Problems updating firmware - now board unconcious!

SimonRafferty · 26 Oct 2016, 16:27

Thanks David - I'll try that now!

Si

SimonRafferty · 26 Oct 2016, 17:21

Sadly - that didn't help!
No obvious corruption. Did a low level format and restored the original files from Github - behaves the same
Tried brand new 8Gb Card with original then my edited files - still nothing.
Any other options?

Si

dc42 · 26 Oct 2016, 17:45

I can think of the following possibilities:

1. Something in your config.g is triggering an undiscovered firmware bug. But if you have restored config.g from github, we can rule that out.

2. A power supply problem. If you are using USB power only, bear in mind that a typical PC USB port cannot handle a Duet + 7" screen, and some USB ports may struggle to handle a Duet + 5" screen.

3. A fault has developed on your Duet WiFi. If you have ruled out (1) and (2), ask your supplier for a replacement under warranty. You will probably be asked to send your Duet WiFi to me for investigation.

SimonRafferty · 26 Oct 2016, 18:51

And the answer is….2 - well, kind of
It turns out that the firmware was running OK. Every time the ESP tries to connect (Blue LED Flashes), the power dips and the board reboots.
I've tried powering the board via a powered USB Hub, with PanelDUE disconnected - and Pronterface then connects.
But......
When you send M552 S1 to connect to the network, the same thing happens and it re-boots. Later I'll put a scope on the Vcc line to see.
Same happens powering the board from 24v.
I think therefore it is a power issue - the USB / On board regulator cannot supply enough juice for the ESP.

Si

dc42 · 26 Oct 2016, 19:00

It sounds like the ESP on your Duet WiFi needs to be replaced. Contact your supplier for a replacement board.

SimonRafferty · 26 Oct 2016, 20:18

That's a shame. Will do.

Thanks for your speedy help.

Si

dc42 · 27 Oct 2016, 07:01

One more thing to check: disconnect anything you have that might be drawing excessive power from the 3.3v rail, such as the Z probe and any endstop switches you have that use the centre pin of the endstop connector. It could be chat the 3.3V regulator is overheating.

SimonRafferty · 27 Oct 2016, 08:43

I put a scope on it last night. The 3.3v rail stays within a few mV - no spikes or drop-outs that I could see.

There's nothing getting obviously hot:

The hottest point top right, is the SD Socket
The alignment of the visual & thermal on this is a bit out - but you get the idea.

These were taken after the board had been powered (from 24v) for half an hour in a pretty cold workshop (16C)

Si

SimonRafferty · 27 Oct 2016, 09:19

I've disconnected the Z probe. None of my end-stops use the centre pin - but I disconnected them as well. No change.

The only thing that makes a difference is to comment out this line in Config.g
;M552 S1 ; Start WiFi

Then it's as normal, but without the Wifi (Obviously).

I think it's reasonable to conclude that the ESP is dodgy!

Si

dc42 · 27 Oct 2016, 09:33

Looks like the processor is warm and so is the 3.3V regulator. That is normal. What surprises me is that the crystal and the SD socket appear to be warmer than either.

elmoret · 27 Oct 2016, 12:58

I'd guess that's IR reflections from the room.

SimonRafferty · 28 Oct 2016, 13:37

My guess is the SD socket is 'hot' because of the residual heat from me holding the SD Card. It's perfectly possible I touched the Xtal at the same time! Mind you, they are only 2C warmer than ambient - so not exactly hot.

I'd like to thank David for his help and Roland from T3P3 for sending a new board (which works perfectly) so quickly. It arrived at mid-day today. I'll return the dodgy one to David by return.

Exemplary service all round!

Si