I think my Duet and/or Duex5 is/are dying

deckingman

Started having some problems and not sure what to do.

Problem 1 is something that I have reported before. It's not big issue and has been happening for about 6 months or more, but it seems to be getting worse. When the Duet first boots, either when first applying power or after editing config.g through DWC, it takes quite a few seconds for everything to initialise. This is audible in that the hot end (thermostatic) fan will run at full speed for between 4 and 10 seconds then settle to it's normal off state (when the hot end s cold of course). The other thing that happens is that establishing a connection to DWC will always times out and take several more seconds before it will connect. This is the M122 that I took immediately after turning on the Duet.

0_1541086487967_first boot M122.txt

The second problem is more recent and has so far only happened 3 times. It manifests itself by things running really slowly, or rather it can take up to 8 seconds between issue a command (such as jog z by 1mm) and the printer carrying out that command. The sequence of events leading up to the latest occurrence was, turn on printer, heat bed, home all. Then I just wanted to check the nozzle clearance at Z=0 so just pressed the Z-1 button and nothing happened - until 8 seconds later when the Z axis moved. On a previous occasion, this behaviour manifested itself at the start of a print (but after a successful home all). Cycling the power cures the problem.

This is the M122 report that I did when the printer exhibited this delayed movement.

0_1541087191777_Slow Movement m122.txt

What is noticeable (at least to me) is that there are 2610 Duex I2C errors. After cycling the power, there are zero I2C errors and everything works as normal.

I've checked for any loose connections but can't find anything. Not sure if it's relevant but all the axis motors are connected to the main Duet board and the Duex has all the extruders.

Firmware is this:

Firmware Name: RepRapFirmware for Duet 2 WiFi/Ethernet
Firmware Electronics: Duet Ethernet 1.0 or 1.01 + DueX5
Firmware Version: 2.01(RTOS) (2018-07-26b2)
Web Interface Version: 1.22.3

Could use some advice as to what else to check and if I need to replace a board, which one (Duet or Duex5). I think mine is the very first Duet Ethernet, or at least one of the very first and likewise the Duex5.

GizmotronX5000

I just had I2C errors myself on a Duex5. Using a shorter (about 3 inches) and thicker ground wire between the Duet and Duex5 solved the problem completely for me. How DC42 first discovered that solution I'll never fully understand.

I had the same slow movement problem and power cycling the system fixed it, for a little while. Haven't had it happen since.

Here is the thread:
https://forum.duet3d.com/topic/7472/printer-slowly-responds-to-g-code

deckingman

@gizmotronx5000 As per my post, my Duet and Duex 5 have been installed since they first came out - probably coming up 2 years or thereabouts without any issue until recently. I saw your thread - my Duet and Duex are installed end to end with about a 10mm gap between them so the link wires are less than an inch long and 2.5mm solid copper. Those wiring requirements are clearly stated in the wiki https://duet3d.dozuki.com/Wiki/Duex2_and_Duex5_Features

So thanks but it's not an installation issue.

GizmotronX5000

It may still be worth reseating those wires. I hadn't changed any wiring for the last few months and the I2C problem spontaneously arose, but went away after replacing the wire. I'm sure I didn't bump anything as I have the board securely located in an enclosure. Everything behaved normally otherwise. I'll keep checking back here for other updates though.

deckingman

@gizmotronx5000 Again, as per my OP, I've checked and there are no loose wires or bad crimps any where - including the Vin and Gnd links. If I tighten the terminal block screws any more, they are likely to break off the board. I've also checked for any signs of oxidisation but there is none - the contacts are all nice and shiny.

In my case, the problem occurs very rarely - about once every 2 or 3 weeks and a power cycle always cures it. The slow boot time is another problem and is always present and getting worse. As both boards were "pre general sale" boards and about 2 years old, I'm thinking that one or other may have developed a fault and need replacing. The question is, which one or both?

dc42

Have you tried replacing the SD card?

I2C errors are usually caused by a bad ground connection between the two VIN terminal blocks; but it's not impossible that the problem is a dodgy ribbon cable, or an intermittent problem on the DueX5.

deckingman

@dc42 said in I think my Duet and/or Duex5 is/are dying:

Have you tried replacing the SD card?

I2C errors are usually caused by a bad ground connection between the two VIN terminal blocks; but it's not impossible that the problem is a dodgy ribbon cable, or an intermittent problem on the DueX5.

Yes I tried numerous SD cards a few months ago when the slow initialising at power up and when editing config.g started to become bothersome. I tried a range of different card sizes, both new and use and formatted to use different block sizes. None of which made any difference.

