Welcome back to the next part of this retro adventure. Last time, I told you the new boards are back from the fabrication house. These are the same logical design (called schematic) as the previous boards, but use a different and simplified layout and I’ve removed some components to make the assembly easier. I also changed my assembly approach.
I’m pleased to say I had a 50% success rate, which is 49% better than last time! The board assembled beautifully.
Unfortunately, I still have some problems but let’s go through the new assembly process before we go further. As I mentioned last time, I intended to assemble the board using a dry layout process first. This means I can take all the time I need to get the components out of their packaging and lay them out on the board and not worry about whether the solder paste is going off.
I printed off a bill of materials, which is a list of the components and sorted them in order of their value. This groups all of the 10K resistors together, and the 10uH inductors so I can go into each package just once to get the components out for placement. If I do this with solder paste on the board I may accidentally smear the paste as I’m adjusting the components, so a dry placement is a good first step. Here’s the dry placement, except for the ‘high value’ components such as the FPGA, video and Atmel processor. The theory being that I can transfer the components from the placement board on to the pasted board very quickly and the solder paste doesn’t have time to smear or spread. Here’s the dry assembly.
Once the dry assembly is complete, I turn my attention to the ‘real’ board and I put the PCB in a jig provided by OSH Stencils, and overlay the stencil on the board, carefully aligning it with the tiny BGA pads as you see below. These are the BGA pads for the 1mm FPGA. The pads are 0.5mm in diameter and the paste mask is perfectly sized.
Here we can see oversize holes for the 0.8mm BGA devices. These pads are 0.4mm, but the holes are 0.5mm. This should fix the problem I had last time where the holes in the paste mask were too small to get the solder paste through and the BGA didn’t self-align.
This is the mask for the Atmel SAM microcontroller.
This is the mask for the flash chip, same over sized holes on a 0.4mm pad with a 0.8mm pitch.
Here’s the paste mask set up in the jig. The board is visible through the holes and you can see the copper colour on the pads.
Also shown above, are the solder paste blobs for screen printing, with extra paste around the video chip and SDRAM to fill the small aperture of the paste mask. The intention is to scree once, so it’s important that there’s enough paste to scree across the whole stencil. Multiple scree can cause the paste to bleed under the mask as it lifts and resettles with multiple passes.
Above is the completed paste spread and it looks very accurately delivered. We won’t know until we remove the stencil and look at the paste deposits under a microscope.
Above, I’ve removed the paste mask and the gold pads are now a dark gray of the solder paste. Looks very neatly done with no bleeding.
Below are microscope captures of the solder paste deposits. A very pleasing result. The first is the 0.5mm deposits on the FPGA BGA pads.
Below we have solder paste deposits on the 0.4mm pads of the Flash memory chip. Note that they are oversized, so there is no hint of the pad underneath. Plenty of paste showing which means the new oversized holes worked! Yay!
Another shot of the paste on the video chip, these are 0.5mm pitch devices, with 0.25mm wide pads. However, I didn’t oversize these as the pad length is about 1.5mm, so the paste has plenty of space to squeeze through. A nice neat distribution.
Placement of the components
So now I had one board with a dry placement of the components and one board pasted up, ready to receive the components. It was ‘simply’ a matter of transferring the components from the dry board to the pasted board. In the end, it turned out to be really simple and effective. I didn’t have to consult spec sheets, check placements diagrams, I simply transferred all components from the dry board to the pasted board. It took me about an hour to load up the dry board, but only about 15 minutes to transfer to the paste board. This time saving is valuable when the paste is spreading and ‘going off’. As I’m right handed, I transferred the components from left to right, top to bottom too, so I’m not awkwardly trying to drop a component onto it’s pads over another larger component. This new process was a major step forward in the assembly process and one I’ll observe next time too.
Once all of the components had been transferred from the dry board to the pasted board, I carefully transferred this to the IR oven and reflowed the solder paste.
This is the moment the oven cycle finished. The board’s still hot! I give it time to cool before removing it from the oven.
Visual inspection after reflow
Visual inspection of the BGA below. Looking good – nice ball profile, no solder spitting and good attachment to the board. This is under the flash memory, 0.8mm pitch, 0.4mm balls.
Again, same pitch under the Atmel controller. The ball shape looks odd because of shadows and it’s difficult to light under the device. This is why X-Ray imaging is used to check BGA placement and reflow.
Balls under the critical SDRAM. This was the problem last time and didn’t reflow properly due to a lack of solder paste. This time, the balls are properly deformed and attached well. Slight reflections coming through between the rows of balls, suggesting no spitting or detritus.
Inspection of the other components, such as the voltage regulator. Looking good.
This was a bit odd, there is some ‘burning’ of the flux and a solder ball on the left:
It’s unlikely that this will materially affect the function of the design, but it was the only component that appeared like this. The solder ball can form if there is moisture on the device when it’s baked, maybe that’s the cause?
Below you can see perfect solder joints, but the resistors themselves didn’t self center. It sometimes happens with the smaller components as the viscosity of the solder can’t pull such tiny devices back in line.
The one below is a bit of a worry. It looks like the three pins on the bottom are sitting on pillows of solder. This can happen if the flux didn’t work properly, but I checked the connectivity and there were no unexpected open circuits, so I’ve left this as-is. Manual reworking would probably make the joint messy.
A perfect placement and reflow soldering of the video chip. Perfectly aligned on the left side:
And the same chip on the top side below. Very important to get this right as the tolerances are extremely tight.
Similarly the USB interface chip below. This has 0.5mm pitch pins and it was really hard to get this aligned, but it reflowed perfectly.
Four capacitors for the power stabilization for the video chip. These are just a millimeter apart and hard to place because if you nudge one when placing the next you have to start again. No issues here.
Visual inspection passed just fine.
Finishing and testing
I hand soldered on the mini USB-B connector to deliver power to the board. Upon plugging in the power, I conducted a “smoke test”. This is an old term that means to apply power and look for smoke. Since modern devices tend to have thermal cut-outs, I touched the 3.3v regulator to see if there is an overload, and sure enough it was very hot. Dang. The same fault appeared on the new board that appeared on the old board. Now I’m suspecting my design might be faulty because the board design is completely new.
I decided to quit for the day rather than spend the rest of the day struggling to find the fault. I suspect the SDRAM chip again, for no other reason that it was at fault last time. I’ll first look at the data sheet of the chip and compare it to the board layout. Maybe I’ve connected a 3.3v line to ground in the device footprint. If I can’t see any issues, I’ll attempt to trace the fault using my multi meter. I’ve read a great deal about tracing shorts on surface mount PCB since my last failure, so I have a few techniques to try. I’ll talk about these and the steps to diagnose the issue next time.
For now, I’m packing up for the day, so here’s the mandatory picture of the device packaging vs the board size. No wonder everyone is so worried about electronic waste. An enormous amount of packaging for such small components.
Although I don’t have a working board, I don’t feel the day was a complete failure. I’ve proved my new assembly process works and can help me to assemble complex boards. The reflow process was almost perfect and I have a beautiful looking board, even if it doesn’t work!
Look out for my next post when I work out the root cause of the electrical short.
2 thoughts on “Retro CPC Dongle – Part 12”
[…] I’ve made quite a few changes since the last build, so I’m feeling hopeful this one will work. Look out for details in a later post. […]
[…] you read that correctly. I managed to get the board working from my last post. I removed the SDRAM chip and the short between GND and the 3.3v line disappeared. Perplexed, I […]