BBB with own cape stops working

Hello all,

I urgently requesting your support! I am desperate!
I am currently developing a radio cape which has two Atmega328p, 3xRFM69 radio modules and some other small things on it. You can find the circuit in the following link:
https://drive.google.com/file/d/0B8Fq4yYMgG7uQ3k3WkJUbG9icWs/edit?usp=sharing

When I use this cape on by BBB, the Beagle stopped working during operation. I tried to reboot it but it now do not boot up like the issue described in the Beagle Wiki as “Improper Power Down”. I never unplugged the power cord from the board. It just stopped working.
When the cape is connected to the beagle the components are sourced by VDD_3V3Exp and VDD_5V. So no external power supply. I additionally just have the Uart4/5 connected to my circuit…

I would really appreciate your help. Because I really do not understand what the matter is!

THANKS A LOT!

Looks like C1 is not gated by the 3.3V rail from the BBB connectors.

.
Gerald

http://www.elinux.org/Beagleboard:BeagleBoneBlack#Expansion_Header_Usage

Gerald

What do you mean with C1?

IC1. Your regulator.

Gerald

Oh I understood, I need to add the C1 on each Atmega 3V3 supply. → blocking capacitor
But could that really be the issue which stopps the BBB running. I can also not boot the beagle up when I have removed the cape.
When I want to boot up the beagle alone the power led flashes up one time…and thats it…

Your regulator must be turned off until the VDD_3V3 rail comes up. Right now it comes up almost immediately as it is powered from either the expansion header or a weirdly designed 5V DC input. The idea of gating the ground to control power will not work as you already have a DC path to ground, assuming that all of your grounds are connected together as they should be.

Gerald

Thanks for your quick answer and very helpful feedback!

I am not sure if I understood right.

Currently I have to options to power the cape.

  1. Use the cape without the Beagle
    → Power supply over the FTDI connector and solder header Pin2-Pin3 connected on the MCP1703 side.
    → 5V solder header Pin1-Pin2 connected
    → 3,3V will be generated with the MCP1703 from the input voltage VIN

  2. Use the cape with the Beagle
    → Power supply all over the beagle headers and solder header Pin1-Pin2 connected on the MCP1703 side. Therefore no voltages will be generated by the cape itself…
    → 5V solder header Pin2-Pin3 connected. 5V from the beagle.
    → 3,3V from the beagle

Therefore the atmegas can only power up when the 3.3v supply of the header is available. The 5v supply is always on thats right. But 5v is only used for LEDs and the Piezo Buzzer and this parts are
disconnected with a BC817 from ground.

Yes all wires which are named as GND are connected.

Do you really see an issue on that? What needs to be changed?

The FTDI connector will not be connected when the cape is connected to the beagle.
All GND signals are connected to each other and they are all connected to the beagle. No other/external GND is connected to the cape.

So your contention is that this is all correct? Do I read you right?

Gerald

You are right, there must be something wrong but I do not know what.
I understood that the blocking capacitor is missing. But I do not know if that is the only mistake and this is the reason why my beagle stopped working because of my cape…

I updated the circuit to that configuration like it is when I have it connected to the beagle.
https://drive.google.com/file/d/0B8Fq4yYMgG7uTFlIRjdjcGRUYlE/edit?usp=sharing

Looks like you have shorting posts for your standalone versus cape mode. If you ever by accident plug in the standalone mode into the BBB, then you will damage the board for sure.

Gerald

Yes of course. I added this option just to run the cape for first tries on a USB port.

Could you find any other problems? I am currently thinking to connect the SYS_RESETn of the BBB directly to the reset pin of the atmega`s?

Nothing else jumps out at me. The 5V rail you are using is on before the board is powered up, but I don’t think that should matter. You could switch to SYS_5V, no reason not to.As you only have four wires to the BBB, you might try adding some isolation there, and use the RESET signal to disable the buffer before until you cone up.

There may be something else in your setup. But, schematic wise I don’t see anything else.

Gerald

I am not sure what you mean with “to disable the buffer before until you come up”.
Directly connect the Beagle SYS_RESETn pin to the pin 29 of the Atmega`s?

What would be the positive effect by using the SYS_5V. I think it is directly connected to the Beagle DC plug. So, it can drive more current?

Two more ideas:

  • would it be helpful to add a 0,1µF capacitor to the beagle voltage supply pins on the header, between VDD_3v3 and Gnd very close to the beagle headers?
  • would it be better to completely supply my cape from SYS_5V and create the 3v for the Atmegas and the other components on the cape with the MCP1703?

@Gerald: What else can be wrong except the hardware design?

It has to be HW or the application and use of the hardware. Maybe you have a short or some other issue with your assembly.

Gerald

Can the 3V3 supply of the Beagle drive all the components regarding max current?
Could that blow the beagle like described at the beginning?

The rail can drive it. No using the 3.3V rail will not cause damage to the processor I/O pins as long as the reset lines is sued to make sure they are not driven until reset is released…

Gerald

Hi Uli,

on your design, some your signals go to the BBB boot mode pins.
For example, 2_D0_RX signal goes to the SYS_BOOT_8 of the BBB which must be Low at power up.
Please check every other your signal likewise, if they do not prevent the BBB from booting.

Normally people just disable all the drivers on the cape until the BBB is up and running.
This is what was initially suggested.