PDA

View Full Version : Having trouble cutting aluminum sheet



fastturbovet
05-24-2005, 12:54 AM
We have an application where I'm cutting .020" 6061 AL sheet. We are doing it on a Bridgeport CNC which can go up to about 3500rpm or so. We're taking a piece thats 9" x 2" and I am cutting out a bunch of channels with a 1/16" endmill. The item is a heatsink and I need the channels cut out so that it will not touch any exposed traces on a circuit board when its installed, so they need to be accurately cut and with as good and "square" a cut as possible.

I am having a couple of problems. The first is that its extremely hard to hold down the sheet. Because I am cutting it, there ends up being small peninsulas of metal which are not very rigid. Any tips on holding it down? Problem is it will be bonded with epoxy to a PCB later, and if I use tape it can leave residue which is tough to get off (since the sheet is thin you can't rub it hard without bending/deforming it a little). I thought about making a custom adapter plate for a vacuum clamp? I also thought about taking a thicker piece of material (maybe steel) and cutting out the same channels in maybe 3/64 width and clamping that down over my AL sheet before milling. We also thought about stacking the sheets maybe 3-4 thick and clamping them down with a top-plate to help with the problems. Any other ideas? Which of these ideas sounds best?

Second problem - getting a pretty bad finish on the cuts. There are tons of burrs and hanging bits of shredded AL all along the edges of the cut. Is this a vibration problem? A tool deflection problem? A feed/speed problem? Or maybe a combination? I am cutting dry with no coolant, and the last time was full depth of cut (.02"), about 3,000RPM with about 4ipm feed. That was a "WAG" )wild-assed-guess) so I could be way off. What could I do to help with this? Different feed/speed? Fixture the part better? Carbide endmill? Less depth of cut? Is my DOC OK doing the full .02 at a time with a 1/16" endmill? Or should I try to take it easier?

And last but not least, I am not necessarily tied to 6061AL nor to .02" material (but I'd like to stay somewhat close). If I went to .05 or .06" thickness AL would it be easier to machine and fixture? I am also not tied to 6061AL. I can get 7075 in the same sheet size, so thats an option too. Being more brittle I wondered if it might flake easier and not burr up so muhc?

Any tips welcome - trying to get this figured out in the next day or so :)

harvard5
05-24-2005, 02:10 AM
Try the 7075 we use it all the time repairing aircraft it finishes alot better than 6061. I don't know much about holding it down but I have used a vacum table to hold down balsa wood to use a router on it and it worked great!
Bill

cbass
05-24-2005, 03:49 AM
Hey, you guys are tied for posts :)

Sorry, couldn't resist (its late, I'm tired etc..)

Spezzaferro
05-24-2005, 02:07 PM
Harvard5 is right, 7075 is a lot better to machine than 6061
I cut thin sheet a lot and I use a double sided vynal adhesive that I pick up
at a sign shop supplier, It's quite pricey but it works great and leaves no residue.comes off all in one shot. I use a Carpet roller to roll on top of the sheet to ensure good adhesion.
Also I use a sacrificial piece of aluminum underneath and . with .020 material
i would go .050 deep, but cut your sacrificial first,

mxtras
05-24-2005, 02:30 PM
This seems like a tough one, that's for sure.

5052 is an extremely common .020" coil and cuts well - another option, maybe.

How many of these do you have to make?

I think I would try a clamp plate arrangement to trap the material between two plates. You could run the part profile in the top plate with a larger diameter tool for clearance, then maybe even relieve the bottom of the top plate to increase the local clamping pressure around the windows.

Vacuum doesn't seem like an option, but I don't have a clear image of exactly what you're doing.

Scott

little bubba
05-24-2005, 08:09 PM
For the burrs I would go with a carbide endmill, your probably bouncing all over the place. Are you climbing, conventional or full width? Generally climbing will keep you almost burr free, if your going full width is it possible to take a .002" or so clean up pass? You didn't say what quantity your making but if its only a few I would go with the double sticky tape, just drop it into a shallow pan of acetone after if there is some residue. If you have to do a whole bunch, a steel template over the top sounds like a good idea, a vacuum chuck sounds like even more fun.

fastturbovet
05-25-2005, 10:51 PM
Thanks for the feedback. I am going to a local metal shop tomorrow to see if I can't pick up some 7075 and also some 1/8" stock to make a top-plate.

I have to make quite a few of these. Its for an electronics doo-dad we're making and we will probably need about 100-150 a month to start, and once we convert some other electronics products over to this new style we will need maybe 200-300/mo or so. Sticky tape would probably work but it gets expensive and its difficult when you consider how many we'd be doing.

I am going to go with the top clamp and 7075 and see how it does. The problem is, imagine if you were taking a 12"x12" square, and cutting (from the center) 5" one way, moving up 1/2", cutting 10" back, going up 1/2", cutting 8" back, etc, etc. You would have these thin slivers of material that were *extremely* flexible. Thats the problem I have - once the machining is done, the part is flexible as a sheet of jello. I will try the double-sided tape and use acetone to remove if the clamp idea doesn't work as hoped :)

Thanks!

Mike

miljnor
05-26-2005, 12:40 AM
when making a sandwich style set up its best not to cut the channle in the top plate first... just sandwich the material inbetween to plates of material and cut everything at once. if your top plate and sub plate are thick enough and the sandwich is nice and tight this will also eliminate your burrs.

The other method is wax but you said you didn't want a residue. and you would still have to melt the wax off. the lower melting point wax melts in boiling water.

