I have two circuits using MSP430 processors that reboot under ‘moderate to severe’ circumstances. One circuit drives a car horn and reboots each time the horn is activated. The second is a lightning detector that reboots when lightning gets very close. In both cases the circuit, is on a breadboard without a ground plane but with somewhat random bypass caps.
In the case of the car horn, the power supply from the MSP430 is COMPLETELY separated from the horn power and fully optically isolated. They do not even share a common ground.
I am not asking for help to debug these specific problems. Although if you have ideas… What I am asking for are references to the proper design and specification of bypass capacitors and or board layout. I will be migrating these designs from breadboards to PCBs and hope to leave these #@#^# problems behind me.
My background is as an embedded SW engineer formally trained as an aerodynamics and mechanical engineer. All my EE knowledge has been picked up on the job, via my home lab, or by reading.
Any web resources, books, fair tales, or other material you have found particularly useful that I could be reading?
You probably need more than just by-pass caps when there are severe transits near your circuit. Notice I said near and not on your circuit. A little study of magnetic fields and current flow will give you an idea of what can happen.
I’m also sure that your circuits as built will not pass CE certification for ESD. (another google term for you)
A device that can help is a TVS (Transient Voltage Suppressor). Although if you have a poor PCB design even TVS’s won’t help enough. Does your PCBs have a good low impedance ground plane?
You now need to discovery what part of your circuit is susceptible and fix that. Hint: You may need more than one fix (ie. cap on one part of circuit and TVS on another).
This is not an easy issue to trouble shoot and fix. Good luck.
waltr:
This is not an easy issue to trouble shoot and fix.
Your not kidding. The car horn project started out running from a motorcycle battery to power the whole thing. By the time I gave up, I had a 9V to run the MSP. The GPIO to run the horn went to an optical isolator, then to a FET to drive an automotive relay (with kickback diode), to drive a car horn (also with kickback diodes). The FET, relay, and horn were on the motor cycle battery. They did not even share a common ground as I was worried about ‘ground bounce’.
I added ferrite beads to the horn leads. I added various caps across the horn contacts. I wrapped the MSP430 in copper tape and located it 3 feet away from the horn. I insured there were no non-committed pins on the MSP. No reduction in failure rates and NO visible glitch on the power line using a 500 MHz scope and 10x probe. ARGHHHHH! I suspect Tachyon beam emissions.
I will look into TSVs as well as researching CE/ESD data (although my device is just for home use).
When I discovered the same sickness in my lightning detector last night I had this overwhelming feeling of dread. Small simulated strikes worked great. Large bolts caused a reboot. But this time, I did see a glitch in the power supply to the MSP. And hence my desire to learn how to spec caps to remove specific size and duration of spikes.
leon_heller:
The use of a solderless breadboard is probably the main cause of your problems; I avoid them like the plague for actual MCU prototyping.
Yes, I suspect that the breadboard is a large chunk of the problem. But the car horn issue was on point to point “doughnut per hole” perfboard. I do wish to transition to some custom PCBs, but still need to learn about design rules for ground planes, bypass caps, routing, …
fll-freak:
I added ferrite beads to the horn leads. I added various caps across the horn contacts. I wrapped the MSP430 in copper tape and located it 3 feet away from the horn. I insured there were no non-committed pins on the MSP. No reduction in failure rates and NO visible glitch on the power line using a 500 MHz scope and 10x probe. ARGHHHHH! I suspect Tachyon beam emissions.
Nothing so strange as Tachyons, just common ordinary magnetic fields. I recommend you locate a copy of:
The copper tape cannot shield out magnetic fields. For that you need expensive mumetal.
Your horn circuit has high di/dt which creates a changing magnetic field which induces a current in nearby conductors. You can minimize this effect by controlling the loop area at both ends. Make sure that in the horn circuit the return wire is parallel to the wire carrying the high current. Twisted pairs are very good here. Distance between the source and susceptible circuit also helps. (If you had the optocoupler located with the other high power devices then the wires to drive it form a loop and you need to use twisted pair.)
On the MSP430 end the usual technique is a ground plane although you could carefully route wires and return paths side by side.