Need some advice on this one, as I am stumped. Question is, why does this 85 reboot on PCB ?

I have this atmega85 (adafruit trinket 5v) ckt that works fine on breadboard but will not when I move it to a PCB. This was my 2nd go at it and I still have the same issue. The only issue is the wake pin.

Pin2 is configured as a input_pullup and the code uses this pin as interrupt ("wake") during sleep mode via pin state change. On my breadboard when I take pin2 to gnd the 85 comes out of sleep and resumes. However, on the PCB when I do the same thing the 85 reboots.

Notes: its 9v batt. A pFet feeds the Trinket, and the 1M/220uF is a timer, essentially cutting power after about 7min, if that cap is dumped out to gnd then it stays running (via switches on toy, etc). The pixel ring power on both sides (+ and gnd path) are controlled via p/n Fet pair, because if you don't do it this way and only pull +V from ring the ring still allows mA to flow. So in essence, pin3 is used to control power to the ring (pin 3 off when 85 goes to sleep, on when 85 wakes). I do it this way because this a AA batt powered toy, and it will go onto the shelf when kid is done playing with it, no on/off switches, hence to save batt power I do it this way, etc.

Also, using my basic DMM I cannot see any voltages dropping too low (batt or the +5v output), so the reboot appears to be inside the 85 itself. I thought perhaps the 85 was choking on the nFet gate charge when pin3 turns back on, but gates are just 11nC each, and I put a 100ohm in series with that and still same issue.

Board & schematic:
excuse the paper drawing, round two so I figured to trace out every wire on bboard and put it on paper, then build that on PCB, etc.

 

Your drawing shows the reset pin floating. I'm not familiar with the Trinket, but with most micros that's a no-no.

 

Your drawing shows the reset pin floating. I'm not familiar with the Trinket, but with most micros that's a no-no.

inputs can be in a mode called "pullup" which has ~50k ohm pullup to +5v
but the issue is, works ok on bboard, not ok on PCB

 

works ok on bboard, not ok on PCB

Perhapse the stray capacitance of the bboard was making the difference?

 

Perhapse the stray capacitance of the bboard was making the difference?

i dunno. pF between pins of bboard doesnt seem like the reason why it doesnt reboot on bboard.

 

The only other reason I can think of is that there's a wiring-error/solder-bridge on the pcb.

 

Your best exposure method will be the photoplotter. e.g. https://pico-systems.com/photoplot.html

 

The only other reason I can think of is that there's a wiring-error/solder-bridge on the pcb.

thats what i thought the 1st time. the 2nd time i made very sure to wire/solder it over a few days, verifying each step, hence why that schematic has green lines, as i marked it as i went along. i looked at the soldering under my hands-free board holder magnifier, i dont see anything wrong with the soldering/wiring.

 

Do you have an oscilloscope? Can you try watching the battery voltage, 5V supply, and reset pins during a wake cycle?

If not, I'd try investigating the floating reset pin and stray capacitance theories with a little trial and error:

• Try a stronger pull up resistor on reset pin to ensure it's not doing something strange.
• Small, fast cap (0.1 to 1uF ceramic?) decoupling 5V line.
• Similar small caps anywhere else that a little bit of capacitance might make a difference.

None of the ideas quite make sense to me, but if it works on a bread board, and if you're sure you transferred the circuit to perf board correctly, there's not much left to do except start doing some empirical testing.

 

i cant change the internal pullup, its cut into the die of the atmega 85, etc. "input_pullup" is an internal function "mode" that activates the pullup, etc.

however, i can try the pin as a std input with external 1500ohm pullup and see what happens.

i do have a scope, just havent used it yet on this problem.

 

well, i dont think any added caps helped, but i changed the pin mode and did a external pull down.

i did add .047uF ceramic and 0.1uF electrolytic to the output of the regulated +5v
i also added the 10k ohm pull down and a 10pF to ground on the wake pin.

now it works. why? i dunno yet, but i am in a rush to get this item shipped.