 2500 mA, and I'm guessing that most users will want the highest setting at that level. The data suggests 1000 bulb lumens, meaning about 700 lumens OTF. That will put some spark in the old M4.

I see a design coming together... :devil:

Of course, I'll want mine warm....unless there's a high CRI XML by the time we get this underway 😉

 

[notrefined]

I'll be happy to do the machine work & mechanical development, but one of the electronic modders will be best qualified to assemble the package. Justin Case, are you available :devil:

that was my plan 🙂

The sweet spot for the XM-L seems to be around 2500 mA, and I'm guessing that most users will want the highest setting at that level. The data suggests 1000 bulb lumens, meaning about 700 lumens OTF. That will put some spark in the old M4.

I see a design coming together... :devil:

Of course, I'll want mine warm....unless there's a high CRI XML by the time we get this underway 😉

 

[precisionworks]

One of our members is sending a tower for me to examine and evaluate. That tower will be compared to the sure fire tower in my M4. After documenting all the dimensions of both towers, a proto type will be machined from Delrin. I am trying to design this "outside the box" and expect that this design will have some significant differences from existing towers. Maximizing total thermal mass and achieving a close fit to the inside of the head are high priority items. Also important are leaving room to stack two drivers so that a drive level of 2000-3000 mA can be achieved. Plus a provision for post length adjustment.
Guessing that model development will require 20 hours or so, spread out over the course of a couple of weeks. That model will be used to write a program for the Haas CNC lathe & machine some parts from C145.
And then I will have a better idea about final pricing 🙂

 

[blackbalsam]

One of our members is sending a tower for me to examine and evaluate. That tower will be compared to the sure fire tower in my M4. After documenting all the dimensions of both towers, a proto type will be machined from Delrin. I am trying to design this "outside the box" and expect that this design will have some significant differences from existing towers. Maximizing total thermal mass and achieving a close fit to the inside of the head are high priority items. Also important are leaving room to stack two drivers so that a drive level of 2000-3000 mA can be achieved. Plus a provision for post length adjustment.
Guessing that model development will require 20 hours or so, spread out over the course of a couple of weeks. That model will be used to write a program for the Haas CNC lathe & machine some parts from C145.
And then I will have a better idea about final pricing 🙂

I am so glad to see this project is moving foward....Thanks to those involved.....Robert.

 

[Justin Case]

The sweet spot for the XM-L seems to be around 2500 mA, and I'm guessing that most users will want the highest setting at that level. The data suggests 1000 bulb lumens, meaning about 700 lumens OTF. That will put some spark in the old M4.

When I drove a 2S2P MC-E with an SOB1227 (about 7.6W draw by the LED, giving about 5.7W of waste heat, assuming 25% efficiency, with another ~1.3W waste heat from the driver, assuming 85% efficiency), the thermocouple temp measured at the interface between the reflector opening and the tower flange hit 162F. IMO, this is a bit hot. Based on the XM-L datasheet, at 2.5A forward current, Vf is about 3.25V, giving a total power draw of about 8.1W. So the XM-L would no doubt run even hotter at 2.5A drive.

IMO, the XM-L sweet spot is where the relative luminous flux vs forward current curve noticeably departs from linearity. To my eye, that occurs at around 1.6A to 1.8A.

I've built several XM-L towers using both the Netkidz and ArcMania/AW tower designs, and 2A is the max drive current I would use. But that might depend on how you use the tower. If you use it for short bursts only, then perhaps 2.5A drive current could be fine. If you also use it for long, continuous runs and require long total battery run time, then something closer to 1.5A might be better.

 

[Justin Case]

One of our members is sending a tower for me to examine and evaluate. That tower will be compared to the sure fire tower in my M4. After documenting all the dimensions of both towers, a proto type will be machined from Delrin. I am trying to design this "outside the box" and expect that this design will have some significant differences from existing towers. Maximizing total thermal mass and achieving a close fit to the inside of the head are high priority items. Also important are leaving room to stack two drivers so that a drive level of 2000-3000 mA can be achieved. Plus a provision for post length adjustment.
Guessing that model development will require 20 hours or so, spread out over the course of a couple of weeks. That model will be used to write a program for the Haas CNC lathe & machine some parts from C145.
And then I will have a better idea about final pricing 🙂

One thing to watch out for if you design the tower for a 14mm Shoppe driver is the relevant Shoppe drivers (BBNG, GD, and SOB) have components close to or right on the edge of the board (most notably the inductor). The ArcMania/AW tower design uses a very narrow shoulder as a stop for the driver. If the shoulder were any wider, you'd risk shorting against some surface mount component on the board. And as-is, things are already close. An anodized aluminum base could solve this problem, but that adds yet another manufacturing step. Currently, I de-solder Shoppe inductors and move them inward.

So merely building a prototype in Delrin might lead you down the garden path if you don't consider the tower as a complete system and account for electrical isolation issues.