Karl_T
05-26-2005, 08:34 AM
Considering how many you're making, look at sending this out to a laser. Tolerance on laser cut is around 0.010 may be a shade better, might be a problem for you.

Karl

ViperTX
05-26-2005, 11:11 AM
If the backside of the aluminum doesn't need to be flat and in the quantities you're speaking off....I would look at some type of forming rather then machining....tool & die comes to mind as well as hydroforming....

miljnor
05-26-2005, 11:16 AM
the company i use for laser can keep +_ .002 on that thin of material

skippy
05-26-2005, 02:26 PM
As vipertx says, I would definately look at a tool and die set up. Considering you have a mill to make the tooling and the product quantities are not large, a manually operated set up would take no time to punch out 100 - 200 units. I guess first compare costs with contracting out to a laser guy.

take a cut
05-30-2005, 08:38 PM
not knowing all the info makes it hard to say although if the quantities are as you say i would probably build a double stage punch to punch this out possibly. The first stage springloaded to clamp and strip the plate and the second power stage to punch??

Scott_bob
05-31-2005, 12:35 AM
2 more cents...

With the quantities you're into, vacuum tooling would definetly be the way to go.
Also, when trying to finish the edges of thin material, this is the one time you'll want to try conventional milling. I say "try" because you almost always want climb cut. The problem with milling the thin material along an edge is the material wants to peal up into your cutter, I think this is what you have described...

Try conventional cutting, it tends not to pull the material up during milling. Not knowing your shape, it is hard to advise clamping. If you possibly can, go for a vacuum clamping method. Can you post a quick image of your shape?

Sincerely,

mxtras
05-31-2005, 09:12 AM
If the laser cutting route is not attractive to you and you are committed to cutting these, you might also want to get your hands on reverse helix end mills to push the material down during cutting. Chip extraction will then become an issue - but it's another option.

Vacuum? I can't see it. Not for what you have described. The thin webs don't offer enough surface area to provide enough clamping force.

Actually - the laser route seems the most logical.

Scott

fastturbovet
06-01-2005, 01:55 AM
update...


Here is a pic of what I am trying to machine (with a quarter next to it for size comparison - the piece is about 8" long and 2" high, give or take, and I upped the thickness to .032 on 6061 and .050 on 7075 which seemed to help a lot since it was more rigid)

http://www.hightechspeed.com/ebay/pcb.jpg

Well, I tried machining out of 7075 and it worked very well. I also tried a carbide endmill and both 6061 and 7075 worked very well. The only problem now is that its taking 30-45min to cut one. I am using 2" dense foam backing and using double sided tape to adhere it. The tape is a PITA to get off though (although soaking in MEK or Acetone helped a LOT).

I talked to a few places about getting it laser cut. I called 4 shops... one said $45/ea, the other said $29/ea, one took my info and never called back, the fourth I could never get them to give me any info on actually getting it done. Given this, I gave up. My budget is about 1/100th their quotes :) Its part of an electronics doodad (the plate is epoxied to a circuit board and there are little metal "ears" around the edge that connect the piece but are cut off after its bonded to the circuit board, to make a bunch of small but seperate 'heat sink pads'). The laser shops I talked to said it would take a long time to machine because the laser is very small so each cut above would actually be a loop on the laser, whereas with a CNC mill I can use a 1/16" mill (the size of the tracks) and do it in a single pass.

I was thinking it may be a good candidate to get stamped? I was thinking of talking to some stamping shops and seeing what they say. From what I know of stamping, a die is made which is $$$ but the per-part cost after that is very low, in quantity. It needs to be flat but thats so it lays flat against the circuit board. If stamping would curl the edges a litlte, I could get it staped from the back and at least the non-flat part would be sticking up (which would be OK if it was off just a little bit)

I need about 100 a month but if the price was low, I'd be happy to buy them several hundred or even a thousand at a time.

Any more ideas/comments? My plan at the moment is to just fixture as large a sheet as I can on my CNC machine and let it run and make these unattended, then leave them in an acetone or MEK bath to dissolve the tape and wash them off afterwards. I would prefer to get them made for me, but really the budget is less than $1-2/ea since anymore would mean I'd have to raise my product price, and I have contracts in place with distributors so I can't do that for 18mos or more.

skippy
06-01-2005, 03:51 AM
Mike,
(A) I went to your website and I have seen what you are trying to do. I am heavily into hi-brite and ultrabrite LEDs also and I am 100% convinced that you do not even need this heat sink! Why? Because the only time you'll be driving the LEDs at their full capacity is when you turn on the indicator (blinker) function or stop light and in the case of the stoplight that's for about 3 seconds maximum. Yes, heat dissipation is an important issue with this style of led BUT only when running constantly at full capacity in high temperature situations. Take your situation (not many leds, not being driven at full current for more than few seconds in the case of the brake, generally in the wind due to being a part of a motorcycle) and now compare that with the example of a fixed, permanently on, speed sign (mph) using 300 LEDs all mounted in a black metal case, at midday in Kansas in summer. Get my point?
For further info log into http://forums.linear1.org/ You'll find Rob very knowledgeable and happy to answer any specific questions you may have regarding cooling. Hopefully you have chosen to use groups of LEDs that are wired in series. Generally about 4 LEDs to a group depending on the mA chosen but you can use the array calculator for that. Each group has the required resistor and groups can then be paralelled together. This is much more efficient than each led running it's own resistor.

