ESP Failed - always "WiFi module is being started" [Fixed]

[Willl]

I have been using Duet2 for a while.

when I brought it the wifi was a bit unstable, but I put new solder on the joints and it was happy for about a year.

Today, between prints and without removing power while my printer and laptop were sharing a isolated network router (no other devices and no uplink) the printer stopped responding to the network. The router shows only the laptop and no other devices.

Usually (IIRC) the ESP blinks very briefly on power and the flashes as it looks for network going solid when it has a connection.

Now it is always dark.

I can connect to the printer via USB and issue commands.

The firmware has not been changed recently, certainly not between the working and the failed state.

M115 gives :

FIRMWARE_NAME: RepRapFirmware for Duet 2 WiFi/Ethernet FIRMWARE_VERSION: 3.1.1 ELECTRONICS: Duet WiFi 1.02 or later FIRMWARE_DATE: 2020-05-19b2

If I use M552 S0 I get

WiFi module is being started

If I try to reinstall the firmware already on the SDCard I get

Trying to connect at 230400 baud: failed
Trying to connect at 115200 baud: failed
Trying to connect at 74880 baud: failed
Trying to connect at 9600 baud: failed
Error: Installation failed due to comm write error

I guess I am out of warranty. Is there anything I can check?

[Willl]

Very weird:

There is a bodge wire between the SDCard and the ESP ground which looks poorly soldered, the sdcard end tests as ground, the ESP end tests as NC.

Taking a wire from the connector to the esp shell makes it wake up and I can then start the wifi from the USB.

Even more oddly, now woken it continues to operate!

[Phaedrux]

Can you post some photos?

[dc42]

@Willl, when and where did you buy the Duet?

[Willl]

@phaedrux

Its not terribly easy to capture what i mean, but this enameled wire doesn't seem to cause continuity between the ESP pad and the SDCard shell. Where the wire enters the solder blob the solder is pulled around the wire rather than wetted to it. I can only assume that the ESP is getting ground from elsewhere, at least some of the time. Perhaps there is a partially soldered thermal pad, perhaps there is a pin that is pulled low by the duet board hard enough to act as the system ground through the parasitic diodes on the GPIO?

[Willl]

@dc42 I think I brought it direct, but it may have been via E3D. It would have been around this time two years ago (probably just over two years ago based on my forum account !)

I should really have RMA'd it when I first had issues, but I had a lot of printing to do and couldn't go back to the old board having butchered the wires.

I know i'm well in to "you're on your own now" land, but it is at least a land I recognize If I had to replace the ESP I do have the tools (and the ESP!) to do so - as long as after doing so I could just tell the board to install firmware and it would do it.

What I really wanted to avoid is soldering it in and then finding that it needs some bootstrapping firmware - but it seems more like a power issue than a dead ESP issue, so I think i may have dodged that as an option anyway.

[dc42]

@willl, it looks it me that it just needs that joint resoldering. If you are not confident enough to do it, do you know anyone who is?

Which country are you in?

[Willl]

@dc42

Thanks. I think that's the conclusion I was coming to, but its very reassuring to hear it from you

Its well in my capabilities so i'll get on with it now!

I'm UK based.

[Willl]

@dc42

The old wire (also pictured) looks even more suspect when removed - so I'm pretty happy I have the culprit.

Can you confirm that the 1.04 board needed this link to bring ground to the ESP? I had expected it to be providing ground to the SDCard shell, which is why I didn't pay it too much attention before.

[dc42]

@willl said in ESP Failed - always "WiFi module is being started":

Can you confirm that the 1.04 board needed this link to bring ground to the ESP?

Yes. There was a ground trace missing on the original 1.04 PCB. This was fixed in 1.04a and later revisions.