Also, different LEDs have different die heights relative to their bases. That leads to different overall tower heights (what I call the stem height). So taking dimensions from some existing tower might also lead you down the garden path if that tower isn't designed from the get-go for an XM-L or whatever LED you plan to target. I would think that you'd want to build a proto tower with the target LED mounted on it and compare that to a stock SureFire MN incan tower to establish the optimum tower stem height.

And even if that target LED is an XM-L/XP-G, will the emitter be mounted on an datiLED 8mm XM-L board or Shoppe 8mm XP-G board (both about 0.8mm thick), or on some larger board (e.g., a Cutter 10mm XP-G board that is 2mm thick)? That will also affect the final stem height.

Leaving room for a double stack of Shoppe drivers is great, but there probably should be capability to accept a single driver as well (e.g., an SOB1500). If the single driver sits too deep in the driver cavity, you might need a different center spring length, which complicates inventory issues somewhat.

 

[precisionworks]

the XM-L sweet spot is where the relative luminous flux vs forward current curve noticeably departs from linearity. To my eye, that occurs at around 1.6A to 1.8A.

The practical side of my brain totally agrees with your numbers. The other side says that the customer is always right :nana:

I'm almost certain that some users will want a sky high drive level, and 3500 mA would make them smile. OTOH, 2400-2500 mA will probably satisfy most users, as long as they understand that this is for "burst mode" only, not to exceed one minute at that level.

A second drive level at 1500 mA would give the "regular" high setting, useful for longer periods of time with good battery life. If possible, a third setting around 500 mA would give a low setting that still had decent brightness plus lots of runtime.

One thing to watch out for if you design the tower for a 14mm Shoppe driver is the relevant Shoppe drivers (BBNG, GD, and SOB) have components close to or right on the edge of the board (most notably the inductor). The ArcMania/AW tower design uses a very narrow shoulder as a stop for the driver. If the shoulder were any wider, you'd risk shorting against some surface mount component on the board. And as-is, things are already close. An anodized aluminum base could solve this problem, but that adds yet another manufacturing step. Currently, I de-solder Shoppe inductors and move them inward.

So merely building a prototype in Delrin might lead you down the garden path if you don't consider the tower as a complete system and account for electrical isolation issues.

Also, different LEDs have different die heights relative to their bases. That leads to different overall tower heights (what I call the stem height). So taking dimensions from some existing tower might also lead you down the garden path if that tower isn't designed from the get-go for an XM-L or whatever LED you plan to target. I would think that you'd want to build a proto tower with the target LED mounted on it and compare that to a stock SureFire MN incan tower to establish the optimum tower stem height.

And even if that target LED is an XM-L/XP-G, will the emitter be mounted on an datiLED 8mm XM-L board or Shoppe 8mm XP-G board (both about 0.8mm thick), or on some larger board (e.g., a Cutter 10mm XP-G board that is 2mm thick)? That will also affect the final stem height.

Leaving room for a double stack of Shoppe drivers is great, but there probably should be capability to accept a single driver as well (e.g., an SOB1500). If the single driver sits too deep in the driver cavity, you might need a different center spring length, which complicates inventory issues somewhat.

I sat down with the owner of the CNC shop this morning & we discussed the tower project. He feels confident that he can program the lathe to cut the base of the tower so that it totally fills the "socket" or cavity that is the forward section of the M4 tube. Since there is always some manufacturing variance between individual lights, a small amount of clearance would be left so that the owner could pack copper tape around the base for a perfect fit.

This would result in a base that is somewhat wider, and perhaps deeper, than the factory tower. I need to take the head of the M4 apart to better see how things fit - any ideas on that? Freezing is my first choice, but I'm open to any option that does not destroy the head or reflector.

If dimensioned drawings are available for the driver boards & emitter board, that would be a great help.

When the standard SF reflector is viewed from the rear, the stem opening seems quite tall. Would it be beneficial to shorten the stem opening (with a corresponding shortening of the stem) so that a taller base can be made. My thought on this is that a taller base has more mass and would act as a better heat sink/heat spreader. It would also give more room for double driver boards.

 

[tx101]

I need to take the head of the M4 apart to better see how things fit - any ideas on that? Freezing is my first choice, but I'm open to any option that does not destroy the head or reflector.

RPM sold some M6 bezel removal tools here

http://www.candlepowerforums.com/vb/showthread.php?210226-FS-M6-Titanium-Bezels-%283-pc.-Added%29-and-Bare-Aluminum-Versions-Available


Actual method/s posted here

http://www.candlepowerforums.com/vb/showthread.php?227625-Removing-a-SF-M6-bezel


If you think taking your M4 head apart is a PITA I have a spare M6 reflector I could donate for R&D 😉

 

[notrefined]

When the standard SF reflector is viewed from the rear, the stem opening seems quite tall. Would it be beneficial to shorten the stem opening (with a corresponding shortening of the stem) so that a taller base can be made. My thought on this is that a taller base has more mass and would act as a better heat sink/heat spreader. It would also give more room for double driver boards.

I guess this is where I'll differ- I would prefer not to alter the stock components of the light in order to make the LED tower work better, because I want to retain the full functionality of the light as an incan as well. for me, that's the primary reason to pursue the tower idea in the first place.

