Need help with M@g heat sink / head sealing for dive light

[arek98]

Hello all

First to clarify, I don't know anything about machining and how properly create CAD drawing to make part "machinable" . I just tried to document an idea. Pictures below are my first CAD drawing that is more complicated than cylinder with hole, so any comments are appreciated.

I was looking for dive light and read about lot o guys using Maglite as a host for conversion. What seems to be a biggest problem is sealing a lens. Original o-ring outside a lens seems to hold up surprisingly well but this is not where one would like o-ring in dive light.

To convert a Maglite to powerful LED light heat sink is needed as well. It seems like heat sinks for multiple LEDs disappeared somewhere and were replaced by lip type ones for single LED (SST-50, SST-90).

So, I have this idea of machining a part that will solve both problems (maybe it's not fresh but I didn't see it anywhere). Below are some of prerequisites/features.

• Lens is 3/16in polycarbonate (I managed to cut 52mm lenses from 3/16 Makrolon AR sheet). Thinner is probably too weak. Thicker would require gap between head and bezel, and only few threads would hold it.
• Will fit 32mm triple XP-G/XP-E board (from Cutter) + Ledil Cute optic (35mm diameter 15mm height). With XP-Es one can get about 10 degree beam. Cute Quad for XR-E will fit as well with Quad XR-E board from Cutter. I was thinking about SST-50 but there is no tight optic for it (one from Carclo is about 16 degrees). Eventually it could be second version of this heat sink.
• Maglite head is sanded a little (about 1.4mm) to remove gap between bezel with thicker than original lens.
• Top of heat sink is flush with head rim (or hair lower, easily solved by sanding little more, just not to much)
• Side o-ring is AS568A-132, top one AS568A-129 (Viton or EDPM)

Most important question is who will machine that? I don't have an access to machining shop and no one will machine just few pieces with reasonable price. Seems like drop-in than Mac is selling has heat sink that is pretty close. Maybe I can convince him to add this one to his offering.

Questions that I don't know an answer:

1. What exactly is the thread size? I believe it should be 20TPI 1.535in major diameter. Is that right? Is it enough information to machine it? What about thread shape, depth?
2. Do you think that there should be some space between Carclo optics and lens? Is it possible that lens will deflect under pressure (say 50m depth) enough to break lens/LEDs under it? Would 1mm be enough? We are talking 3/16" thick 35mm (unsupported) diameter polycarbonate.
3. What do you think about fitting it into Maglite head? Given an o-ring on side it should be pretty tight (even that pressure is low for an o-ring) but I don't know Maglite tolerances. If to close it may not fit in all heads.
4. Current drawing has 10mm space for driver (assumes it would be poted). Do you think it would be better to make it longer (about 1in) and machine flat surface parallel to tube for driver attachment? How much harder would it be to machine (not circular feature).
5. Do you think that narrow part (where driver sits) can be matched in size to ID of Maglite tube? Is there a risk that it may cause problem assembling it?

Thank you
Arek

 

[350xfire]

Having built about 4-5 Mag heads, I can tell you that you are pretty much over thinking this. Yours is a very nice design though, but will require way more machining that you need to make a mag head work.

What I do is:
1. Machine about 1.7mm of the head to allow for thicker lens
2. use 5mm lens. I use boro glass custom cut.
3. Use 48mmx2mm(I think) for both head o rings and a 124(I think) in the tail cap.
4. Use an off-the-shelf heatsink and reflector.
I have had this set up in my pressure pot without silicones or sealers of any kind without leaks!

Oh and yes 20tpi and you will need to cut the barrel and get rid of the factory switch.

 

[arek98]

350xfire

Thanks for suggestions. I agree with you that this may be overkill but it would be nice.

For the lens I think I prefer polycarbonate. It will not break and AR coating seems to be holding pretty well (not scratched intentionally though). Borosilicate transfers more light but if overstressed it will break in an instant while polycarbonate will bent. It is also easier to crack it when hit something.

I was looking at metric o-rings but I cannot find a place that sells them to public in small volumes. And I don't want Nitrile o-rings. They are least resistant to sun light and weather, I'm looking for EPDM or Viton (probably over thinking this one as well). Do you know where I can buy them?

So, you are saying that Mag head after replacing o-ring on side of the bezel is sealed well enough to let say 50m? I feel uneasy about original lens o-ring.

I don't want a reflector. Too much spill. I don't like aspheric (don't know exactly why). I prefer optics that gives larger hotspot with less spill.
I assembled one head using 3 Ledil LXP optics but they are little too high, they barely fit (with LED on star under it). I poured poting epoxy around them seal everything but it is kind of messy.