(B) Second issue is that heat dissipation is achieved via the pins/legs. If I understand correctly you want to have a circuit board with LEDs mounted in the usual manner except to fit the heatsink on the non copper side of the circuit board. Even if heat dissipation were an issue, which I believe it's not in this case, having the heat sink on this side of the board where there is no heat won't achieve anything. My advice to you is to use the thickest FR4 board material you can find (for stability against the vibrations of the bike) with the thickest copper layer you can find to dissipate the heat. From memory, 0.124" thickness with 1oz laminate thickness is available but you'd have to do some more research on that. At least you're in the "anything is possible all it takes is money country" which is a good thing.

(C1) Alternatively, buy straight FR4 board (without copper laminate) in +/- 2mm thicK and also 1mm copper sheet. Get the FR4 and the copper guilotined to just over 8" x 2". Bond the 2 pieces together. Modify your circuit board layout so that all you are doing is cutting the separator tracks and nothing else so as to leave the maximum behind. Make a vacuum hold down arrangement specific to your job which should be very simple, fast and cheap to do. The heat sink and circuit board will be one and the same with this method. Route it all out followed by assembly. You could even put some grooves in the copper to "supposedly" add cooling.

(C2) As per C1 but get one die to punch out external shape of FR4 board, another die to punch out 1mm copper shape as per your picture, bond both together, again using copper as the circuit board and away you go soldering/assembling the board with your router only being required to do the hole drilling.

(D) If you feel you need to have some type of visual gimmick such as a heatsink to give you a "perceived" advantage over any competition you may have, think about a "dumby power management chip" Doesn't work but looks hi-tech. I'm not having a dig at you here as I have enough commercial background to know that sometimes it can be advantageous to have a gimmick of some kind. Sorry I've raved on so much.

SRT Mike
06-01-2005, 08:11 PM
Edit...

fastturbovet
06-01-2005, 08:13 PM
Mike,
(A) I went to your website and I have seen what you are trying to do. I am heavily into hi-brite and ultrabrite LEDs also and I am 100% convinced that you do not even need this heat sink! Why? Because the only time you'll be driving the LEDs at their full capacity is when you turn on the indicator (blinker) function or stop light and in the case of the stoplight that's for about 3 seconds maximum. Yes, heat dissipation is an important issue with this style of led BUT only when running constantly at full capacity in high temperature situations.

I am 100% sure I need the heat sink :). I am not using SuperFlux LED's on this board, I am using a different type of LED. They *require* 1sq-in of supplemental heat sink per LED. Also, they have a very low R theta j-a of about 100C/watt, and a max junction temp of 125c. They run at 150mA and they will toast themselves in about 2 seconds without a heat sink. Also, it's critical that the LED's be able to be run at 100 duty cycle for extended periods of time without burning out. I have about 6,000 dealers of our products in the USA so if I send something out that even fails in 1% of cases, I am (as we say) "screwed" :). Furthermore, the LED I am using has such a low max junction temp that it cannot even be soldered to a circuit board, it must be mechanically attached to a metal substrate which provides the mount as well as the heat sinking.



Take your situation (not many leds, not being driven at full current for more than few seconds in the case of the brake, generally in the wind due to being a part of a motorcycle) and now compare that with the example of a fixed, permanently on, speed sign (mph) using 300 LEDs all mounted in a black metal case, at midday in Kansas in summer. Get my point?

I think the bikes are much hotter... most new motorcycles (at least sportbikes) have undertail exhausts. Its very common for the light to be less than 3" from the exhaust pipe, and we've seen temps of well over 150 degrees in testing.


For further info log into http://forums.linear1.org/ You'll find Rob very knowledgeable and happy to answer any specific questions you may have regarding cooling. Hopefully you have chosen to use groups of LEDs that are wired in series. Generally about 4 LEDs to a group depending on the mA chosen but you can use the array calculator for that. Each group has the required resistor and groups can then be paralelled together. This is much more efficient than each led running it's own resistor.

Thanks for the info - I checked out the site. I'm pretty familiar with the subject though :) We are working directly with the engineers at Lumileds on this project - we sell many thousands of lights per year and this new design is a complete departure from what they normally do, so everyone is pretty psyched about it. We have about 80 tail lights we manufacture but this is the first one using the new LED which requires the metal substrate, and its a smaller volume product so its a good test-bed. The LED's are run each on an individual circuit, because we use a microcontroller and PWM to modify the brightness level of the light, but we are also using a switcher regulator which is pretty efficient and precludes the need for beefy resistors at each LED. The problem with stringing too many LEDs together is the variance in brightness due to fluctuations in vehicle voltage, but the switcher solves that.


(B) Second issue is that heat dissipation is achieved via the pins/legs. If I understand correctly you want to have a circuit board with LEDs mounted in the usual manner except to fit the heatsink on the non copper side of the circuit board. Even if heat dissipation were an issue, which I believe it's not in this case, having the heat sink on this side of the board where there is no heat won't achieve anything. My advice to you is to use the thickest FR4 board material you can find (for stability against the vibrations of the bike) with the thickest copper layer you can find to dissipate the heat. From memory, 0.124" thickness with 1oz laminate thickness is available but you'd have to do some more research on that. At least you're in the "anything is possible all it takes is money country" which is a good thing.

No, the LEDs are not mounted in the traditional way. A lot of work has gone into the determination of what conditions the light will be used under as well as the exact amount of heat-sink material required. The LED's will be mounted to this heat sink and this heat sink will then be mounted to a PCB. The heat sink is the conductor and the heat dissipator. The copper on a normal FR-4 isn't nearly enough to dissipate the heat from the LEDs in this case.



