TS-7970 Errata

From embeddedTS Manuals
Issue Status Description
RS232 prevents booting on Rev. A Workarounds available, fixed in Rev. B

Early TS-7970 Rev. A boards may fail to boot if RS232 is connected before the board is powered. A small amount of RS232 idle negative voltage leaks from the transceiver to the FET controlling the switched 5V. The FET will not toggle while the output has a negative voltage, so the 5V rail never comes up. If this is a concern or if the issue is seen, we can rework the board to have a 2.5 ohm resistor from SW_5V to 5V_A on U47 pins 4 and 5. Rev. A boards shipped after 01/13/2016 include this fix.

Boot is prevented if the USB Device port (not host) is plugged in without the console enable jumper. Fixed in Rev. C

If the console jumper is not installed the silabs has USB VBUS, but no data signals. This puts the USB device into a locked up state while it waits to communicate on this bus. Due to this lockup it is unable to monitor voltages and turn on the SW_5V to the reset of the board. The fix is to disconnect VBUS from the silabs which is done on the Rev. C PCB. A cable without VCC can be made to work around the issue, or submit an RMA.

Watchdog behavior Fixed with BOM change on Rev. H

As part of the Rev. H changes, the 32khz that was being supplied by the discrete RTC is no longer available. If the CPU does not receive a 32khz input, it instead switches to an internal ring oscillator in the CPU. This clock in the CPU can be routed to PWM, RTC, or Watchdog, but in our software this is only used for the Watchdog. The early Rev. H units did not short the CPU's 32khz clock input to ground, and may not reliably start the ring oscillator. This can result in the Watchdog either tripping, or not tripping when it should. This can be fixed with a BOM change to connect this signal to ground. For units experiencing this issue, submit an RMA here.

The default software does not feed the watchdog. This will only affect users who are using the /dev/watchdog interface in their application.