Quick question about boot order with an RPi and Duet 3
-
My understanding is that if I connect my RPi, I don't then have the SD card in the Duet main board. So the question is, when power is first applied to the machine, does config.g get read in immediately like it does in stand alone mode? If not, how long is the delay?
The scenario that concerns me is as follows. For whatever a reason a print goes haywire and bad things start to happen. I need to act fast so I hit my big red emergency stop button which kills all power thereby cancelling the print and turning off the motors, heaters, fans and everything else . The bad things cease to happen. I realise that the hot end is still full of molten plastic but the fans aren't running so I restore power. Config.g gets read in and the thermostatic fans start up within a second or so of restoring power so all is good.
BUT, if there is a significant delay in the fans not starting, heat creep will likely cause blockages in my heat breaks. My new hot end uses 6, Slice Engineering Mosquito style heat breaks which don't take kindly to rough handling (I know this from painful experience) and cost over £200 to replace so I dare not take any risks.
So to be more specific, if using an SBC, how long is the time before power is applied and any thermostatic fans starting up?
-
@deckingman said in Quick question about boot order with an RPi and Duet 3:
So the question is, when power is first applied to the machine, does config.g get read in immediately like it does in stand alone mode? If not, how long is the delay?
No there is a delay; would depend on the SBC and SD card in use (and other software also starting at boot, network delays might be a factor as well). 5-10 seconds or even more.
-
@bearer said in Quick question about boot order with an RPi and Duet 3:
No there is a delay; would depend on the SBC and SD card in use (and other software also starting at boot, network delays might be a factor as well). 5-10 seconds or even more.
Ahh, that's bad - very bad.......
OK. So how about if my emergency stop kills the 24V which kills the Duet but the RPi remains "live" because it's powered from a separate 5V supply? Will that a) prevent the delay and b) not have any adverse effects? That is to say will any software that is running on the RPI get screwed up if I interrupt power to the Duet boards?
-
I had this conversation here briefly the other day. With thermostatic fans running so I could time how long it took to recover, I power cycled the whole system and found it consistently takes just over 24 seconds for the fan to start again. This is with a Duet3 and Pi4 as the SBC.
edit: while I haven't timed it, resetting just the Duet3 seems to have about the same recovery time as if you weren't using a SBC
-
@ChrisP said in Quick question about boot order with an RPi and Duet 3:
..............and found it consistently takes just over 24 seconds for the fan to start again.....................
That's terrible!
-
@ChrisP said in Quick question about boot order with an RPi and Duet 3:
consistently takes just over 24 second
is that with the lite or desktop image?
-
Another alternative that I could live with at a push - can I configure these 12 volt fans to be always on? Older Duet boards used to have always on fan connectors but I'm not seeing anything like that on my Duet 3 expansion boards. I run 24V and the fans are 12V so I can't simply connect the fans to the input terminals of a board.
EDIT. If I connect the positive side of the fans to Vfan and the negative side to gnd on the input terminals of the board, that should effectively make them run as always on.
-
@deckingman said in Quick question about boot order with an RPi and Duet 3:
@ChrisP said in Quick question about boot order with an RPi and Duet 3:
..............and found it consistently takes just over 24 seconds for the fan to start again.....................
That's terrible!
Indeed,its not ideal. My suggestion was the possibility to have an SD with config in the D3 with just setup for the sensors and heaters with thermostatic fans, then wait for the SBC for the rest of the config.
@bearer said in Quick question about boot order with an RPi and Duet 3:
@ChrisP said in Quick question about boot order with an RPi and Duet 3:
consistently takes just over 24 second
is that with the lite or desktop image?
That is indeed with a desktop image. However, for users with that setup, that's the time. I'm also not sure how much that impacts as I don't think the service waits for the desktop to load before starting.
-
@ChrisP Is there a downside to keeping the RPi powered from a separate 5V supply (that's what I have in any case - just never yet connected it). Does anything get screwed up if power to the Duet board is interrupted in that case? Expanding on that, is there any downside (apart from using a small amount of electricity) in keeping the RPi permanently powered and just switching off the 24V to the Duet boards when not in use?
-
@ChrisP said in Quick question about boot order with an RPi and Duet 3:
I don't think the service waits for the desktop to load before starting.
No, it doesn't explicitly wait but it'll still be contending for the resources to start (and its not just the desktop). How big an impact I don't know, I tend to use the light image for all things pi and didn't time it, 10 seconds was just a stab in the dark, and 5 seconds an estimate for a faster SBC with SSD or something.
Maybe it could be worth looking into moving DCS up on the start priority, it wouldn't technically have to wait for networking to start, only storage really. DWS/DWC could start later.
-
@deckingman said in Quick question about boot order with an RPi and Duet 3:
Does anything get screwed up if power to the Duet board is interrupted in that case?
check out dc42 @ https://forum.duet3d.com/post/148283 and the rest of the thread. (tl;dr theoretical possiibilty)
edit: to ensure that's not a problem you could consider trying https://forum.duet3d.com/post/148888 but its not tested
is there any downside (apart from using a small amount of electricity) in keeping the RPi permanently powered and just switching off the 24V to the Duet boards when not in use?
the main wear would be flash writes (and possible cooling fan). but when the system is idle neither should be necessary, so not a huge factor.