(D) If you feel you need to have some type of visual gimmick such as a heatsink to give you a "perceived" advantage over any competition you may have, think about a "dumby power management chip" Doesn't work but looks hi-tech. I'm not having a dig at you here as I have enough commercial background to know that sometimes it can be advantageous to have a gimmick of some kind. Sorry I've raved on so much.

no insult taken, but I can assure you the metal plate is very functional - no gimmicks! We're kind of on the bleeding edge of LED automotive applications at our shop, so there is a whole bunch of thought and engineering in this product. The method being used currently is the best approach. The LED's require the heat sink to function correctly, and we move tens of thousands of lights per year, so the product must be 100% reliable, even if used at 100% duty cycle in the desert in the middle of july for 30 minutes straight full-blast brake light.


Still interested in feedback on whether stamping would be an appropriate method of having the aluminum heat sink made? Anyone?

mxtras
06-02-2005, 02:49 AM
Stamp it. Be done with it. Get some sleep.

I think you are going to kill yourself any other way - you already proved that to yourself, I think. Machining - hold down and efficiency issues. Laser Cutting - too expensive. Casting - uuh....no. Extrusion - no. Praying to God that you wake up and find a thousand of these things under your pillow tomorrow AM?

I think that you should get through your always painfully slow prototyping process and then stamp em out. Until then - did you try machining more than one at a time using the clamp plate? Just curious.

Keep us posted. You have my curiosity now.

Scott

skippy
06-02-2005, 02:51 AM
Hi Mike,
As you can see I put a fair bit of time into that answer thinking that you were a "back-yard-wanna-be" (home brew hopefull??) of which there are plenty in manufacturing, or at least a small operation however that's clearly not the case so sorry about that. I drew my conclusions too fast. Certainly fitting that type of led in an automotive application never been done so no wonder you're excited about it. You'll probably get sued for blinding the drivers behind said bikes! hehehe

Punching for me would be the way to go both on the aluminium heat sink and also the circuit board external shape which I assume will be what holds everything together and also ties it to the body of the light which I assume is plastic.

The only problem I see is that after punching you'll have a jigsaw of pieces and then you must find all the pieces, orientate them and bond them in perfect alignment on the substate. Yes jigs can help but I wonder if it wouldn't be better to punch as suggested above but leave two narrow little bridges on each piece to hold it all together, bond it, place in mill to remove bridges. However that makes one more process so I don't know, just an idea..... I'll think on it a bit today.

lerman
06-02-2005, 03:08 PM
Year (well, actually decades) ago, when I worked at Grumman, they used a process referred to as "chem milling" to "machine" aluminum sheet.

That was a chemical process to remove mass from aluminum.

I would think that you could bond your aluminum sheet to the pc board material, apply resist, and then etch it in place (in much the same way as a pc board is etched) with the proper chemical. My guess is that lye (sodium hydroxide) would work well. That assumes you can find an appropriate resist.

Ken

Scott_bob
06-02-2005, 08:30 PM
Can this even be stamped?
Aren't those outlines 1/2 way thru the material?

If you get a quote for stamping or fine blanking, you'll be shocked...
Granted the cost will be low per piece, after you pay for the die but not for quite awhile.
You're not going to recover your cost (of a die) till after 10k pcs (at least).
IMO, you need higher RPM. Buy a reasonable high speed spindle for your low RPM performance CNC, say NSK pneumatic 40,000 RPM...
Invest in a vacuum system, nest your parts as many as can fit on your vacuum fixture, and yah, run it unattended!
Get some help here on what it might take to automate broken tool detection after every part and let it go... Overnight...

Have fun,

Mortek
06-03-2005, 10:36 PM
We hot glue ceramic strips to a plate for grinding. I think this method would work well for this application. Make a fixture plate out of say 1/2" aluminum. Put on a hot plate to heat. Use a clear glue stick and spread out on fixture plate. Place your .020 sheet on the fixture plate with a weight on it and remove from the hot plate. Allow to cool and then go place it in your mill and cut away. You could make up several fixture plates and be mounting more as you are waiting for one to cut. Put back on the hot plate and remove the finished part. Soak in acetone to remove the glue residue and you have a perfect part. Believe me I was shocked at how well this works. I'm certain it will work for you.

Here's a hot plate like we use for sale on ebay.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=48729&item=7520683236&rd=1&ssPageName=WDVW&tc=photo

You can get larger hot plates

ViperTX
06-04-2005, 12:32 AM
Does Waytek sell these LED taillights??? I know they have a line....

skippy
06-04-2005, 06:53 PM
Mike, I have a fair bit of product manufacturing experience under my belt but I always tend to trust Bob_scott's opinion more than my own when it comes to process/method cost/time advice. If you stick with the present process, the multiple part vac fixture using high speed spindle seems to me to be the logical way to go both in time and cost keeping in mind the relatively small numbers you are talking about. (I mean it's not a 10k run in one hit)

However that still doesn't address the jigsaw puzzle alignment problem you will be faced with at the time of bonding the pieces to the pcb. In my humble opinion, this will be as bigger challenge as the problems already discussed in this thread. I know you said you are working on this in conjunction with the lumiled engineers and I don't profess to know more than you or the lumiled engineers HOWEVER, don't be afraid to think outside the box either. Í my not come up with an answer here but just some thoughts and a bit of encouragement to do some lateral thinking.

I think I know the style of emitter you are working with and if I understand correctly the base of the emitter connects to the heat sink via thermally conductive epoxy. The base of the emitter and the heatsink are not connected to the + or - voltage (i.e. they're isolated). You only need the channels so that the heatsink "will not touch any exposed traces on a circuit board" From this we can deduce that the circuit board is probably a single sided FR4 board with the copper tracks facing the side of the emitter.

