6502.org

Not simultaneously, but you can mount one side first, reflow it, and then do the other side. Unless components are really heavy, they'll keep hanging on the bottom while the solder melts. I've even done it with fairly big inductors and 1812 ceramic caps.

Quite the opposite, in an oven it is possible. On a hotplate, you really can't. However, I was considering mounting BGAs and top-layer devices on a hotplate (just because I am so happy with the results), then flipping the board over, applying paste and bottom layer and sticking into the oven for the bottom layer.

I would have to use a special jig to make sure that the top components don't touch the rack, but that's not too hard.

How do you keep the parts from falling off, when you reflow for the opposite side?

Surface tension keeps them on. Sometimes they slide around a bit, but BGAs have so many balls that I don't expect a problem there.

Like Arlet says, surface tension. And being very, very gentle.

My oven, a convection toaster oven, has a nasty vibrating fan. I really have to rewire and PID it soon.

Of course, the easiest thing to do is place components on top only.

Gentlemen, great info! So with a hotplate one is pretty much limited to have BGA type IC's on one side of the board. Then will have to hand solder the opposite side. This is not too much of a limitation for a design IMO...

I will have this hotplate way before I even start designing a BGA board. But I think I will practice dismounting BGA devices from old motherboards to get a feel for things and temps required.

Have you ever thought of maybe building another hotplate and spacing it ontop of the original to make your own oven?

Typically I put all the ICs on top, and only put some small stuff at the bottom, such as (decoupling) caps, resistors, ESD protection. If there's ever a problem with components shifting or falling off, it's usually not a big deal to fix them by hand.

Yes. The hotplate is not the best radiating surface though.

I also considered cutting the bottom of the oven out and placing it over the hotplate like a hat. Haven't settled on the best idea yet.

They still use their own custom software, with no gerber output (which locks your design into their service), right? If standard Gerber and Excellon files are produced by the CAD and used in manufacture, then even a cheap CAD's drill files can be edited easily by hand. I would copy the drill file to another file, then in the first one, delete the text lines with the X-Y data for the holes you don't want plated through, and delete everything but those in the second file. If it's just a few holes, it's pretty quick. Obviously it's not a practical solution if you want to do this with hundreds of holes.

Unfortunately, I was informed by oshpark that American fabs will make any holes through copper plated. Chinese fabs seem to respect unplated holes - goldenphoenix will do it 4 different ways - you can designate certain hole sizes unplated, send in a separate drill file, and I forget what else...

We (the company I work for) have been using Imagineering in Illinois for boards, so I checked their capabilities page, and about 60% of the way down the table it says the hole-size tolerance for NPT (non-plated-thru) holes is +/-.001"; so obviously they will do a second drilling if asked. They're not the best for price on hobbyist quantities, but for low production volumes (like a few dozen boards to a few hundred), they're really good, and they don't have the limitations of some of the ultra-low-cost houses that cater to hobbyists.

Great to know! I may try them if I ever get to low production quantities.

I ordered another 8 5/8" x 6" x 1" aluminum block for this purpose of having a future little mini oven. I held off ordering anything else until I get the 1st hotplate working.

For the mini oven, the upper hotplate could do double duty with a board sitting on top for the first stage. And a board in the "oven" for the re-flow and soldering of the opposite side, so electricity wouldn't be wasted. Spacing for the upper and lower aluminum blocks would be tighter than .5", but would be adjustable. Also, after experimentation and observing the results, the 4 sides would be attached also with 1" thick aluminum. 3 sides would be permanent, the front removable. The disadvantage is that one can't see what's going on. But we'll see how programmable the PID controller is.

Also, I am going to pursue filled in vias. Maybe best to use the higher melt temp of silver bearing solder to fill in the vias. Also use the finest grit sandpaper to level everything out afterwards, before soldering the BGA. Also, as enso pointed out, fill the solder in from underneath, with the board upside-down, so gravity will act on the solder.