DuetWIFI migration

mhackney

Thanks David. I've done everything up to step 4. I can connect but it constantly drops the connection within a few seconds. Once this happens, the web interface won't connect but I can run from paneldue. I'll go through your procedure verbatim from scratch. Can I use the .bin files on the CF that came with the card?

dc42

Yes you can use the .bin files that came with the SD card. What we do to program a new board is this:

use SAM-BA to put the main firmware on the board
connect Pronterface via USB (note that the M552 S1 command in config.g is commented out, so it doesn't try to start the wifi module)
send M503 to make sure the SD card is working
send M997 S1:2 to install both DuetWiFiServer and DuetWebControl
send M552 S1 to start WiFi

mhackney

David,

So I started from scratch and followed the above to the letter EXCEPT I used the latest release you put on DropBox yesterday. I also made sure to disconnect USB and reboot the Duet between critical steps. My Pronterface log was clean:

[[language]]
>>>M997 S1:2
SENDING:M997 S1:2
 success
Erasing 295248 bytes...
Uploading file...
5% complete
10% complete
15% complete
20% complete
25% complete
30% complete
35% complete
40% complete
45% complete
50% complete
55% complete
60% complete
65% complete
70% complete
75% complete
80% complete
85% complete
90% complete
95% complete
Upload successful
Trying to connect at 460800 baud:
 success
Erasing 3125248 bytes...
Uploading file...
5% complete
10% complete
15% complete
20% complete
25% complete
30% complete
35% complete
40% complete
45% complete
50% complete
55% complete
60% complete
65% complete
70% complete
75% complete
80% complete
85% complete
90% complete
95% complete
Upload successful
>>>M552 S1
SENDING:M552 S1
WiFi server starting up
DuetWiFiServer version 1.01
Flash size 4194304, free RAM 31512 bytes, WiFi Vcc 3.34V, host name: Maxie, reset reason: Turned on by main processor
WiFi server connected to access point HackneyNet, IP=192.168.1.242

Upon success, I used an Incognito chrome browser so there was no cached cruft. It connected fine and I see that the firmware packages were updated:

Firmware Name:	RepRapFirmware for Duet WiFi
Firmware Version:	1.15-beta3 (2016-07-09)
Web Interface Version:	HTML: 1.11b-dc42, JS: 1.11b-dc42

Although it stayed connected a bit longer (30 seconds) I get the same AJAX communication error I've been getting:

Communication Error

An AJAX error has been reported, so the current session has been terminated.

Please check if your printer is still on and try to connect again.

Error reason: timeout, retries 1

Once this happens I can't reconnect via the web interface, I get this:

Error

Could not establish a connection to the Duet firmware! Please check your settings and try again.

However, I can use PanelDue with no problems.

This is all very similar to the experiences I had helping test Christian Hammacher's DWC code earlier this year to eradicate the disconnect problems many of us were having.
–--
So my next step is to use the original binaries - I saw you posted on another thread about this. I'll also capture a network trace with wireshark.

mhackney

Ok, I've now started from scratch as per your post above David - installed the pre-production firmware via SAM-BA, etc. I then installed the pre-production DWC.bin and DWiFiS.bin. Pronterface reported all of this installed successfully. Once these were installed, I setup Wireshark to capture traffic to my Duet WiFi's ethernet address. I then enabled WiFi in Pronterface.

I then launched an Incognito chrome browser and connected. I verified that all the pre-production version numbers were displayed on Settings->General. I feel confident that I have successfully back-graded everything to the pre-production binaries. I then homed from DWC. It lasted longer but I got a timeout and was able to capture the entire episode from initial turning on the WiFi to the failure. Since we can't attach files here, here's a link to the Wireshark trace on my dropbox: https://db.tt/FcFk6ZiA

mhackney

Note that when I capture network traces for my other Duets with external WiFi (TP-Link) I don't see any of the retransmissions and other issues seen in this trace. The printers sit next each other with very strong wifi signals to my router. I haven't had disconnects from any of the other 4 printers running earlier Duets with these TP-Links since helping Christian test his updates earlier this year that seem to have eradicated disconnect issues on that release.

mhackney

An update, after letting the disconnected board sit for 5+ minutes I tried reconnecting (press the Connect button). It actually did connect but then disconnected within 15 seconds. I did not get a Wireshark trace but in the future I'll run Wireshark before any similar attempts.

mhackney

Also if I turn off the WiFi server and re-enable it in Pronterface (M552 S0 / M552 S1), it recovers and allows me to reconnect (but then fails eventually again).

mhackney

and here is my M122 dump after the above:

[[language]]
SENDING:M122
Diagnostics
Used output buffers: 1 of 32 (3 max)
Platform Diagnostics:
Memory usage:
Program static ram used: -445744
Dynamic ram used: 57704
Recycled dynamic ram: 3016
Current stack ram used: 428696
Maximum stack ram used: 431780
Never used ram: 51548
Last reset 01:14:58 ago, cause: software
Error status: 0
[ERROR] Error status: 0

Bed probe heights: 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Free file entries: 10
SD card interface speed: 20.0MBytes/sec
SD card longest block write time: 0.0ms
Slowest main loop (seconds): 8.235451; fastest: 0.000000
Move Diagnostics:
MaxReps: 3, StepErrors: 0\. Underruns: 0
Heat Diagnostics:
Bed heater = 0, chamber heater = -1
GCodes Diagnostics:
Move available? no
Stack pointer: 0 of 5
macro is idle
http is idle
telnet is idle
serial is ready with "M122"
aux is idle
file is idle
Network Diagnostics:
WiFiServer is running
SPI underruns 0, overruns 0
Webserver Diagnostics:
HTTP sessions: 0 of 8

dc42

Michael, thanks fort all the tests you did and for the Wireshark dump. The dump show thart there were no transmissions from the Duet after the last successful status request, except for the usual [RST,ACK] that closed the connection.