So, what would happen if you made your circuit board the other way around, i.e. with the laminate facing the other side (i.e. facing the front of the bike). This would mean that the heat sink could be one piece and wouldn't need channels to be cut (= better heat sinking for the LEDs) and would therefore only need the external shape to be cut plus simple holes (or slots) where the terminals connect to the pcb. A substantially bigger cutter could be used as a result. Next is to bond the heatsink to the fibreglass side of the prepared pcb, finally bond and screw emitters to heatsink/pcb. Or perhaps you were planning to use a double sided pcb? If that's the case, (a) just redraw the board so that it is single sided with some jumpers if need be or (b) use a second small pcb board piggy backed to the first or (c) a double layer board. Personally I'd imagine that option (a) should be possible.
If this method is possible, your pcb will have holes already drilled where the emitter attaching screws will screw to the pcb and your heatsink will have somewhat bigger holes in the same location. (bigger so that the terminals don't make contact with heatsink) You then use the two most opposite holes to fix (+ align) heatsink to pcb during the bonding process. You would use a step shanked screw for this assembly process. In saying that, I wonder if the heatsink has to bonded to the pcb anyway? I'm thinking on the run here. Pcb with 2mm holes for 2mm screws/nuts (sorry I can't talk in fractions of an inch), 3mm holes in heatsink where emitter terminals attach, step shank plastic washers so that current is transmitted from emitter terminal to pcb without making contact with heatsink. These screws also serve to hold heatsink to pcb.

EDIT Another variation on the above would be: instead of starting off with the boomerang looking piece cut from sheet aluminium, get it die cast in alloy complete with the terminal mounting holes already in place, plus a slight recess to position/accept the body of the emitter (= faster assembly time) and possibly a little pin at each end of the boomerang to line up with corresponding holes in the pcb to make for faster positioning during assembly with pcb. I'd also imagine that the board doesn't have wall to wall emitters so you therefore use up the left over real estate with traditional cooling fins to once again increase cooling.

EDIT2 Up until now, I've (we've?) been assuming that the pcb is the central piece of this project because the emitters and heatsink/s are fitted to it and it's the pcb that is screwed to the body of the light. (i.e. pcb acts as the mounting plate or fixture). I may be wrong but this is what I've been assuming. Now, following on with the one piece die cast heatsink idea. What if the heatsink becomes the central piece? i.e. the heatsink is what mounts to the plastic body of the light and also holds the pcb and emitters?

The heatsink would have all the previous features plus the holes or tangs or whatever is used to fit to the plastic body of the light. The same step shank plastic washer system would be used for the terminals of the emitter HOWEVER the pcb would use stand-offs and if a double sided pcb is to be used, it would depend on which side of the pcb you put your plastic or metal washers. The added advantage of this is the +/-2mm air gap between the heatsink and the pcb which will increase cooling even further. My whole reasoning here is based on my assumption that maybe not all, but certainly the majority of heat is dissipated from this type of emitter via the flat underside (of the emitter). If this is not correct, please let me know and I'm sure we can still come up with something.

EDIT3 Another thought occurred to me during the night, if for whatever reason the one piece cast heatsink isn't possible, what about if you mill a heatsink as per your picture BUT mill it from the pcb side not the emitter side and you cut the external shape, holes, and slots to full depth but only cut the tracks half way through using a bigger cutter (= faster). This would keep it as one piece (= faster = cheaper) and because the tracks are wider, alignment so that heat sink doesn't touch pcb tracks becomes a non-issue. Even if you increased the material thickness slightly so that milling half way through was easier, this method would eliminate the jigsaw problem plus greatly improve the heatsinking ability of the system. (in case you don't understand that statement, say you have 3 +3 blinkers, 5 for the stop/tail light = 11 emitters in total. If stop light is activated, 5 emitters will be dissipating heat to the total area of the heatsink instead of each just having it's own little heatsink.) Using any of the above methods makes it unnecessary to use MCPCB which you may or may not have been thinking about using.

Well what do you think? As you can see, this one's had me thinking and as I said earlier, maybe none of these will be the answer but I'm at least trying to broaden the possibilities.

gar
06-05-2005, 08:23 AM
050605-0712 EST USA

fastturbovet:

Copper has about twice the thermal conductivity of aluminum. There is also thermal resistance in the bond from the pc board copper to the aluminum heatsink.

In any event if copper was used for the heatsink material, then you would need about 0.010" thickness.

You might investigate the pc board processing and see if the vendor could copper plate 0.010" copper in the heatsink areas. If so it might solve many problems.

.

skippy
06-05-2005, 11:36 AM
"Copper has about twice the thermal conductivity of aluminum." True

"There is also thermal resistance in the bond from the pc board copper to the aluminum heatsink." If Mike chooses the heatsink bonded to the circuit board route, I'd imagine he'll be using thermally conductive epoxy which although it does have some thermal resistance, overall is pretty conductive.

"You might investigate the pc board processing and see if the vendor could copper plate 0.010" copper in the heatsink areas. If so it might solve many problems" I still maintain that if he separates heatsink from circuit board with an air gap of say 2mm, it will mean that the heatsink will do its job to get rid of heat (and more efficiently because both sides of the heat sink will be exposed to air) and the expensive circuit board and microcontroller can be kept away from the heat source.

