PPP - Palm m500

Some m500 series quality problems ?
.

For writing my reviews about the m500 and the m505 I had two early devices, not from Palm, they don't support me. The m505 was a preproduction unit and the m500 was an early production unit from North America. I found in both units a quality flaw which I think is worth to be mentioned here. I want to get that right, I wouldn't bother about a single flaw in one Palm. And I wouldn't care, if it would be an obvious problem everyone could call Palm support for. But I'm producing hardware myself and I guess I know when I see an upcoming 'nightmare' for a user, most likely quite a time after he bought the unit.
.
The picture below shows a little part of the printed circuit board (PCB) of a production m500. For a magnification, click on the image.
.

.
The yellow circles are marking some of the fatal problems, especially the 3 most left circles (I didn't mark all of the tiny ones, you can find alot more on the magnified image). The little solder 'pearl' in the center of each circle is not belonging there. In fact the cupper pathes underneath the 3 solder pearls on the left are in danger to be short circuited. For all readers who don't see a PCB every day: The darker green areas are isolated board material (the cupper layer is removed there), while the lighter green areas are cupper pathes. The green paint itself is the so called solder stop (or solder stop mask). The solder stop is masked out only for component pads. It normally prevents solder to stick to the rest of the PCB during production which would cause shorts between the high density pins or pads. But back to the problem. In case of these three pearls on the left, the fatal fact is, that they are lying right over a ground path and the raw input line (unfused!) from the LiIon battery. The unit I took this photo from works in the moment, but only because the super thin solder stop paint is isolating the pathes. Additionally, the balls are not really touching the PCB. They are rather 'swimming' in puddles of flux material that is molten during the solder process and solidifies when it cools down. But the bigger problem comes weeks or months later. That flux stuff is of course no permanent glue, it's just sticky. But it dries out over the time and becomes 'brittle' and then these solder pearls are getting free. That's when a users nightmare becomes true. These balls are big enough to cause shorts between any 2 pins of the high density chips like the RAM or the Flash when they are freely moving inside a Palm. Meanwhile the cycle time of the Palm is below 100 nSec. ( = 0.0000001 Sec.!), so even a rolling by solder pearl is touching these pins long enough to short several 100 cycles. That again causes dead sure a bus error in the Dragonball and you'll get a Fatal Error. After a reset the Palm works again for days or weeks before the next pearl comes by. A permanent damage is very unlikely if only logic pins are shorted. Palm support - you might want to have the unit warranty repaired by then - won't find anything as long as they don't test the unit for days and keep shaking it. You'll get it back and you'll see the next crash soon. That's what I call the worst problem for a PDA.
.
Below you can see the image of an m505 PCB, also having superfluous solder sticking to it. It's a harmless location, but it becomes loose too, earlier or later. Then it causes the same problem as described above (again, click to magnify). Don't forget, the pin pitch of high density chips is as small as 0.5 mm (Flash chip for example). It doesn't need big junks of solder to short them.
.

.
Now why do I think that this is no single problem? Well, modern mass production is depending on a well calibrated production cycle. A critical part of it is the wave soldering. When a PCB like the one in a Palm is produced, robots are first applying tiny drops of glue in the component places, then they place the components themselves. That fixes the parts enough to run the whole board through a wave of molten solder, soldering all pins and pads in one time (that's when the solder stop does its job). This solder 'bath' has alot of critical parameters. Solder temperature, flux-solder ratio, process speed are only some of them. If they're not optimal, you'll get results like the ones on the images. But you don't realize it immediately, or the process just takes a while to settle. In that time quite a few of faulty units are produced. Additionally modern mass production doesn't use single PCBs, but multipanel PCBs. Alot of Palm layouts are arranged on one huge PCB, processed all at once and separated afterwards at intentional points of fraction. You can find the left burrs at the edges of most smaller PCBs nowadays that are mass produced.
In other words, until the wave solder process in that time was under control, thousands of PCBs may have left the solder station with that flaw. And of course the automated test process doesn't 'look' at the PCB (at least not for that problem), it just tests the electrical function. So the PCB passes all tests. I only wonder, why the PCBs are not 'washed' after production. Normally wave soldered PCBs are washed afterwards to remove the flux. Then solder pearls that might stick in the flux would be washed away too. Reasons for not washing a PCB are of course costs or, more often, parts that are not wash resistant, like switches or the piezo or the IrDA tranceiver's lens. The alcoholic fluid might damage them.
.

A final word about this page: I don't say and I never meant to say that Palm produces bad quality devices. I just wanted to show, that so called spurious crashes and 'Monday' units are reality and their weird behaviour has a reason. I get quite some mail from users all around the world describing problems with their Palms which might well comes from such flaws. They are desperate, of course. How can a technical untrained user find out such a problem? If you happen to have a device with spurious effects, don't give up bothering Palm support. I have alot of sympathy for mass producing manufacturers. It's not easy to always handle the necessary, complicated processes. But I have even more sympathy for customers, paying quite a few hard earned $$$ and insisting on flawless quality. There is that nice sentence: It can happen, but it must not happen! Bottomline is, that PCBs like the ones on the images should have never found their way into a Palm.
.

.
.