For barel I think I would like to have all options open. Remove threads from head, cut battel and glue them together or screw it as is with piezo or something in place of factory switch, depends.

 

[precisionworks]

who will machine that? I don't have an access to machining shop and no one will machine just few pieces with reasonable price.

Most machine shops are pretty reasonable, considering they've invested tens of thousands of dollars in machinery, tooling & shop space, plus invested enough years in training to do a job like the one you want.

Quite a few people on this board can make your parts. I'd guess roughly 10 hours set up & machining time to make one. Figure most shops will charge $60-$100 per hour & you can calculate the cost.

To me, a one-off custom dive light for under a grand is a bargain.

One of the board members is working on a dive light, which would probably cost less.

 

[arek98]

precisionworks

Maybe a one-off custom dive light for a grand is a bargain but this is a heat sink for Mag not a custom light. As for someone that wants this as a hobby your estimate seems very unreasonable to me (it would probably be reasonable if I was a company building prototype for production run). Could you, please, elaborate a little what exactly takes 10h? How much of that time is a setup? What would be a time necessary to make 10? I'm assuming you need to do setup before each operation with lathe. Does CNC need a setup, or can you just feed it with output from CAD (e.g., Solidworks)?

As I said I have no idea about machining. That's why I'm asking.

For reference I got a quote from rapid prototyping company and they will make one for me for $446 with 1 business day turn over (still way too much).

To be clear I'm interested in few pieces (4 at least, most likely never more than 10) and I'm not willing to pay more than $50 a piece. I know I can't expect to go to machining shop and get that price. This is why I'm asking here if someone would do that. Maybe just because likes an idea and wants few as well or maybe someone will see an opportunity to sell some here to make it profitable. Maybe someone just bought a lathe and wants to learn. For a grand I can as well buy a lathe and try to do this myself (probably will not do any good but I will have some fun at least).

If it can't be done for that price then I'm okay with that but I had to ask to know, right?

Drawing I made is how I would like it to be. If cost is to big maybe someone may suggest how to change it to make it easier/less expensive to machine? What seems to be a hardest feature to make?

I've seen that sonodeliri is working on whole light (actually a light head for canister light) and maybe I will buy 1 or 2 from him when he is done and has them for sale. But this is completely different thing. What I'm trying to get is a heat sink only for cheap Mag conversion.

Thank you

 

[wquiles]

I know I can't expect to go to machining shop and get that price.

So you "do" know $50 is not reasonable, but yet you don't want to pay more than $50. Yet, you want to still find somebody that can either do this at a complete loss, or basically do it for "almost free".

Am I the only one reading this right?

It is a completely custom heatsink, with two custom O-ring grooves, and a threaded end. It does not mater that it is for a Mag light - $50 is not even close to being remotely reasonable. My own estimate would put the custom heatsink at about $600 each, so the shop that gave you the $466 was actually a very reasonable offer.

Will

 

[arek98]

I'm looking for someone that may be interested taking design, machining bigger number of heat sinks and selling it to people. I would buy few of them. I'm not telling, I'm asking if this is possible. If not then be it. I'm not asking for someone to make 4 of them just for me and charge me $200. This would be obviously charitable donation.

Again I don't know how hard is to machine different features. Let say Shoppe is selling Mag heat sinks for about $20. What makes mine impossible to make for $50 price tag? Are groves for o-rings that expensive? Two different diameters inside? I don't think that fact of it being higher makes a big difference. Other than that it looks like PES heat sink (PES has identical threaded end).

 

[wquiles]

I'm looking for someone that may be interested taking design, machining bigger number of heat sinks and selling it to people. I would buy few of them. I'm not telling, I'm asking if this is possible. If not then be it. I'm not asking for someone to make 4 of them just for me and charge me $200. This would be obviously charitable donation.

Again I don't know how hard is to machine different features. Let say Shoppe is selling Mag heat sinks for about $20. What makes mine impossible to make for $50 price tag? Are groves for o-rings that expensive? Two different diameters inside? I don't think that fact of it being higher makes a big difference. Other than that it looks like PES heat sink (PES has identical threaded end).

All of those heatsinks for $15-30/each are made on a CNC machine. You would have to pay a couple hundred dollars just for the initial setup, and then a smaller charge for additional units.

So if it was $200 for setup, and you are making at least 10 pieces, then your fixed cost per unit is $20, plus whatever the CNC shot will charge per unit (materials, shop time, machine hourly rate, etc.) so it is easy to see how you can get to $20-30 each, although I suspect that I am way off on the CNC costs and that you need a much larger run before you can buy them for $20-30 each.

