Hacker News Reader

Comment by eternityforest | Hacker News Reader

You could probably build one at home for a few hundred if you really wanted and did not care at all about time.

You would just need to 3D print a case and somehow cast your own silicone keypad. DIY electronics is actually not too bad cost wise.

View on HN

You could probably design a very similar one of tbese wirh a standard enclosure ($5) tactile switches, and a custom PCB for the top. Custom PCBs are ridiculously cheap now and even pick and place has gotten affordable for small runs.

You can even run 4-layer impedance controlled boards that support BGAs for ~$60 and get 10 of them. Problem is I now have a bunch of unpopulated PCBs I don't really intend to use. I made a few PCIe x1 to SMA adapters for an FPGA devkit - even building a single one and considering the rest of the PCBs as write-offs was less than a quarter the price of premade ones.

Can you verify good flow of solder on a BGA without an x-ray? Do you just reflow it for the prescribed time and see if it works?

It is already the way you do it for parts like QFN (at least the way I do it): if it works then all good! If it does not, then heat it up again. I heard BGA is actually pretty simple as you cannot apply too much paste, but I don't have so experience with it.

I'm doing this as a test mostly. I bought some 0.8mm 121 pin lattice iCE40HX FPGAs ($5 a pop) to test with with. Few enough balls that you can do a 4 layer stackup without cheating (no traces in the power plane layers).

I hand-placed some 0.4mm QFNs before (LVDS to TMDS converter IC) (admittedly, only a ~90% success rate) so this has to be at least twice as easy.

"Easier than QFN" isn't a glowing recommendation but makes me hopeful. QFNs are the bane of my existence.

I should mention, when I did QFNs, I used solder paste, a stencil, and a hotplate. I don't think I could ever manage to hand solder one. Passives on the backside were done with solder paste, a stencil, and a hot air reflow station.

Do you have a reference to a hotplate you like for QFN? I have been considering one because I just don't get good results with hot air, and a reflow oven seems like a bit too much for my use case.

If the board layout permits it, you can inspect the outside balls around the perimeter- if they are fine, the inside ones will most likely be fine as well. Sometimes, a small piece of front-surface mirror can be used to inspect the sides of the chip obscured by taller components.

Even PCBA companies usually just place and reflow the chip and may only use x-ray to validate the initial build- and then rely on process consistance + functional testing at the end of line.

I've been doing some test runs with HP multijet fusion prints, and I think that would be a great technology to use for case like this. You likely would need a couple-few tries to get it right, but it cleans up very nicely. I really don't like the look of FDM printed parts, so that aspect of it really appeals to me.

I'm sure you could buy a silicone keypad on Alibaba for a few dollars.

Keypad could probably be done with a flexible filament too.

Reply on news.ycombinator.com