Quote from: RHA on November 30, 2013, 07:30:23 PM

Quote from: vs3 on November 30, 2013, 09:23:11 AM

Quote from: WynX on November 28, 2013, 06:07:32 PM

Yes, heat is going to be substantial, but 50W is not too crazy for a chip with a large heatsink. Pictures of testing heatsinks have been posted as Carlton is mentioning.

On the first question - you probably haven't looked at the specs. (...)

As for the heatsink - I'm not talking about what they plan to put in their retail products. The question is specifically to the chip's packaging. The final product design is a consequence of the chip design.
If the chip has exposed silicon - they you'll see heatsinks on top of the chip (e.g. like BFL). If the chip is meant the send all the heat via the PCB - then you'll see a large heatsink on the back of the PCB (e.g. like Avalon and Bitfury).
So - the question is which of the two options have they picked (and if that's been decided already).

You know very well they use FCBGA packaging. You specifically asked them if they could switch to QFN:

Quote from: vs3 on November 28, 2013, 01:10:59 AM

After taking a quick look at the specs - I have some questions (and recommendations) for the development team:

Do you HAVE TO use a FCBGA and 700 pins? That's quite a pain for troubleshooting and not to mention much more expensive for in the manufacturing process.

If it isn't too late - could you consider a QFN (or[T]QFP or something else non-BGA)?

Do not ask to switch to QFN packaging just before tapeout and for a 50W chip.
Such a proposition clearly shows you either have quite small knowledge about chips or you try to troll the topic.

RHA - had you bothered to look at my profile and previous posts you might have gotten a better idea of my knowledge about chips and my desire to engage in pointless trolling.
And since you appear to be much more in-the-know about their chip - would you mind to please post for me the link where they said they've that tape-out already happened or when is it going to happen? (aside from other people's previous guess that it might happen in mid-Dec, which given their hesitation to switch designs 20/28nm that isn't quite clear either)
And as for the switching part - depending on how far they are in the process it might be just the right time to ask for that which is exactly what I asked (or to quote for the reading-challenged: "If it isn't too late - could you consider a QFN"). So an answer of "it's late" would've been just fine - and note that it only matters if it comes form them (unless you've been appointed their mouthpiece of which I could not find any indication).

Now, without trying to insult you or anyone else - instead of lecturing me "not to ask to switch packaging" you should've asked why I brought the question in the first place.

The simple reason is that FCBGA packaging is much much more expensive in the manufacturing process than QFN. I'll give you one example - with the QFN/bitfury chip I made a design that can use the cheapest possible 2-layer PCB, AND on top of that the assembly house said they can use their old/cheap machines for the process. Final result - very cheap to manufacture stuff with it. Now compared to a FCBGA - you HAVE TO use a much more expensive PCB (and sometimes you consider yourself lucky if 4-layer could do), thermal and thickness requirements for the PCB are much stricter (as if PCB starts bending and twisting when heated that will break the solder balls - a common "disease" for cell phones for example) and then your assembly factory would have to use a more expensive process to position and inspect the boards (they may have to literally X-ray each board to check the solders).

On top of all that is the HEAT issue. Don't get me wrong - I wasn't happy with Avalon and Bitfury's choice to send all the heat on the other side of the PCB. However, QFN is the cheapest packaging option, so I understand why they picked that route (or in other words - that gives them very cheap chips). The other option is to expose the silicon (like what BFL did) which lets you easily attach a proper heatsink. Now, the downside is that "attach" word - you're almost guaranteed to bend the PCB in that process, so you have to consider that when making the PCB design (just look at PC motherboards and the CPU assembly) and you'll have to make sure that stuff doesn't break in that process (because you can literally break the PCB or some components - like the case of I think KNC's exploding capacitors).

So, in my opinion the disadvantages of QFN outweigh the advantages of a BGA packaging (and it will make it simpler to use - remember the KISS Principle?). And for the record - I don't insist to be QFN - any other non-BGA can also be considered for the same reasons.

I'd appreciate if BlackArrow could chime in on the question whether it is too late to consider QFN (or any other non-BGA) packaging.