-
Never trusted thermostatic fans for similar reasons. Corrupt SD = blocked hotend. Best thing would be to adjust the firmware so one fan connector defaults to on until the config is successfully read in. That would have the downside of not being able to take anything on that specific pin that was dangerous if on, such as heater or laser...
-
@DocTrucker said in Quick question about boot order with an RPi and Duet 3:
Best thing would be to adjust the firmware so one fan connector defaults to on untilthe config is successfully read in.
or like the Duet2, lacking pull downs; as such one could try removing some..
-
Forgot the main point of replying! I thought the Slice engineering hotends were designed to be ok in a fan failure, but clearly not. Or is that just due to the rest of your hotend setup?
-
@bearer said in Quick question about boot order with an RPi and Duet 3:
or like the Duet2, lacking pull downs; as such one could try removing some..
I've lost you a little there!
-
@DocTrucker The Slice Engineering claim is valid where there is forward movement of the filament. They have printed for many days with no fans - even without the copper heat sinks and just using the thin tubes - and with PLA which is the worse filament in this respect. BUT, (and it's a big BUT) if filament does not move forward and is static, then the filament itself will effectively bridge the heat break. That's just simply physics so no matter how effective a heat break is, you can still get heat creep through the filament itself if it is static.
-
@deckingman Is this with the Toolboar or the expansion board? Looks like the toolboard doesn't have pull downs on the output fets as such it should be possible to behave more like the Duet2?
-
@DocTrucker said in Quick question about boot order with an RPi and Duet 3:
@bearer said in Quick question about boot order with an RPi and Duet 3:
or like the Duet2, lacking pull downs; as such one could try removing some..
I've lost you a little there!
Duet2 Fan 1 will spin until the firmware loads and drives the pin low, the other outputs have pull down resistors to ensure outputs stay off until the firmware is running.
-
@DocTrucker Ref the always on fan thing - I think I'll do what I mentioned above. That is, connect the positive side of the fan(s) to a fan header, but connect the negative side to gnd on the input terminal or some other convenient non-PWM switched gnd point. Given that PWM switches the negative side then connecting the negative side of the fans to permanent gnd should make them always on should it not?
Edit. So the fans will run whenever DC voltage is present - no waiting for config.g or anything else to get read.
-
@deckingman yes, taking from a 12V regulated source straight to ground would be fine. I know wiring heaters in series isn't a great move but what about pairing up the fans and running from 24V?
At the end of the day if one fan has gone you've had a bad day, you may as well make it worse!
I concede Plan A - regulated 12V - is better. 
-
@bearer Thanks for the links. tl:dr but they seem to be related to powering the RPI from the Duet or vice versa, or powering over USB. But in my case, I have a completely separate 5V supply. The gnd side of all DC supplies are connected together and also connected to chassis earth. There is only one connection to mains and the earth side of that too is connected to chassis earth. Not my field of expertise by any means but I don't think there is much likelhood of ground loops? Yes? No?.
-
@deckingman said in Quick question about boot order with an RPi and Duet 3:
@bearer Thanks for the links. tl:dr but they seem to be related to powering the RPI from the Duet or vice versa, or powering over USB. But in my case, I have a completely separate 5V supply. The gnd side of all DC supplies are connected together and also connected to chassis earth. There is only one connection to mains and the earth side of that too is connected to chassis earth. Not my field of expertise by any means but I don't think there is much likelhood of ground loops? Yes? No?.
the point dc42 made was having a powered rPi with a unpowered Duet could theoretically cause damage to the rPi's SPI circutry.
the link to the isolator was a proposed solution to remedy that - getting full galvanic isolation as a side effect.
(but you'd still run the risk of ground loops with your setup IF say the 5v GND connection were to get interrupted; as is the case with all ground loops, they're contingent on a fault)
-
@DocTrucker said in Quick question about boot order with an RPi and Duet 3:
@deckingman .............. I know wiring heaters in series isn't a great move but what about pairing up the fans and running from 24V?......................
Ohh man. I did that - with the exception of wiring the fans in series
. That is to say, I forgot to move a jumper on one of expansion boards and put 24V on two expensive, near silent, Sunon Maglev fans. They didn't even attempt to turn - just gave a discrete "pop". Still waiting for the replacements to arrive.......... -
@bearer said in Quick question about boot order with an RPi and Duet 3:
the point dc42 made was having a powered rPi with a unpowered Duet could theoretically cause damage to the rPi's SPI circutry.
the link to the isolator was a proposed solution to remedy that - getting full galvanic isolation as a side effect.
Ahh, OK. Thanks. Sounds like that's best avoided then...
-
@deckingman said in Quick question about boot order with an RPi and Duet 3:
Sounds like that's best avoided then...
It is easily avoided by powering both from the same (5v) supply through the ribbon cable; direction doesn't matter, both will always be powered together. (just make sure to set the jumpers correctly)
I had other reasons for wanting to try full isolation.