 

[Justin Case]

I agree to leave the TH reflector stock so that interchangeability with SF incan towers is retained.

Also, it is clear that the tower base can be enlarged in diam and still be compatible with SF THs and 6P-style necks. The base of the reflector opening also has a hemispherical cavity, which reduces the contact surface between the tower base and the reflector. Thus, the tower base could be machined to have a matching hemispherical bulge, instead of having a flat top face.

However, I think this hemispherical mod to the base is unnecessary. See my back of the envelope calculation here, for example.

 

[notrefined]

I guess that would depend on the relative cost (time and extra material) to machine the hemispherical vs. flat surface...if the cost is small it might be reasonable to do it even in the face of small benefit. Of course, like most of these things, the cost will probably turn out to be quite large lol

 

[Justin Case]

I guess that would depend on the relative cost (time and extra material) to machine the hemispherical vs. flat surface...if the cost is small it might be reasonable to do it even in the face of small benefit. Of course, like most of these things, the cost will probably turn out to be quite large lol

Based on calculations, a hemispherical top is a total waste of time and money.

 

[Justin Case]

RPM sold some M6 bezel removal tools here

http://www.candlepowerforums.com/vb/showthread.php?210226-FS-M6-Titanium-Bezels-%283-pc.-Added%29-and-Bare-Aluminum-Versions-Available


Actual method/s posted here

http://www.candlepowerforums.com/vb/showthread.php?227625-Removing-a-SF-M6-bezel

Mettee sold a bezel ring removal tool along with his replacement smooth reflector. See here.

 

[precisionworks]

I also agree that the hemispherical surface would be of little value. (Although it would be easy to program for that feature and would add very little cost.)

I do understand the reasoning behind leaving the reflector stock. No need to fix something that isn't broken.

 

[notrefined]

sounds like something I would do just because I could...so it's probably a good thing that I can't 😉

 

[precisionworks]

LOL 🙂

I will cost out that feature and expect it might add a dollar or two. The additional thermal mass should not hurt anything & it makes a Chinese clone that much more difficult to manufacture.

During the late 1800's and early 1900's, L S Starrett in Athol, Massachusetts, manufactured their measuring instruments using non standard screw and tap sizes. Even though nothing was offshored at that time, Starrett understood that no other manufacturer would go through all the time and expense of having special tooling made in order to produce a knock off. Not a bad idea.

 

[Justin Case]

If you add the spherical cap to the tower base, I hope that SF's machining of that cavity in the reflector is consistent. The incan towers stop on the flat part of the reflector opening, not the spherical cavity. So if that cavity depth and/or diameter varies enough so that the spherical cap on the tower prevents full insertion of the stem, you may run into problems with the tower focus.

Perhaps you can make the cap undersized enough to account for any reasonable amount of machining tolerance. You'll get a gain in additional thermal mass, but basically no additional gain in contact area (which looks unneeded anyway).

I mic'ed the spherical cavity in one of my KT1 THs, and got about 0.61" wide and about 0.12" deep (to the edge of the reflector opening). Technically, the spherical cavity would go deeper than that. If I stay with 0.12" deep, then the spherical cavity actually has a flat at the apex. But I think as an approximation, this is close enough.

So assuming that the spherical cavity is 0.61" wide (0.61" chord length) and 0.12" deep, then the matching spherical cap on the tower base will have

c = chord length = 0.61"
h = spherical cap height = 0.12"

Doing some magic algebra and geometry, the radius R of the sphere that has a spherical cap with a chord length of 0.61" and a cap height of 0.12" is given by

R = [(c^2)/4 + h^2]/(2h) = 0.89"

Using more magic algebra and geometry, the volume V of the spherical cap is given by

V = (1/3) * pi * h^2 *(3R - h) ~ 0.04 cu. in. = 0.66 cc

However, you have to subtract the volume of the part of the stem that would pass through the center of this cap to get the final value for the additional volume added by the cap. Again, I'm going to assume that the top of the cap is flat to approximate the volume of the stem to subtract off.

The main part of the Netkidz tower stem is about 0.373" diam = 0.947cm (0.4735cm radius). The height of the section of stem to subetract off is 0.12" (0.305cm). Thus,

Vfinal = V - (pi*0.4735cm^2*0.305cm) = 0.66cc - 0.21cc = 0.45cc

To visualize how much additional mass this is, let's see how much 0.45cc is in terms of additional tower stem length.

0.45cc is equivalent to an additional stem length of about

h = 0.45/(pi * Rstem^2) = 0.45/(pi*0.4735^2) ~ 0.64 cm.

Thus, adding a spherical cap to the tower base would be like adding about 2/3 of a cm of length to the tower stem. That seems like a fair amount of extra mass. The total stem length is about 2cm (and part of the top of the stem is actually a smaller diam), so the cap is like adding about 30% or more of additional stem mass.

My comment previously in post #75 that a spherical cap is a total waste of time was based only on any additional performance benefit from the added contact area to the TH reflector. As the above calculations suggest, there is also added thermal mass, which appears to be potentially significant.