I use a good number of load cells for various things but recently have had quite a few load cells fail. For reference I’m using a 5V supply and I attached the schematic for reference.
[attachment=0]LoadCellAmp.png[/attachment]
I was able to deduce that the cause was either a 10uf (C7) and or a 0.1uf(C3) capacitor that is connected to ground. When the aforementioned capacitors were on the board the load cell amp power draw was ~2 watts to ~25 watts. When I removed the the capacitors the power draw went to normal and the load cells amps seemed to work as expected. I’m curious if you have seen this issue before / something similar. Also, have you had any production issues with the capacitors not being rated for 5V?
Note: I tested the HX711 chip and transistor on the high power draw load cell amps by soldering them onto another load cell amp and they seemed to work fine.
My money would be on C6 and/or C7; and stress cracks within their internal structure. And those having internal stress fractures probably because they are sitting pretty close to the edge of the board, and get stressed when the board is de-panelized the. If you replace C6 and C7 you are probably back in business.
For a more detailed explanation…
Those two caps are 10uF ceramic in 1210 package. That is a very high capacitance per volume for a 1210 package. You can see from the board picture they look “pregnant”. Many years ago that type of high capacitance structures were build as “stacked” ceramic caps (imagine two or three smaller value caps stacked on top of each other and repackaged in order to make the higher value capacitance) which had pretty bad failure rates. In more recent years the cap manufacturers have been able to create the same high capacitance in one singular package with much better yield rates.
But there is a catch; a capacitor is simply two plates with some dielectric in between the plates. Distance, area, and dielectric give you the capacitance. For a fixed area (i.e. package size - for example 1210), the only way to increase capacitance is by adding more plates (i.e. building upwards). That is the basis of MLCCs (multilayer ceramic capacitors). To try to keep the volume within reason, and for same dielectric type (i.e. NPO, X7R) the only variable you have to play is with the thickness of the plate. So the higher capacitance, the more layers (vertical height), and in order to keep the vertical height within manufacturing limits the plate thickness has to be reduced. So for a 0.1uF on a 1210 package you can use thicker plates within the manufacturing volume limit. But for a 10uF (100x more capacitance) you need a lot more layers and the plates in those layers must be pretty thin. Now if you apply some twist or mechanical stress to that capacitor structure the plates will be stressed, and of course the thinner plates will be more prone to cracking. In most cases you would get an open condition, but in cases where the dielectric is pretty thin (which would be the case for the highest layer content caps) two or more of those plates may touch and if enough of those plates touch you get a short.
An where does that stress comes from: it usually comes because those caps sit next to the edge of the board. The board is panelized (normal printed board manufacturing process to reduce cost), and it is assembled (components mounted via pick-and-place) as panels (also normal printed board assembly process to reduce cost). After the panel is assembled it gets de-panelized to yield the individual boards. The de-panelization can occur in a few ways: routing, v-grove, etc. There are certain methods that create a lot more stress on the components in the edges. Some cap manufacturers are better at making their MLCC structures more robust than others. It is very likely that the breakout boards you are using were assembled prior to being de-panelized, and that they used something close to v-grove for de-panelizing which increases the stress on those components near the edge of the board. And that given the current shortage of components the very same caps were previously sourced from better manufactures but now are being sourced from some reasonable alternate that although capable may not be as good.