Gar, I think your idea of plating the pcb is a good idea but I'm just saying that maybe a bit of lateral thinking is required.

p.s. I looks at though Dave's not the only one that has left the room. Mike has too!

fastturbovet
06-07-2005, 12:12 AM
"Copper has about twice the thermal conductivity of aluminum." True

"There is also thermal resistance in the bond from the pc board copper to the aluminum heatsink." If Mike chooses the heatsink bonded to the circuit board route, I'd imagine he'll be using thermally conductive epoxy which although it does have some thermal resistance, overall is pretty conductive.

"You might investigate the pc board processing and see if the vendor could copper plate 0.010" copper in the heatsink areas. If so it might solve many problems" I still maintain that if he separates heatsink from circuit board with an air gap of say 2mm, it will mean that the heatsink will do its job to get rid of heat (and more efficiently because both sides of the heat sink will be exposed to air) and the expensive circuit board and microcontroller can be kept away from the heat source.

Gar, I think your idea of plating the pcb is a good idea but I'm just saying that maybe a bit of lateral thinking is required.

p.s. I looks at though Dave's not the only one that has left the room. Mike has too!

I'm still here.

I appreciate the thoughts and ideas on how to do the product, but really, I am already decided on how it will be made. The problem is that there are many many design considerations involved here, which include the equipment we have available, cost of parts, fabrication time, packaging constraints, LED specifications, PCB packaging requirements, external component setup and many other things. I have spent quite a lot of time on the design of this light, along with lots of other folks and this is the best solution. I can't go into explicit detail on each aspect of the design, partly due to it being proprietary info, but also because I am not looking to design a new product as a group - it's already designed and it will be made in the way it has already been designed. I sincerely appreciate the input, but really I am not looking for a way to design the product - that part is done, what I need is the best solution to produce the metal plate pictured. The LED's being used are not the type that has a thermal pad on the bottom, and the heat sink is a conductor for the power and ground in addition to being a heat sink. There are so many different aspects of the design, including the temperature range, installation requirements, vibration resistance, LED cost, other parts cost, cost of consumable materials and so many other factors that approaching the design from a different standpoint is really not helpful and I can't give out every piece of data that led us to this design, but suffice it to say there are a lot of smart and knowledgeable people involved and this design is the best way to go, given the parameters at hand.

I hope that doesn't sound rude - I really appreciate the enthusiasm from folks about this project, but the critical thing is quickly producing the part at hand.

I have had pretty good luck machining them on our CNC mill. I have played around with feeds/speeds and gone from a HSS bit to a solid carbide with good luck. I am cutting dry, and using double-sided tape to hold the AL sheet to a piece of dense/hard 2" foam which I clamp in the vice. This is workign well. The hardest part is that it's a PITA to remove the tape after the cut, and it needs a new piece of foam cut every time and mounted.

Thanks to the previous poster for the hot glue suggestion - I am going to give that a shot. I have tried spray-on contact adhesive but it seemed to gum up the bit and cause it to heat up and break. The tape is messy but isn't causing breakages. I think the hot glue idea may be the best so far - any other ideas on holding down the piece? I have looked into stamping, chemical etching, laser cutting, waterjet and blanking and they are all more expensive over the expected product lifespan. Problem is, since its a motorcylce part, they change so quickly that I would never get the ROI from stamping, and laser/waterjet is too $$ per part compared to milling. I can run the mill unattended and fit 9 of them on the table at once, run if 3-4 times a day if need be, once overnight and do 40-50 a day no problem. I need about 100 a month to start so no worries.

Scott_bob
06-07-2005, 10:16 AM
An alternative clamping method...
I have used this and it does work well, it's just a bit messy like the double stick tape.

Well worth a try though, a lot less work. Just heat, cool and you're ready to cut, then heat again to melt the adhesive. Works well.

http://www.miteebite.com/
(then click to the mitee-grip)

Good luck,

skippy
06-07-2005, 12:29 PM
Hi Mike,
Thanks for filling us in a bit on what's what and no it wasn't rude in the least and yes we all realise that one can't publicise every little detail about a product otherwise someone else will be making it before you do.

I now understand why the heatsink can't be made in one piece ("the heat sink is a conductor for the power and ground in addition to being a heat sink")

One thing you didn't answer is the question about how you are going to take all the jigsaw pieces and bond them to the pcb with correct alignment. Sounds time consuming even with a jig? Did you give any thought to (at least I think I mentioned this idea earlier) cutting the outside shape of the aluminium with perhaps two holes that coincide with the pcb, bond ali to pcb using your thermally conductive epoxy (if that's what you're going to use), then mill out the grooves of the heatsink. This would solve a lot of problems. It would be very easy to do if you are using single side FR4 and yes you may have to true the squareness of your mill first but even if you are planning on using a double sided pcb I still think this would work as you'll have the thickness of the epoxy as a safety margin. I've milled to this tolerance previously simply by making the multiple part, vac hold down set up discussed earlier in the thread.

Failing that, as mxtras says: post-prototype, stamp it and get some sleep

Your mates here at cnczone only ask one favour, send us a post card from the Carribean Islands when you get rich and retire there ok? hehehe

JavaDog
06-07-2005, 12:33 PM
suffice it to say there are a lot of smart and knowledgeable people involved and this design is the best way to go

Too bad they didn't figure in the manufacturing issues into the design...

Mortek, the hot-glue idea is slick. Going to have to remember that one myself! Have you ever had the heat from machining re-melt the part though?

Mortek
06-07-2005, 10:13 PM
nope, always using coolant though. That should stop it on the thin sheet.

fastturbovet
06-09-2005, 02:46 AM
Hi Mike,
Thanks for filling us in a bit on what's what and no it wasn't rude in the least and yes we all realise that one can't publicise every little detail about a product otherwise someone else will be making it before you do.

