Here’s what my colleague said to me
We were wondering if there’s any known underlying issues with the IRQ signal that might be worth looking into with this type of unit.
- Does he have another ABCC to try with? In case he has broken this one.
- Can his software see changes on the corresponding input pin on his microcontroller if he unplugs the ABCC and ties the IRQ pin directly to either 3.3V or GND via a wire? In case he has messed something up with his I/O definitions and this pin on his microcontroller has been configured as an output rather than an input.
- The IRQ signal is in the top row on the CF connector, if he unsolders that one from the board and connects the scope directly to it (so that the rest of his hardware can’t affect it) what will he see then?
- A short-circuit on their side? Or a broken module due to ESD or a short-circuit on their side?
Also note that a pull-up resistor needs to permanently wired up to the IRQ. The IRQ is a normal GPIO pin on our side, it is floating while the ABCC is in RESET, and if there is no pull-up resistor to stabilise it in the inactive state between when the RESET deactivated and the ABCC has booted up and assigned a high level to it any noise might be interpreted as a low signal on his side, and then his microcontroller will start talking to the ABCC too early. This is not the cause of the present problems though, the oscilloscope image confirms that since the IRQ goes high but never low, but it might be an issue further on if this resistor is not already there.
Also, the IRQ goes low appr. 85ms after the RESET is deactivated, with SPI being used, on an ABCC40-ECT I have here, so the EPL should not be too far from that.
Is there any particular condition that could prevent nIRQ from toggling?
No, not apart from hardware-level issues or an incorrect Operation Mode.
If I power it with 3.3V the module signals 99% times exception state. Is this ok?
It is not that he gets a ‘Fatal’ immediately at RESET deactivation rather than an EXCEPTION at a later time? (Double-check the LEDs, one solid red LED -> EXCEPTION, two solid red LEDS -> Fatal.)
If the Operation Mode pins have an illegal combination the ABCC should go to Fatal more or less immediately when the RESET is deactivated.
My fault! I forgot to connect the 50pins connector shield to ground! Now everything runs smoothly with 3.3V
Should not matter unless there is something strange going on with the power pins. If all of the power pins are properly wired up he should not need the GND spring just to get things running. That spring is more for EMC stability. It might be an idea to have a closer look at the power supply connections.
all DIP switch pins are not connected
all LED interface pins are not connected
That should be corrected in the final version of their board. No inputs on the ABCC should be left floating, they must be tied to stable inactive level. And the active-low open-collector/drain outputs should also have pull-up/down resistors to give them a stable voltage in the case the ABCC does not presently drive that output.