But you are asking for a couple, maybe 4 of them. A CNC shop would have charge a lot more per unit. It i s even worst on a manual setup, since you have to do each one by hand, which is a lot slower, therefore the cost has to be even higher since most all shops charge based on an hourly rate.

 

[arek98]

Thanks Will, so it seems like a lost cause .

Do I understant correctly that for short (10-20 pieces) runs CNC is least expensive? Is it possible to do something to make CNC setup easier? Like preparing input file in particular format? I guess it would depend on particular machine, right?

 

[precisionworks]

Does CNC need a setup, or can you just feed it with output from CAD

Wouldn't that be sweet ... send the CAD file to the machining center, the machining center pulls all the correct tooling out of the automatic tool changer, then sets the speeds, feeds, depth of cut, and does the profiling, threading, chamfering, etc.

Reality is that you send your drawings to the NC shop of your choice, they select tools, tool paths, sequence of operations, etc. For those "simple" hobby parts, probably 4-5 hours of setup charge ($400-$500) plus $40-$50 per part in quantities of 100 or more. The shop that quoted $446 is insane, but insanity isn't criminal ... they most likely need something for their machinists to do to avoid laying them off, so they do the job at zero profit. You should send them payment tonight, before they take a second look at that quote.

I'm not willing to pay more than $50 a piece

I can appreciate that. I'm not willing to pay more than$50k for a Porsche Carrera GT, even though it sells for 10X that amount. Both of us want to pay 10% of what something is worth, nothing wrong with being unrealistic.

Let say Shoppe is selling Mag heat sinks for about $20.

Call Wayne, send him your design, see what he charges. He's in business for the same reason I am - to make enough profit today to be here tomorrow.

Probably your best bet is to buy a lathe. With any luck, your cost for one part will be under $2500, meaning $250 each for ten. That's if you pay yourself nothing and are doing this "because you just bought a lathe and want to learn" :eeksign::eeew:

 

[350xfire]

It's a quanity thing! The more you buy the cheaper they get. So the set up for something is most of the work. Once you come up with cash to order lets say 1000 pieces, your price goes way down since the set up cost is spread out over 1000 instead of 4. That's just the way it is.

Again, nothing wrong with the stock mag design, in the right hands it can be made to be what you want. This is why DIY people love the mags.

 

[gadget_lover]

Here's why you can buy a $20 heatsink, but a custom one is hundreds,

For the discount model...
1) The design is whatever is simple to make
2) The sizes are what is easiest to set up and cut
3) The tolerances are whatever comes out of the machine.
4) The material is whatever was at hand or whatever best fit the bill
5) There is no way to make a mistake, as the customer is expected to to make it fit.
6) Since they are made in quantity, jigs (to hold pieces you are working on) and fixtures (to hold tooling) are created to make it easier to setup and cut.
7) As few tooling changes as possible go into the design.

Now compare that to the custom heatsink.

I see at least 3 lips, 1 through hole, 3 bores, 1 thread to a shoulder, 2 grooves, More than 4 concentric surfaces. That will require putting the part in at least twice, possibly more, and several tool changes. A mistake on any one of those steps means the part is trashed and you start over.

When you buy your own lathe the time you invest is 'free'. It is a great learning experience. I highly recommend it.

Daniel

 

[precisionworks]

I see at least 3 lips, 1 through hole, 3 bores, 1 thread to a shoulder, 2 grooves, More than 4 concentric surfaces. That will require putting the part in at least twice, possibly more, and several tool changes. A mistake on any one of those steps means the part is trashed and you start over.

+1

That's what I saw as well. It's a pretty complex little part, especially so for something simple & fast to knock out

When you buy your own lathe the time you invest is 'free'.

+1

Since no profit has to be made, it matters not how much time is taken on each part. If the part takes 10 hours or 10 minutes, it's all the same. If you get the process down to 30 minutes or so (where it would need to be to sell for $50) post that information here. Many of us, self included, would pay you to do our machine work.

 

[archer6817j]

You might try simplifying it like this:

I don't see any need for it to have threads since it's captured by the bezel. Also, all of the "steps" on the inside and outside are costing you money and they (as far as I can tell) serve no functional purpose. Even just the face groove requires buying or grinding a special tool.

You can post your request on the RFQ section of www.cnczone.com or directly to www.rfqwork.com (no affiliation with either but I'm a member on cnc zone). There are usually loads of people willing to do jobs for far less than they should Just be sure to post some dimensions, tolerances, and material requirements otherwise people will freak out.

