3070's melted.

captainmorgan

Well-Known Member
How many people check their heatsinks for flatness. All the heatsinks I've bought are concave but within industry standards. I sand all mine flat but I'm not sure everyone is checking for this. I could see uneven pressure from a holder over a concave surface putting to much stress on a Cree COB.
 

Airwalker16

Well-Known Member
I have put together literally 1000's of cxb's and hundreds of cxa's(old substrate)...as well as dropped them, tossed them, crammed them in drawers, stepped on them, poked them with screwdrivers, sprayed them with foliar, hot plugged them...just about everything you are not supposed to do to them. Most of them with pads I might add. Never once have I cracked one, "blown up" one, or damaged a bond wire.

"thicker" pads do not mean thick...it means thicker than before.

Pads come right off like a perfect sticker. They are not fragile, and can be reused. I have a bunch on personal projects.

Cree actually said that they prefer user to use plastic holder vs metral ideal...because of the possibility of over torquing(too much pressure) is easy with them. I've never had it happen..and screwed them down pretty tight...but could see it happening.

I had one cxb3070 DOA in a 500 pack one time. Cree took care of me and then some.
Well

@greengenes07 if youd make bjbs for 3590's, id be using them.
 

bicit

Well-Known Member
I brought these for Vero 13. Was rubber like and broke during application. After that I decided to stick with thermal paste.
http://www.digikey.com/product-detail/en/TG6050-10-10-1/1168-1568-ND/3042380
http://www.digikey.com/product-detail/en/TG6050-10-10-0.5/1168-1559-ND/3042372

Any recommendations for vero 29 or 3590 pads?

@bicit 0.07g. What thermal paste
MX-4, and that's aiming for a .05mm thick layer of TIM.

If you want to use a different paste you'll need to find the density of your desired TIM. It's usually listed on the package or on the manufactures website in units of grams per cm³ (g/cm³). MX-4 Happens to be 2.5 g/cm³. After that find the surface area of the chip you're working with(presumably in mm²), multiply by the desired layer of TIM, and you'll have the approximate volume of TIM needed in mm³. Convert to cm³ and multiply by the density of your TIM. You'll have the approximate amount of thermal paste needed in grams. So for example.

1: MX-4 is 2.5 g/cm³
2: A cxb 3590 is (34.85mm)x(34.85mm) or 1214.52mm²
3: (1214.52mm²)x(.05mm)=60.72mm³ or .06 cm³ of thermal paste needed.
4: (.06cm³)x(2.5 g/cm³)= 0.15 grams of TIM

Looks like I slightly understated my original values :oops:. In all reality if your heat sink is reasonably flat you don't need much TIM.
 
Last edited:

Greengenes707

Well-Known Member
I don't make BJB's...BJB does. And they do not make 3590 products. If they made 3590 holders...I would have them.

I do have 3590 thermal pads in route. So within the week PLC will have the same sur-seal pads they have for the 3590's. They perform and use really well compared to what I have used/tried. I used berquist for a while, but were even pricier and didn't have a real edge in performance.

Grease is great, just messy and time consuming. I don't like to do 100's of cobs and let them sit before mounting. WIth grease I lather and mount as one process. Instead of with pads being able to pad up all your cobs then go to mounting. On a small scale/DIY, that probably doesn't matter to anyone.
 

superbak3d

Well-Known Member
How many people check their heatsinks for flatness. All the heatsinks I've bought are concave but within industry standards. I sand all mine flat but I'm not sure everyone is checking for this. I could see uneven pressure from a holder over a concave surface putting to much stress on a Cree COB.
I sand my heatsinks whenever applying new paste. Reduces the chances of air bubbles when the paste spreads from pressure.
 

Airwalker16

Well-Known Member
I don't make BJB's...BJB does. And they do not make 3590 products. If they made 3590 holders...I would have them.

I do have 3590 thermal pads in route. So within the week PLC will have the same sur-seal pads they have for the 3590's. They perform and use really well compared to what I have used/tried. I used berquist for a while, but were even pricier and didn't have a real edge in performance.

Grease is great, just messy and time consuming. I don't like to do 100's of cobs and let them sit before mounting. WIth grease I lather and mount as one process. Instead of with pads being able to pad up all your cobs then go to mounting. On a small scale/DIY, that probably doesn't matter to anyone.
OH dude, SWEEEEET. Pads for the 3590's?? Im all for it. I'll be buying some FOR CERTAIN.
and yes, I'm aware you don't make the BJB's. But considering how popular the 3590's are now days, I'd assume they'll start manufacturing them sooner or later?
 

Abiqua

Well-Known Member
It's passive. 3 @ 2100mA on a 48" heatsink.
I don't mean to harp on you personally, but here is where DIY truly exists...

Did you actually do any thermal calculations before and after running the cobs initially....I know these calculations can be difficult intially to figure out terms....but this is a real aspect of DIY .

I know that everyone likes to think that DIY is as easy as screwing some bolts together and figuring out your vF, but the real DIY'ers busted their asses in past posts to outline alot of these critical components and how and why and when they will work together ......that is the beauty of DIY, the ability to build yourself, but some things are skipped over repeatedly by some of the "newer" generation of Cob users....and unfortunately I chalk this up right now to operator error.

Another is building safety systems.....Even a one way temp fuse would have probably stopped the problem.......maybe even a relay controlled resetable switch thermostat that will cut power once heat gets to a certain degree....

Something to keep in mind when going forward....there is some real critical parameters than each DIY builder has to take into account for their own setups, regardless of how much beaten' to death a particular build has been done.....:peace:
 

grouch

Well-Known Member
When you tapped the screws into the heatsink, did you take the screws back out and sand the heatsink flat again?

When I built mine the heatsink bulged out a little around the screw holes from the screws displacing the aluminum.
 

Abiqua

Well-Known Member
Well, the heat gun read like 101*F on the top if heatsinks. I will probably throw active cooling on it eventually
That's a start, but as captain eleborated, laser guns get funny readings from shiny metal....Do you remember what the grow room ambient was too, ballpark?


If you can research later and get some of these numbers, in theory you could ballpark a little closer, especially if you match with Supra's vF rating .... but you really need to be able to have a second # to complete an equation side...... a TC measurement would be nice :)

Lets start with basic equation where most assume radiated black body..[this is important because we know diode energy isn't radiated immediately and some suggest increasing Rth [heatsink resistance] by 20% or more in some cases....more on that later....


Lets get most of the equations out of the way....



TC
Case Temp
Tj
Case to Junction
Rj-c

Resistance of Junction to Case [Diode datasheet]

Pd
Power dissapation
[Efficency x Wattage]
Tb
Heatsink Temp
Rh
Heatsink Resistance
Rb
Thermal material Resistance

TC = Tj - [Rj-c x Pd]

Rh = [Tb-Ta]/ Pd

Tb = Tc - [Rb x Pd]


So if you had the heatsink temp [Tb] and a Tc measurement, then you could at least find the resistance of your Heatsink [Rh] and figure out if your heat is rising too much per watt....but you would need to quickly consult the datasheet for your Rb [graphite sheet? resistance]
 
Top