I designed a device using an Arduino Leonardo, but now I am trying to design a embedded version using the same micro-controller on my own PCB.
I have copied most of the necessary bits for the ATMEGA32U4, however when I plug in the USB port the uC heats up dramatically.
I am hoping someone on here can look at my schematic and see whatever silly mistake ive made.
Attached is an excerpt of the schematic focusing on the uC.
I don't see an obvious problem - but the chip is trash now, whatever happened to it (when an AVR micro is abused such that it gets hot, that event pretty much invariably trashes the board).
I figured that much, I did prototype this in a batch of 5 so i can still play with the other ones.
I have attache an excerpt of the PCB design, i dont know if that'll help.
No ground plane, no decoupling caps, no crystal (depending on what USB configuration, you may need one, see the datasheet), traces changing width in the middle without good reason. None of those are the reason it heated up, but lacking decoupling caps will result in unstable/unreliable operation, and lacking and a crystal if using USB not in low-speed-mode (as I understand it, the arduino core does not use low speed mode) would wreck functionality.
Also why are using fuses to keep the RTS and HWB lines low, but then including pull-up resistors that won't do anything until the fuses blow?
The fuses are not soldered on, they are open pads to use for entering the bootloader. and C3 is a board decoupling cap. and Im using the USB configuration that dosent require the XTAL crystal.
ComatoseChimera:
The fuses are not soldered on, they are open pads to use for entering the bootloader.
Gotcha.
ComatoseChimera:
and C3 is a board decoupling cap.
This won't be causing the overheating problem, but you should have smaller decoupling capacitors for each of the VCC/AVCC pins on the microcontroller. For some reason the Leonardo only uses two (C1 and C2) but best practice is to have a 100nF capacitor for each power pin, situated as physically close to the pin as possible.
BJHenry:
Gotcha.
This won't be causing the overheating problem, but you should have smaller decoupling capacitors for each of the VCC/AVCC pins on the microcontroller. For some reason the Leonardo only uses two (C1 and C2) but best practice is to have a 100nF capacitor for each power pin, situated as physically close to the pin as possible.
Indeed - a board decoupling cap is all well and good (I would use a larger value than 1uF though), but it does not replace the 0.1uF decoupling caps for individual parts (which must be as close to the part as possible).
One can often get away with using fewer decoupling caps than Vcc pins if the traces are short (most pro mini/nano clones have only one, even though there are two Vcc pins and an AVcc pin), though best practice is to use one on each power pin.
I would remove one of the failed processors and use a multi meter to measure each pin to ground, verify the reading makes sense. But first I would inspect the chip connections with a magnifying glass.
If you are going to continue working with SMD boards you should get a 5X and 8x or 10x eyeloop.
What you are describing sounds a bit familiar, I just went through the similar type of problem, the schematic was correct as was the assembly. The problem was the 5 volt regulator, the 12 volt input was connected to the power supply through a short wire, the output of the regulator was connected to the 5 volt circuit. The oscilation was strong but I could not lock on with a 500 Mhz scope. The solution was to put a bulk capacitor maybe in the 50uF to 100uF range less then 1" from the regulator, another on the output. Then we put .22 SMD caps soldered directly to the pins, the problems disappeared on about 200 boards so far, before less then 1 out of 100 worked. This response is to help you get started in solving your problem, not solve it for you.
Good Luck & Have Fun!
Gil
gilshultz:
What you are describing sounds a bit familiar, I just went through the similar type of problem, the schematic was correct as was the assembly. The problem was the 5 volt regulator, the 12 volt input was connected to the power supply through a short wire, the output of the regulator was connected to the 5 volt circuit. The oscilation was strong but I could not lock on with a 500 Mhz scope. The solution was to put a bulk capacitor maybe in the 50uF to 100uF range less then 1" from the regulator, another on the output. Then we put .22 SMD caps soldered directly to the pins...
What regulator were you using? Most linear regulators need input and output capacitors.
