"It's not hard, you just need hundreds of dollars in equipment and some practice!" Sounds like "hard" to me.

TQFP can be done with common and cheap items like a cheap soldering iron, a flux pen, some copper wick and a pair of tweezers.

C'mon now, anyone seriously into electronics --even as a hobby-- ought to own more than a cheap soldering iron, a flux pen, copper wick and a pair of tweezers. A toaster oven is, what, $30? Yes, you have to get a prototype stencil made. These are also cheap. Google it.

Here's someone who wrote about the proposed approach: http://www.instructables.com/id/Toaster-Oven-Reflow-Solderin...

Yup, flux is great, under appreciated stuff. First you apply the flux to the area you'll be soldering, then you put just enough solder onto your iron and drag it along the pins. The evaporating flux will pull the solder onto the pads. The hard part is in judging the right amount of solder to use.

There's another technique I've used in the past for fine pitch TQFP's. I call it "impact soldering".

It goes something like this:

  - Mask everything around the component to be soldered with masking tape
  - Flux as usual
  - Place component down and align it
  - Tack on the corners in order to ensure alignment
  - Don't worry about solder bridging
  - Now, apply solder to the pins
  - Use a healthy dose
  - You actually want an entire row to be brideged
  - You should see a solid strip of solder across all pins
  - Get all four sides done the same way
  - Now, take the iron and heat up one of the beads of solder to the melting point
  - Without any delay, hit the board edge-wise on the table
  - The molten solder will come flying out
  - A small amount of solder (just enough) will remain on the pins
  - With practice you can get perfect factory-looking joints with no bridging whatsoever

It takes a little practice, but I have found that if you need to do a lot of TQFP's this technique, once mastered, works far better than trying to apply solder precisely. It's messy at first and you might even ruin some parts. Once the technique is perfected it works amazingly well and it is very fast.

Obviously, if you have a lot of parts on a single board you are entering territory where reflow soldering in an oven is a far better idea.

I was excited by this chip because like others have mentioned it reduces the tolerances and therefore cost of PCBs, which is great for hobbyists. What are the specs for your BGA boards and how much do they cost for small runs?

The specs wouldn't be any different than they would be for a good quality board for TQFP with a pitch in the order of 0.5mm. If you need to support DDR3 (as the A10 does) and high speeds you need a good board anyway in order to make sure that the transmission lines remain true to design parameters. In other words, signal integrity is a concern with these designs and you just can't use a dirt-cheap board. It doesn't matter that the A10 is hand-solderable. You still need to attend to signal and power integrity and that means that a certain class of board is required.

Having done boards with FPGA's as large as 1152 balls (definitely not soldered by hand) I can tell you that good board houses produce good product and the difference wouldn't be much, if any between a controlled-impedance board for an A10 project versus an equivalent-sized BGA device.

This is important, so I'll repeat it: The A10 requires a controlled-impedance board.

EDIT: Just learned that the A10 requires DDR3. So BGA's are the order or the day. Hand-soldering with a soldering iron is simply not a reality for such a board. If you want to assemble one yourself you have to use a stencil and an oven or hot plate for reflow. Not that hard for small quantities.

I should also note that I've only done this when absolutely necessary. It makes a lot more sense to pay to have SMT boards assembled. It's not that expensive and assembly shops have all of the right equipment to get it right. With BGA's you really need to have full inspection (including X-Ray) in order to ensure proper attachment. This is particularly true in the case of high-speed boards.