I now understand why the heatsink can't be made in one piece ("the heat sink is a conductor for the power and ground in addition to being a heat sink")

One thing you didn't answer is the question about how you are going to take all the jigsaw pieces and bond them to the pcb with correct alignment. Sounds time consuming even with a jig? Did you give any thought to (at least I think I mentioned this idea earlier) cutting the outside shape of the aluminium with perhaps two holes that coincide with the pcb, bond ali to pcb using your thermally conductive epoxy (if that's what you're going to use), then mill out the grooves of the heatsink. This would solve a lot of problems. It would be very easy to do if you are using single side FR4 and yes you may have to true the squareness of your mill first but even if you are planning on using a double sided pcb I still think this would work as you'll have the thickness of the epoxy as a safety margin. I've milled to this tolerance previously simply by making the multiple part, vac hold down set up discussed earlier in the thread.

Failing that, as mxtras says: post-prototype, stamp it and get some sleep

Your mates here at cnczone only ask one favour, send us a post card from the Carribean Islands when you get rich and retire there ok? hehehe

You're right! Aligning them was a bit of a PITA :)

What I have done is all those little parts you see in the pic I posted have "ears" that connect them. Basically little half-circles that connect each segment to the one on either side of it. As you can see the part has holes (and so does the PCB). So I have made a jig which is just a sheet that I can place a bunch of the PCB's on with pins extending through the holes. I lay the PCB's on it, then I put it in our auto-dispensing machine which splurts out conductive epoxy, then each heat-sink sheet is layed down on top over the pins which precisely aligns it on the PCB. I put another sheet on top which is basically a weight and put it in the over to set the epoxy. Then it goes back on the dispenser which puts more epoxy on to bond the LED's.

CHeck out this pic

http://www.hightechspeed.com/warrior1.jpg

I control each LED individually so they cannot be connected together. The OEM of the LED's provides a punch/crimp method for attachment, but the machine to do this is many many tens of thousands of dollars, and since it deforms the metal mount, is not suitable for our use (since we need it to be flat for the bond procedure). You can see in the picture the remnants of the metal "ears" that connect each segment. This is a rough prototype so they were hand-cut to test this prototype board. In the actual implementation, after the LED's are bonded and heated to set the glue, the fixture plate is re-mounted in the CNC and the ears are milled off, producing the board.

hope this clarifies what the application is!

The suggestions on machining the sheet have been excellent, I bought a hot-plate and will give the glue method a shot!

fastturbovet
06-09-2005, 02:53 AM
Too bad they didn't figure in the manufacturing issues into the design...


Actually, they did. If you have a better idea, I am all ears.... but if you don't understand the application, how can you say that the manufacturing issues were not addressed? The question was not "hey can anyone tell me how to make this product" but rather it's already been spec'd and the task at hand is the best method to cut the board. I can run a batch of these now - totally unattended. My issue has been holding down the work correctly. Right now I use double sided tape which works, but is a pain to clean off (although I can just soak in acetone and it's unattended).

I have the tools I have at my disposal. The constraints are that it must be cost effective as my final price is already determined because of my contracts with distributors. There are many ways to solve the problem - using a different LED would be one way, but there are also constraints on the lens which requires discrete point sources of light. I can use something like a 1W+ emitter but the price isn't there and they output too much light for what I need. I know the electronics side of it *very* well and there is nothing that compares in price/performance to what we're using. Production is more difficult than our other products, but at least its mostly automated, so the hands-on time for production is less than most of our stuff.

If my design does not take the manufacturing issues into account, I am sure you could do it better and kick my butt in the marketplace! We sell many thousands of these products a year, so the money is there... and with a superior design at a better price point, I couldn't compete. I'll wait to see your superior design in stores around the USA, yes? :)

No flame intended, but it's hella presumptuous to think we're idiots who didn't think of what it takes to make the product (since the cost to manufacture comes outta my own pocket)

fastturbovet
06-09-2005, 03:10 AM
Your mates here at cnczone only ask one favour, send us a post card from the Carribean Islands when you get rich and retire there ok? hehehe

Skippy,

Well, working for yourself is all about being rich right? I know when we (my and my biz partner) started this company about 5 years ago, we instantly became right and have been trying to find the time to spend those millions! :) Haha... well, maybe not. What are you doing in terms of CNC/electronics? I would be happy to give you lots of details on what we do, if you're interested, didn't mean to sound snippy prior, just I have a deadline and need to get this cranked out. I could talk electronics/software/machining all day long! We started back in '99 with an idea for a bike part... I sold all my "toys" I had accumulated over the years, rented office space and got to work. First couple of years were VERY hard... but it paid off. If I had a buck of every time I sat in a cheap hotel room at some trade show in some city thousands of miles from home after working a 15 hour day (the third, fourth or fifth in a row!) I'd be rich already. We stuck it out for the first couple of years and the business grew and grew. Now the problem has been finding the time to do production as well as finding time to get new ideas off the ground. I don't have to worry about paying the bills anymore (thank god) but it's still a challenge finding time to work on new stuff.

