I have a sick robot. .

GoPiGo
1

I have noticed a problem with Charlie:

Loads on the GPIO (physical) pins 18 and/or 20 cause the robot to fail to start.

I suspect that the problem is with the st2378e level-shifter, (or the Atmel processor?), not being able to source the required current to multiple loads.

  • The GoPiGo controller will work by itself.
  • The Waveshare e-paper display will work by itself.
  • The combination of the two loads causes the robot to fail to start.

Ultimately I will probably need a new controller board for Charlie, but that’s not happening anytime soon.

Unfortunately these kind of piece parts are virtually unavailable here, and replacing them would be difficult to the extreme.

Good news:  Things that don’t load those pins, (pure i2c), don’t affect Charlie’s operation.

More good news:  Charlene doesn’t seem to have this problem.

2 Likes
2

18 is the rpi_reset Should be “off limits” ( DI intended use only).

Pin 20: Does it react correctly to GPIO setpin states? Input/pullup/pulldn/outputhigh/outputlow when not connected to display?

perhaps you could reroute other unused GPIO pins for the wave share display?

RE:”I have a sick robot” - he’s just not comfortable displaying his emotions.

2 Likes
3

It doesn’t even get that far - the robot won’t even start.

1 Like
4

Here’s the test jig I used:

A 40/40 pin isolation adapter:

20241113_192849
20241113_1928491920Ă—865 217 KB

 

20241113_192944
20241113_1929441920Ă—865 165 KB

 

Using this I can isolate any one or more pins by flipping the respective dip-switch(es).

The black lines indicate each tenth pin.

The two green lines indicate pins 13 and 26 on the old-style 26-pin interface.

2 Likes
5

no cold joints, no reversed, no off by one in 120 junctions? Impressive!

no burned fingers?

2 Likes
6

TL:DR - This isn’t my first rodeo. :wink:

I did this in groups of 12 pins, (the dip-switches are grouped into twelve switches each except for the last one which is four switches), and checked each connection under my microscope after the end of 12 connections.  Any questionable connection was measured with my meter.

There was a fair amount of rework where wires broke or connections didn’t work, but by grouping them into groups of 12 connections, it was easier to keep track of them.

I also did a rather quick-and-dirty continuity test from one header to the other after completion.

1 Like