Please can you try again, but this time monitor the connected wireless devices in the status page of your wifi router. I need to see whether the Duet is still a connected device after the error occurs and you find you are unable to connect. You will probably need to press a Refresh button in your router's web interface to see the latest list of connected devices.

mhackney

David, I posted this on another thread here but I think I discovered the problem was my network:

This is interesting, using some network monitoring tools and the admin console for my Verizon Quantum router I discovered that I had a few odd things going on on my network. Several devices which I could not identify were issuing a lot of packets and that seemed to be causing timeouts. Also, the Windows laptop that I VPN in into work was chewing up a lot of bandwidth. The Quantum router supports dual frequency but none of my devices were connecting the the 5GHz. So I created a new SSID for 5GHz and connect my lap worktop to it. Now, I haven't had a single disconnect in several hours from the Duet WiFi! And the Wireshark logs looks clean as a whistle.

However, to answer your question - after the disconnect, my router status page shows the Duet WiFi as disconnected. It was precisely because I was looking at the status page that led me to see the odd activity on my network as I worked through identifying each of the 28 devices connected! They add up quickly.

mhackney

David, I have now updated to the latest package of files (1.15-b3 etc) and things are quite stable.

I aslo tested the DWC.bin that I built with the shell script I wrote and it works fine too. I think you can include it for further testing. I put a copy on my dropbox here: https://dl.dropboxusercontent.com/u/45673517/dwcspiffs.sh

cheers,
Michael

dc42

I'm glad you solved it. I think I mentioned before in one of these threads that once the Duet becomes disconnected form the access point, it doesn't attempt to reconnect. That's why you had to stop and restart the wifi module in order to connect again. In the next release I'll add support for monitoring the connection status and attempt to re-connect if it becomes disconnected.

Thanks for the build script, I've added it t github.

mhackney

Great, thanks David.

I'm commissioning one of your IR probes on this Rostock Max with the Duet WiFi. I can do head to head comparisons with the FSRs. But I've been chasing an issue with FSRs on this machine as I described in an email I sent a few days ago.

Like all things, there are tradeoffs. I like the simplicity of mounting FSRs and the fact that the nozzle tip is the probe, but the downside is the "force".

dc42

The important things to achieve when not using the nozzle as the probe are:

1. Reduce varying effector tilt with XY position as far as possible, because such tilt changes the relative heights of the nozzle and the probe. I now have a bulls eye spirit level mounted on the effector so that I can see any tilt.

2. Mount the probe as close to the nozzle as is safe and practical to do so, to minimise the effect of that tilt. That's why I made the IR sensor shallow enough to fit below the heatsink of an E3Dv6 hot end.

Probing with the nozzle itself has the advantage of not being affected by effector tilt - although effector tilt is a sign of geometrical errors that will cause calibration inaccuracies anyway. But I'm not sure I would want to probe my PEI-covered bed with a hot nozzle.

mhackney

I'm almost done with my mount. I have a very non-standard setup with a CNCd effector and a modified E3D V6 that is water cooled and fitted with a Volcano. My mount is adjustable both up & down and in & out so I can dial in the position to optimize it. I'll post some photos once I have it complete.

Do you find the bulls eye spirit level is sensitive enough to show tilt?

dc42

The bulls eye spirit level was sensitive enough to reveal a tilt when the effector was close to one of the towers. The rods are accurately made to equal lengths, so I corrected it by rotating the carriage on the carriage truck.

mhackney

Here's the IR Probe mount I made for the TrickLaser MetalMax effector along with my custom water cooled E3D V6 hotend and volcano. The clearances look closer in the photos, I'll put Kapton tape on the backside of the board t be safe but I think I'm ok. The mounting system can slide up and down (slots in the printed red mount) and forward back (a slot in the mounting ear on the red mount) so I can fine adjust the probe position.

Next step is to wire it up to the Duet WiFi and get it running.

dc42

It looks like you could trim the component wire stubs on the back of the board a little more.

One concern I have with that cantilevered sensor mount is that the plastic part only needs to soften and warp very slightly to change the relative heights of the sensor and the nozzle.

mhackney

Thanks David, yes I did that and then Kapton tape. I'l keep an eye on the plastic part. It is mounted to aluminum - a big heat sink and it is very far away from the heater. Worse case I'll machine one in aluminum but I wanted to try this out quickly.

mhackney

David, I seem to be missing something trying to setup the IR probe. I followed the instructions on your blog. That all went well. https://miscsolutions.wordpress.com/mini-height-sensor-board/

When I get to the part to test and configure the probe I do as you say:

heat the nozzle (and bed)
home
lower the head until the nozzle touches the bed (paper test)
issue G92 Z0 (and I see Z set to 0 in DWC)
raise head 5mm
issue G30 S1
the Z value I read off in DWC is 1.0mm

I use this value to update my config.g:

M558 P1 X0 Y0 Z0 H5 F500 T12000
G31 P400 Z1.0

And reload config.g when asked.

Now, I home and run auto calibrate. When done, it seems like my Z=0 is set about 1mm above the bed. What am I missing?

FYI: bed is PrintBite