RaspberryPi Soft Switch PCB check

photomankc

So this is my first modestly complicated SMT PCB design and I wanted to get feedback from the hive on what I have thus far, Basic theory is that a PICAXE 08M2 is used to monitor a push-button switch to signal when to turn on. After turning on, the Raspberry can send a KILL signal to turn off it's own power 10 seconds after the KILL line floats back up to logic H. The user can press the button briefly to tell the Pi to do a powerdown, or the user can hold the button down to force an immediate power down. The PICAXE switches to 32KHz clock when it's in the OFF state and draws about 35uA. MOSFET I had on hand was a logic level P-Chan that has a max rating of 25A. As this is I suspect 2-3A would be the maximum.

The ADC allows for the Pi to sense battery voltage and shutdown while there is some reserve remaining. Tried to move the ADC as far from the MOSFET as possible to reduce any temperature variation in readings from the local heating that MOSFET might generate. On-board regulator for 3.3V is very low quiescent current but adds about 15uA to the off-state current drain. The ADC adds another 1uA when not doing a conversion.

TOP


BOTTOM


SCH



Let me have it!

Mark_T

Unless you are planning big heat-sinks, don't choose MOSFETs by current rating, choose by Rds(on) -
you need to calculate the heat dissipation (the current rating is merely an indication of the max power
rating under infinite heat-sink conditions - there are devices where the current rating is larger than the
current that damages the packaging / bonding wires!)

[btw layout looks OK - you could get the board size down a bit, but if you are hand-soldering it is
probably wise not to. Perhaps a few more ground vias?]

photomankc

Mark, I appreciate the feed-back. have made sure that each group now has it's own local GND pad rather than snaking around to other via points. I will be hand soldering this board so It is best to keep it more open, but also I needed it to keep a 1" mounting hole pattern since that is the hole grid I added to my robot chasis. Well it's actually 1/2" grid but getting the layout I wanted into 1.5" x 1.5" was not working out. I may look at making a modified version that stacks onto the PI later on but this is for my robot and I won't be stacking it in my design.

The MOSFET actually has a little bit lower than average RDS(on) so it burns less wattage passing the current. I calculated the heat based on the case-to-air thermal specs and the numbers say 4A should be safe. 4A is 800mW and a junction temp of 102C. My small bot is a stepper based system that draws a maximum of about 2A, for 200mW and junction temp of 60C. I have plans for a larger bot later but it would have a separate motor supply with switching that was tied to the state of the logic power but not directly connected.

Again, thanks for giving me some feedback. I'm going to start on the prototype this weekend. The SMD PICAXE's are on their way. I really wish a US outlet would start supplying those. I know I could go to a bare PIC or AVR but this is nice for getting the overall design worked out and they are easy to interface to.

photomankc

PCB is done, electrical checks out on it and the switching works as intended, so tonight I will mount and program the PICAXE SMT part (lower right). I have been able to create a version that is about 1/2" shorter on one side. I also have found another MOSFET that is more appropriate for the job and it's smaller to boot and has lower RDS(on) and thermal resistance. I'm going to look at making a version that stacks on the Pi's GPIO header and see if I get any interest in it.