To save me having to keep describing the connections, here is a picture

For the slow initialisation, I just tried disconnecting the ribbon cable from the Duet to the Duex and turning on the power - made no difference. It still took a few seconds to initialise then a further few seconds for DWC to connect. I'm guessing that rules out the ribbon cable and Duex 5.

gavatron3000

Are the supply cable and jumper between board both solid core?
Im personally not a fan of solid core conductors in screw terminals due to reliability, all industrial control stuff I work with is all bootlaced multistrand or flex as we call it. I groan when I do work on a house wired with solid there's always something that has a drama with it. Might be worth changing cables like the other bloke said

Edit: there also might be a possibility that the terminal solder joints have fatigued due to jumper cable not being able to flex and flexing the terminals ever so slightly over time. Just another thing to check I guess

BPisLife

Which firmware are you using? I noticed that at least in 2.01 or any of the 2 series firmware releases (haven’t tried the very latest yet) that it does seem slower. One of my printer even has noticeable jitters in all motor movements as if the motor control is not buffered and starts and stop rapidly.

I should point out that I too use duetwifi with a duetx5.

As far the specific problems you are seeing, I have not seen those. If the board is older visually inspect any electrolytic caps (if the board has any) as they do die overtime and their end of life is usually indicated with a leaking electrolyte.

dc42

Can you pin down which module is holding it up? For example, if you comment out the M552 S1 command in config.g, does it start up at normal speed (as indicated by the hot end fan)?

You could also try connecting a PC terminal emulator via USB as soon as you restart the Duet, in case it is trying to send a message to USB. Take the usual precautions against USB ground loops.

deckingman

@gavatron3000 No space to fit ferrules so I'll stick with the solid core as per the Wiki Instructions.

deckingman

@bpislife Firmware is as per my OP. No jitters or movement issues other than when the 8 second delay problem manifests itself (once every 3 or 4 weeks).

deckingman

@dc42 said in I think my Duet and/or Duex5 is/are dying:

Can you pin down which module is holding it up? For example, if you comment out the M552 S1 command in config.g, does it start up at normal speed (as indicated by the hot end fan)?

You could also try connecting a PC terminal emulator via USB as soon as you restart the Duet, in case it is trying to send a message to USB. Take the usual precautions against USB ground loops.

Thanks - I'll give those suggestions a go.

deckingman

I've just tried commenting out M552 to disable the network and it makes no difference to the slow initialisation (as indicated audibly by the hot end fan). I've also pulled out the daughter boards - I have PT 100 daughter board on the duet and a thermocouple daughter board on the Duex5 (but with nothing connected to it). They make no difference either. I can't tell if any of that makes any difference to the slow movement as that only happens once every 3 or 4 weeks and the printer is behaving itself right now.

Could you expand on what is meant by connecting a PC terminal emulator via USB?

The other thing I forgot to mention which might be relevant, is that it always does a sort of double blip of the hot end fan. That is to say, when power is first applied, the fan will run for between about 2 and 8 seconds, then it stops for about a second, then it does second run for maybe a second or two before stopping. Is that normal?

dc42

Hi Ian,

No the double blip is not normal. It sounds like it is resetting twice. After power up, what does M122 report as the reset reason?

By chance, today I was debugging a startup issue in our new automatic test equipment firmware, and I found that the order in which things were initialised was causing the watchdog timer to time out. The same issue potentially affects RRF, so I've applied the fix to RRF too. It will be in RRF 2.02RC4. So please try it when I release it. Although this could fix the startup issues, it's not connected with the slow running issue - that one is related to the I2C errors.

deckingman

@dc42 Hi David,

The M122 immediately after start is linked in my OP. It's the first one entitled "first boot m122". I appreciate that if you are on your tablet, you'll not be able to scroll up to see it so here it is again. 0_1541180301298_first boot M122.txt

TBH, it's done that "double blip" thing for as long as I can remember but usually, the fan just runs for about a second or two each time. It's just lately that the fan can run for much longer - up to 8 seconds.

I'll try 2.02 RC4 when you release it and let you know if that improves anything.

Many thanks

dc42

@deckingman said in I think my Duet and/or Duex5 is/are dying:

Could you expand on what is meant by connecting a PC terminal emulator via USB?

I meant connect the Duet USB port to a PC and use a terminal emulator program (e.g. YAT) on the PC to connect to the Duet. Just like the way we suggest communicating with a Duet WiFi to set up the SSID and password.

brunofporto

@deckingman said in I think my Duet and/or Duex5 is/are dying:

Aren't those jump wires too thin?

I had some issues with my PC memory from time to time. After some months if a move the cabinet then the memory had issues.

I solve that by cleaning all contacts and reinsert the memory. My hypothesis is that it oxidized a thin layer around the connection and when I moved the PC it reallocate slightly the connections causing the issue.

Maybe the ribbon cable connection have this issue. Try cleaning them with some proper product (not sure the name in English, here is "contact cleaner").

deckingman

@brunofporto said in I think my Duet and/or Duex5 is/are dying:

@deckingman said in I think my Duet and/or Duex5 is/are dying:

Aren't those jump wires too thin?

They are good for at least 20Amps, so 480 watts@24V. Given that I only use a 200Watt PSU for entire printer and the Duex only has 5 extruders and a few fans connected to it, I'd say they are more than adequate.

elmoret

Those brown wires are good for 20A? generally 20A means at least 14ga. Those don't look like 14ga wires.

I don't think that's your problem, but I wouldn't put 20A through what looks like 20ga wires.