Surface Mount Tombstone Experiment

0201 Resistor Tombstone

What is a tombstone?

The tombstone effect is a common failure in surface-mount technology. Smaller chip components such as resistors and capacitors suffer the most. They have a small mass, and the surface tension from melted solder has enough force to flip the part up on its end.

What causes the tombstone effect?

The usual suspects for the tombstone effect are:

  • Lack of thermal balance - one pad melts before the other.
  • Insufficient solder dam - solder flows from a surface-mount pad into a nearby hole
  • Uneven application of solder paste - different surface tension on each pad
  • Placement offset - part not correctly centered during pick and place

Who is to blame for tombstones?

Thermal balance and solder dam are design problems that a designer should know about and eliminate during the design process. If the board has these design issues, it can lead to problems with fabrication of the board and manufacturing delays.

Uneven solder paste and placement offset are challenges for the contract manufacturer. The CM needs an accurate pick-and-place machine and a good solder paste process to solve these problems.

The tombstone test board

If we are assigning blame here, it would be good to know which effects are most important! I designed a test board to find out.

I totally made all those mistakes on purpose!

The top row of resistors has the IPC-based footprints provided by my friends at SnapEDA. The next row uses the 0201 footprints from the Panasonic library in Altium. The rows below are the same, except that I added copper pours to create an intentional thermal imbalance.

Close-up of above layout showing zero solder dam, grid points are 1mil

I left out the 2 mil solder dam between the pad of C3 and the via pad. The solder mask keep-outs touch, and the solder dam width goes to zero.

The classic tombstone-look of a flipped-up resistor is not the only failure mode. It is also possible to get a bad connection without visibly flipping the part. To detect possibly-hidden open circuits, I used chains of resistors that I could easily measure. I tried to be clever and make my circuit light the LEDs if the resistor connections are good.

Any bad resistor connections, and the LEDs won’t light

The results

The folks at Tempo Automation built my test boards. I uploaded the Altium pcbdoc file and an Excel BOM, and added a Fab Note, telling the DFM team not to fix my design, and the assemblers not to fix any tombstones! Since Jeff at Tempo was buying, I checked the someone-else-will-pay box.

After the boards came out of the oven, there were no tombstones anywhere! The lack of copper balance on this layout is not enough to cause tombstoning. There was flow of solder where the solder dams were too narrow, but not enough to cause tombstoning problems there, either.

To try to get some tombstones for-sure, the Tempo crew reprogrammed the solder jet machine to reduce the solder on just one of the two resistor pads. Using 10% steps, they went from 90% all the way down to just 10% of the recommended solder amount. There were still no tombstones, and all the solder joints were good. Amazing!

The word from Tempo is that they do see tombstoning when there is a huge thermal imbalance, such as diodes connected on one side to a large, thick ground plane with just a small trace on the other side. Then, they get 100% tombstones. However, with the relatively minor problems on the test board, they just couldn’t get the resistors to flip up!

Via too close to pad — Solder down the drain!

I also added vias too close to the pads on C1–5. Without sufficient solder dam, when the paste liquifies it gets sucked into a via.

There needs to be a 2 mil solder dam between the pad of C3 and the via pad. In this case the solder dam width goes to zero. There is still plenty of solder still on the capacitor pad, but it is easy to imagine that of more of the solder could flow into the hole and create a dry solder joint. To ensure that there is enough solder on the pad, use the correct solder dam!

Accurate placement is important

To create the tombstones shown in the picture, the Tempo crew programmed the pick and place to intentionally place the parts off-center. Without this intentional placement offset, the accuracy of the Tempo pick and place machine is sufficient to reliably place 0201 resistors.

0201 resistor tombstone caused by intentional placement offset

You should still follow the guidelines for thermal copper balance and solder dams. Yields will be better if you follow the design rules. But based on my experience with this board, there is no need to sweat the finer details of copper balance, at least at Tempo. It is more important to have an accurate pick and place process.

No more Mr. Nice Guy!

Okay, Tempo can load 0201 resistors all day and not make tombstones. How about 01005 resistors? I have been assured that these will be no problem. I intend to find out in my next design.

I’ll continue to make design mistakes here on purpose so that you can learn about them without it costing you delays and board turns. Stay tuned to see if I can become the “Worst PC Layout Designer in the West.”