PeroThe soldering process went on mostly well, no tombstones, no misplaced components and no excesive solder...however, some bugs were found and caused me a couple of hours of annoyance:
- 3V6 and 5V voltages missing. For some reasong the switching power supply L7980 didn't deliver the said voltage. After some time of examination, I found the root cause, and it couldn't have been more stupid than it was - feedback resistors R60/R61 and R62/R63 were converesely set. So, instead of 5 or 3V6 volts at the output, the voltage divider was set to 0.75V. The reason of this folly is that in my Assembly Drawing the resistor designators were misplaced. Replacing the said resistors solved the problem.
- Short circuit at 3V3 RF. After I got SMPS to work, I noticed increased heating of the chips. Soon, I realize that 3V3 at the PLL of MAX2871 is shorted. To find out the location of the short, I lifted the U8 chip (LP5907) and broke it by accident. But at least, I've found the cause of short circuit. Under the MAX2871 chip there is a ground pad and 3V3 trace very close to it. For some reason, solder mask doesn't cover it completely. So, during the soldering, some paste went over it and made a short. Reworking the MAX2871 solved the problem. However, I've lost LDO chip and have to replace it now.
- 6-ball BGA didn't solder. This one was really anoying. The small RF logarithmic detectors LMH2110 didn't solder to their place, so I needed to do it manually. First, they are increadibly small (IC size smaller than 0603) and it was Sisyphus work to put them on their place. I managed to loose two of them just by clasping them a bit too hard with my tweezers. Second, I couldn't solder them easily because, I assume, the solder balls are lead-free. I wasn't able to heat them up with my hot air gun (nor reflow oven) because the air flow was set too low because any higher air flow would blow them away :/. I plan to use hot plate next time when I solder them, that is, when the new batch arrives because I lost all those that I had :(
- Filter lost. Simple as that, I lost the filter LP0603A1880ANTR.
These are all bugs I found so far. I examined the board and believe that it main parts should work properly. I could test out the VCO, attenuator and switches, but I cant solder the Teensy on board before all my reworking is done. So, when new detectors arrive, I'll give it a test try.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
You want to reduce the amount of solder paste to the big thermal pad under your QFN. Usually it should be around 60-70%.
Vacuum pickup would help a bit on handing tiny parts.
You have to increase the temperature profile of your reflow as the RoHS parts are supposed to be soldered at 260C instead of 240C. Could also be uneven heating so the area of your DSBGA might not reach the needed temperature. i.e. shadowed by taller parts or just uneven heating because of oven design.
I have done 4 or 5 similar 3x4 package (RoHS) from TI with hot air rework tool. It takes a while if the PCB isn't perheated or have a lot of copper under it.
Your solder paste would help the part stick a bit better than me doing that without solder paste as I am too cheap.
https://hackaday.io/project/4993-dual-channel-battery-chargeranalyzer/log/17291-keeping-up-with-the-times-prototyping-with-qfn-and-bga
Are you sure? yes | no
hanks for the feed back, those were my findings as well. I
don't think I'll use reflow oven at 260°C to fix this, as it's a self
made and I'm not really sure if it can reach that temp. I'll try with
the hot plate and air gun.
Btw, nice documentation of your
earlier project...regarding that chinese company (they seem terrible!)
silkscreen, in my case OSH Park also screwed up a bit. You can clearly
see that silkscreen is off centered from the pads on this BGA chip.
Are you sure? yes | no
The hotplate would help a lot as that's essentially what the commercial setup do.
As for OSHW, their PCB supplier has much better registration. My recent batch however, they having some issues with missing gold plating on ENIG. That one escaped their QC.
The Chinese place is hit and miss, but is a lot cheaper. I have changed my silkscreen in my layout to allow for the slop.
More on my SMT stuff:
https://hackaday.io/project/6929-smt-assembly-on-the-cheap
Are you sure? yes | no