Qbone sudden loss of power

[Andrei Warkentin]

Dear all,

I've been witnessing some strange behavior recently. I'm curious if others have seen it and what the cause/remediation might be.

After turning on the system, I see the QBone starts powering on. You can hear the relay click. The green led is on and there are blue leds on the BeagleBone. At some point -all- lights go out on the QBone. The rest of the cards in the backplane have their power, so it's not a PSU issue.

After turning system off and back on, the QBone boots successfully. I wasn't able to find anything in the logs. Could the relay be "unswitching" itself? Why would it do that, and why would it only happen on a "cold start".

The system:

- H9278-A backplane, ATX PSU applying power via 4-pin molex.

- Front panel with all the correct cabling for the H9278-A, and ATX power on wired into the power swtich

- KDJ11-D/S with cab kit

- other cards

A

[Joerg Hoppe]

Hi Andrew,

if the relay is defective, you should see +5V on the BBB P9.5 and .6 go to GND.

If +5V on BBB input is OK, but +3.3V on P9.3, .4 is dead, then either the BBB is defective, or
+3.3V is overload due to an soldering failure.

The  green LED is powered by +3.3V.

There may also be an failure on die H9278 PSU, causing a slow or erratic power-up ramp.
Or one of these power-Molexes has a bad contact while cold ... they fail sometimes, are they "brown" ?

In any case, an oscilloscope trace on BBB P9 +5V is needed now,

kind regards,

Joerg

[Chris Cureau]

I don't know why this didn't post properly when I did a reply-all, but here's what I sent to Andrey yesterday:

This exact issue actually happened to me recently. While I haven't had the chance to trace the exact reason, it seems to be connected to the auto start function.  When I stopped any configs from loading at boot, the qbone stayed on. I was able to manually configure and boot from the simulated disk after that. When I have a few extra minutes I'll try to do more troubleshooting.

That being said, I haven't done any more than check voltages (my 5V rail is at 4.77V), but I'd be happy to understand how better to test.

My current setup: 

 - H9270 backplane, wire wrapped to 22-bits

 - M8189 (11/23+) in slot 1

 - M7551-BC in slot 2

 - QBone in slot 3

These are powered by a 400W ATX power supply which provides +5 and +12 to the backplane. BDCOK and HALT signals are set up through a 7404, as in Diane's page (http://www.diane-neisius.de/pdp11/index_E.html).

Cheers,

Chris

[Andrei E. Warkentin]

I do have autostart enabled, and in terms of the time until the "thing" happens - yeah, sounds about right... it's as if it boots up all the way to the autostart.sh script, and then the board goes dead. It's just extra strange that the lights on the board go off as well.

My 5V is around 4.5. Definitely on the low end. Gonna focus on that for now.

N.B. I did try to get serial logging out, but it's a bit confusing. The BeagleBone serial pins (ttyO0) are only used until /dev/ttyS1/S2 get brought up, and after that I haven't figured out where the console printk's go. Definitely not on any of the UART headers. So if something weird happens, those logs seem to go nowhere.

A

On Mon, Apr 1, 2024 at 3:12 PM Chris Cureau <cmcu...@gmail.com> wrote:

This actually happened to me recently. While I haven't had the chance to trace the exact reason, it seems to be connected to the auto start function. 

When I stopped any configs from loading at boot, the qbone stayed on. I was able to manually configure and boot from the simulated disk after that.

When I have a few extra minutes I'll try to do more troubleshooting.

--
You received this message because you are subscribed to the Google Groups "UniBone" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unibone+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/unibone/7f6252c1-e487-4d98-8c41-40dd36ffa9a6n%40googlegroups.com.

--

A

[Chris Cureau]

This actually happened to me recently. While I haven't had the chance to trace the exact reason, it seems to be connected to the auto start function. 

When I stopped any configs from loading at boot, the qbone stayed on. I was able to manually configure and boot from the simulated disk after that.

When I have a few extra minutes I'll try to do more troubleshooting.

On Mon, Apr 1, 2024, 2:12 PM Andrei Warkentin <andrey.w...@gmail.com> wrote:

[Joerg Hoppe]

Am 02.04.2024 um 16:03 schrieb Chris Cureau:

I don't know why this didn't post properly when I did a reply-all, but here's what I sent to Andrey yesterday:

I just received your messages as "spam report" and had to confirm them manually  ...

What does your autostart script contain?

4.77V is low, but still acceptable, is that measured on the BBB pins ?

kind regards,

joerg

--
You received this message because you are subscribed to the Google Groups "UniBone" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unibone+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/unibone/0497b5e4-ab4c-433c-b8b1-9e8537ba4502n%40googlegroups.com.

[Andrei E. Warkentin]

Thank you Joerg and others,

Unfortunately not quite setup yet to collect cold hard data on voltage/signal levels (beyond using a multimeter on a QBus breakout), but I've reason enough to believe the QBone is fine. I've "improved" my situation by backfeeding the other Molex connector as well, so the setup with an M7554 is stable enough, but it's naturally not a solution. I need to mate the proper ATX 5VDC voltages into the proper backplane power connector first...

A

To view this discussion on the web visit https://groups.google.com/d/msgid/unibone/e1133e92-602d-4dca-876f-3a03b0034f7b%40gmail.com.

--

A