The other thing I'd watch out for is the actual o-ring fit (depth/width of groove). In my experience "doing it by the numbers" on a one-off part doesn't usually work. If you find someone to machine this I'd send them the mag head and o-rings you want to use so they can test fit as they machine. That way you won't have a bunch of paper weights. You can buy o-rings from mcmaster. You won't find anyone who will sell you less than a bags worth.

 

[arek98]

Thank you Archer. This is what I was hoping, someone to say how to modify it to make it less expensive to machine. Step inside is because 35mm bore cannot go deep enough to fit LED board and optics. I think it would leave a wall (one where thread on my drawing is) too thin. On the outside back could be flat, ending where threads end (and drop threads of course).

Do angled walls for grooves serve any purpose, or they may be 90 degree as well? Are rounded corners important? I think at least one where o-ring will be pressed into would help sealing. Are such fillets expensive to make. What if groves were removed and replaced with cut out corners?

 

[archer6817j]

The o-ring grooves don't "need" to be tapered, although technically they should be. The corners at the bottom of the grooves should NOT have a radius. You may have already, but you can look up the standards for o-ring groove dimensions for a particular size o-ring. Basically there needs to be enough extra depth and width for the o-ring to "squish" the appropriate amount.

As far as rounding external corners, I would call out "break with file" on your drawing which just means knock the sharp edge down with a file. Chamfer or radius will cost money.

Also, you still have two changes in diameter (steps) in the outside diameter (towards the tail) and two steps on the inside bore. It looks like you will need to keep the two steps on the tail end; but keep in mind, every time you change diameter this is something the machinist needs to measure and adjust for...so basically twice the work. If you can make the inside one diameter it will be faster and cheaper.

 

[arek98]

Hmm, all o-ring grove/glands I've seen call for 0.005-0.015" radius on the bottom. Are you saying they are not needed?

How about that? It cannot get any simpler I think. It would just seal a head. Some "lip" type heatsink would be needed for LED like from Britelumens or somewhere)

Is it harder (more expensive) to make grove as opposed to cutting out a corner?
Thank you

Head would look something like that:

 

[archer6817j]

I don't think there is any real problem making it one piece. I think that's where you draw the line between function and simplicity.

It's an interesting idea to just use a lip for the o-ring seal. I've actually been toying with the same thought so you should try it out! The thing to keep in mind is that the o-ring will be captured by (a) the sink and (b) the body of the light. So, the body is acting as one wall of the o-ring groove and that "lip" should be sized accordingly to get enough pressure on both faces of the o-ring. So, basically the grooves should be shallower. However, this does leave a small issue at the head end of the heat sink because you will either need another step or something to provide radial pressure on the o-ring.

You are right about the o-ring groove. I guess I meant the groove itself should not be a radius but a square groove. I believe the very small radius is recommended for strength, rather than o-ring performance. That should not be a factor here. Keep in mind that a lot of cutting tools, especially on the lathe, have a "nose radius" and when working into a corner you will see that radius. Is it easier to machine a lip than a groove? Maybe depends on what tools the machinist has on hand.

I think you are getting close to a good compromise solution!

Oh and you would be best served by having some air space between the lens and the optic.

 

[arek98]

I think I like an idea of splitting it into two parts. Having separate part for sealing would allow use of existing heat sinks for SST-50 and SST-90. My original heat sink was to shallow for any reasonable SST-50 or SST-90 optic or reflector. It will still limit optic diameter to 35mm but well, something has to give. Other advantage would be that having heat sink with LED's attached to tube there is no problem with twisting wires when screwing on a head.

Disadvantage is that the only thing holding it in place would be lens (bezel) but I hope it will be enough. Maybe diameter may be made to make a real tight fit. I was thinking about using epoxy between sealing part and Mag head but I'm not sure about that.

Do you think that I really need something on the outside of the lens o-ring to hold it? If I make a grove/lip OD hair bigger than ID of the o-ring it should hold close to it. Any pressure from outside will push against sealing part as well.

Thank you Archer for help.

 

[precisionworks]

all o-ring grove/glands I've seen call for 0.005-0.015" radius on the bottom

Parker is the authority on O-rings, and you can download their O-ring handbook: http://www.parker.com/literature/ORD%205700%20Parker_O-Ring_Handbook.pdf

O-rings are installed into square grooves 99% of the time. I see (every few years) an application where the manufacturer cast or milled a half round circle to hold the ring - just replaced an O-ring on a Wehmer 108 dental trimmer, and that groove was cast into the backplate. Love that design, as the O-ring has no where to go, gets overcompressed easily, and has to be replaced on a regular basis when it no longer seals & water runs onto the floor