Heat Transfer Fluid distributors in Europe?

[Kremmen]

I am in the process of designing an FPGA based device where the chip comes in a 256 ball BGA package. Soldering those can be a bitch and I decided to try vapor phase process to maximize the odds for success. I hit an unexpected snag however - can't find anyone selling the heat transfer fluids needed for the job. Galden at least is a mfg and specs are all over the place, but just try to find a reseller for the stuff?!
So my question is, if anyone has found a seller for this or any other brand of transfer fluid in the soldering heat range (~220 -240 deg C)? Any hints gratefully accepted.

[jahonen]

Have you tried asking Perel, they seem to have the stuff you are asking?

http://www.perel.fi/osastot/factoryautomation/materiaalit_galden_nesteet.htm

Regards,
Janne

[Kremmen]

I guess i will ask them. While the company comes up in a google search, i couldn't find the actual product on their website. Maybe i have missed it repeatedly...

[mikeselectricstuff]

I suspect you may find a fairly high minimum order qty, and this stuff is not cheap.
 
As long as the solderpaste print is good, and placement is close enough first time (no dragging around!), the soldering is the least critical process.

Bear in mind that a 256BGA will usually be on at least a 4-layer PCBs, and the inner planes will help keep the temperature distribution even over the area of the device even if the oven isn't very uniform.

Let us know if you get a price - I did ask for pricing a while ago - don't recall the price or MOQ but it did definitely steer me away for VPR pretty much immediately.     

[JoannaK]

It's been some years since, but IIRC:  Those FPGA-routings + via placements are quite devious stuff. I'd say that at a ballpark you'll get max 2 rows per layer routed out from the chip. 256 BGA (that's 8*8 balls) you'll need 4 layers full of copper traces for IO lines and at least 2 layers just for Power/Gnd planes.

I don't really think that kind of soldering is the best thing to do at hobby-home-owen. Not that it might not work (eventually), but there's very little margin for errors and testing bad solder/shorts etc. without proper tools (Xrays + automated machies) will be next to impossible. So I would suggest asking some specialized company to do the soldering. Afterall, 4+layer board have to be professionally made anyhow.

That said, I have not asked the price for a small run, nor have I asked if those companies are willing to take work from someone individual (not a ocmpany).

[mikeselectricstuff]

That said, I have not asked the price for a small run, nor have I asked if those companies are willing to take work from someone individual (not a ocmpany).

The  problem is not who they will deal with  but finding someone oriented towards smaller jobs.
One issue is most production places are oriented to making stuff in big panels, so you need to pay for a large, usually framed stencil (GBP150-200) before they can do anything at all.
If you're lucky you may find someone with a  Mydata solderpaste jet printer or precision dispenser which would avoid the stencil cost.
It may be easier and cheaper to just order a few spare boards and chips... 

[Kremmen]

Xilinx has put some thought into the placement of signals & power + gnd in the BGA array. 2 rows per layer is more or less right, but for this 256 ball array, there is just enough space, and I have made an initial escapement routing on a 4 layer board with just 2 signal layers. It does require 0.15 mm tracks and clearances but that shouldn't be a problem.
The vapor phase process was specifically to raise the odds for a successful solder. That way heat transfer is supposed to be efficient while avoiding the risk of overheating. The boiling point in atmospheric pressure of the fluid ensures the vapor temp and makes overheating impossible.
But i am also starting to worry about the price of the fluid. It has been curiously difficult to find prices anywhere and when i finally managed to find one, my first impression was that the decimal point must have slipped a couple digits. So i'll have to see if that is an option after all.
The boards themselves will certainly be made by the mfg i normally use. Of course i can have the boards assembled aswell but then we are much deeper into calculating production runs in detail and that was not my original plan. But we'll see.
Thanks for all your replies, they have been useful. If i get a price i'll let you know.

[mikeselectricstuff]

Just found this price :
http://www.oritech.com.au/productDetail.aspx?productID=40254
5kg about US$1330.
A problem is there is only one manufacturer so they can pretty much charge what they like.

[amyk]

Just found this price :
http://www.oritech.com.au/productDetail.aspx?productID=40254
5kg about US$1330.
A problem is there is only one manufacturer so they can pretty much charge what they like.

The MSDS for that states that it's perfluorinated polyether (also looks to be rather inert and nontoxic), and if you search the CAS number you can find that it's also being sold as vacuum pump lube. Not any cheaper though, the price seems to be around US$200/kg.

[mikeselectricstuff]

Just found this price :
http://www.oritech.com.au/productDetail.aspx?productID=40254
5kg about US$1330.
A problem is there is only one manufacturer so they can pretty much charge what they like.

The MSDS for that states that it's perfluorinated polyether (also looks to be rather inert and nontoxic), and if you search the CAS number you can find that it's also being sold as vacuum pump lube. Not any cheaper though, the price seems to be around US$200/kg.

I think the problem will be finding someone that will sell you less than 5kg.

[SeanB]

And of course the carbon tax on it doubles the price........

[Kremmen]

Just found this price :
http://www.oritech.com.au/productDetail.aspx?productID=40254
5kg about US$1330.
A problem is there is only one manufacturer so they can pretty much charge what they like.

Yeah, i found a similar price indication somewhere. A nice idea but maybe not at that price level. You probably con't need very much fluid per run, but there will inevitably be some losses so looks like this will be too expensive to consider. A pity, really because it would have been easy to do while maximizing the odds. Oh well, back to the drawing board...

[mikeselectricstuff]

Just found this price :
http://www.oritech.com.au/productDetail.aspx?productID=40254
5kg about US$1330.
A problem is there is only one manufacturer so they can pretty much charge what they like.

Yeah, i found a similar price indication somewhere. A nice idea but maybe not at that price level. You probably con't need very much fluid per run, but there will inevitably be some losses so looks like this will be too expensive to consider. A pity, really because it would have been easy to do while maximizing the odds. Oh well, back to the drawing board...

As long as you stick to leaded paste, a pizza oven isn't nearly as scary as you'd think.

[Kremmen]

Oh i use one all the time, and leaded paste as well. Some time ago already, i modified a pizza oven by moving the heaters all on top (that way the first side components do not drop off), added insulation, a convection fan and most importantly, a sequencing kiln controller + contact thermocouple for ramp control. It works beautifully for all "regular" smd stuff, but haven't tried it for BGA yet.

[mikeselectricstuff]

Oh i use one all the time, and leaded paste as well. Some time ago already, i modified a pizza oven by moving the heaters all on top (that way the first side components do not drop off)

Surface tension should hold bottom-side parts on. If heating from one side stops this then you probably have a pretty big thermal gradient, which is probably going to be bad. The whole PCB wants to be at pretty much the same temp throughout, being heated mostly by convection - too much IR will be absorbed unevenly and can cause scorching.

[Kremmen]

Surface tension is iffy for larger smd components. Some smd transformers i have used will definitely drop off. Of course it would be usually possible to arrange them on the second side. Anyway, i don't want to reflow the already soldered components. Also, the thermal gradient will be not that great since the board will be soaked to more or less uniform temp before going for liquidus on the top side. The sequencer controller will take the soak temp arbitrarily close and then peak for the time i want. I have used the Altera temp profile which has worked beautifully. So the effect will be to just avoid reflowing the bottom side. The fan is there to provide convection and avoid hot spots on the top side. This system has produced zero failed boards so far. At least due to temp gradient induced solder failures or cracks in ceramics etc. I just haven't tried it with a BGA.