I can say the BIGGEST factor in our success has been being resourceful about getting the job done. You'd probably laugh if you saw some of the machines we make parts on... been replacing it with more high end stuff but I'm proud of what we accomplished. Hell, I have a pick-and-place machine we paid $800 for and retrofitted our own CNC control on, the Bridgeport (which I posted about on here) that I got for $1300 and my partner built a whole CNC control for it, the epoxy dispenser that I machined a mount for a rotozip to make it into a CNC router, and on and on. Its fun though and sure as heck beats working for "the man" :)

skippy
06-09-2005, 03:44 AM
Hi fasturbovet, unless I'm mistaken Javadog wasn't criticising you but instead saying that the led manufacturer sure is making you work for it due to their design.

The "ears" system above is what I was talking about when I said "I wonder if it wouldn't be better to punch (or mill) as suggested above but leave two narrow little bridges on each piece to hold it all together, bond it, place in mill to remove bridges. However that makes one more process so..... "

"Well, working for yourself is all about being rich right?" No, I was just joking. For me, the greatest "buzz" is the challenge, particularly of developing a new product, then seeing it go into production, and then seeing that product in use (in your case on someone's bike). I have always believed that it's the love of doing it that comes first (no matter if you're selling real estate, manufacturing something or providing a service) and from there success is almost guaranteed and the money is a natural follow on from success.

Neat product, all the best with it and I'll pm you re the other.
p.s. someone should start a thread about where each person's avitar came from. I'm sure there'd be some interesting stories. Yours, well that's obvious but howboutapicy, please?

JavaDog
06-09-2005, 06:55 AM
Hi fasturbovet, unless I'm mistaken Javadog wasn't criticising you but instead saying that the led manufacturer sure is making you work for it due to their design.

Glad someone understood that. Just seems they they don't make that particular product easy to integrate...or maybe it was just the way it needed to be integrated in your design that made it tricky. Either way, good to see you worked it out.

FYI, if I had wanted to call you an idiot - I would have. Honestly, I hope I am half as succesful as it sounds like you guys are when I start getting my ideas off the ground. Kudos for making it work.

fastturbovet
06-10-2005, 12:42 AM
"Well, working for yourself is all about being rich right?" No, I was just joking. For me, the greatest "buzz" is the challenge, particularly of developing a new product, then seeing it go into production, and then seeing that product in use (in your case on someone's bike). I have always believed that it's the love of doing it that comes first (no matter if you're selling real estate, manufacturing something or providing a service) and from there success is almost guaranteed and the money is a natural follow on from success.

Neat product, all the best with it and I'll pm you re the other.
p.s. someone should start a thread about where each person's avitar came from. I'm sure there'd be some interesting stories. Yours, well that's obvious but howboutapicy, please?

I was just joking about the being rich part... I laugh about it because everyone who hears you started your own outfit thinks "wow you must be rich", but if only they knew :) As we know, we are the LAST ones to get paid! At least I got a chance to see the world (through the murky windows of cheap hotel rooms and 5 hours of sleep a night) :)

Your idea of the mounting ears is a good one, I can say its working well on the finished product. Skippy I will post another pic in a minute so you can see the brightness level. Are you familiar with the type of LED on that board? They are desigend for automotive use but not to be used in the way we're using them... when I first talked to Ken at Lumileds he thought I was nuts, but over the past few months they have gotten pretty interested in this particular usage and its working great. The biggest hurdle was "how do you take and LED that's designed to be crimped onto a substrate in groups in a no-heat forming process, and use it on a PCB board where everything is reflowed". But I enjoy the challenge as well, so I am usually having fun doing this stuff.

Check out the pic I'll post - the LEDs are *bright*. P.S. If you ever need any LED's, let me know. I buy a bunch of 'em and I have really good relationships with most of the manufacturers. You can even get the Lumileds 24lm emitter for $0.85/ea now! They have a new white emitter coming soon for headlight applications. Confederate motorcycles is incorporating it into their 2005 MY headlights - cool stuff!

fastturbovet
06-10-2005, 12:48 AM
Glad someone understood that. Just seems they they don't make that particular product easy to integrate...or maybe it was just the way it needed to be integrated in your design that made it tricky. Either way, good to see you worked it out.

FYI, if I had wanted to call you an idiot - I would have. Honestly, I hope I am half as succesful as it sounds like you guys are when I start getting my ideas off the ground. Kudos for making it work.

Sorry if I came back a little harsh, you are right, this particular LED is not originally designed to be used this way. The problem with high-power LED's is they decrease the thermal resistance to the LED die which means they run cooler for a given amount of heat and conduct it away from the die very efficiently, but it also means they either can't be soldered or they require big heat-sinks to a non-conducting pad on the LED. THis particular one was designed for mounting to a metal board in a crimp/punch process but it didn't suit our needs. Our options were super-expensive LED's that were too bright, or clusters of cheaper ones. The clusters screw up the optics in the light and don't end up being cheap when you add them up, so our challenge was to take an LED with a very good cost/brightness ratio but use it in a way other than it was designed for. If we could solve that challenge, we could overcome our biggest obstacle the past year, which was cost, brightness, and function. That's what this product is designed to do, and it was a major PITA to get it all working right... but with a lot of trial and error, haggling, and the resourcefulness of some great minds it was able to be done, and I am saving money over my previous cost on lesser-brightness products. The only problem is the time it takes to cut these aluminum bits, and the fixturing of them. Time is not a huge big deal, I can run 'em and walk away. The fixturing is not a huge deal, and the suggestions here have helped a ton.

It would be a good idea to start a business forum on this site, I think, for discussion of making money with your CNC setup.

Mortek
06-14-2005, 11:33 PM
By the way, that hot glue isn't the craft glue that you buy for your glue gun. You can find it at this address

http://www.aremco.com/a9.html

We use the stick crystalbond 509

Hope you haven't tried the craft glue already.