Thermistor Failure - Kill Printer

Is it possible to add a feature to the firmware that would go something like if extruder/heated bed changes temperature by X amount in Y seconds, kill printer.

I had a thermistor fail and show temperature swings of 130C but the cartridge happily continued supplying more heat because these swings were going below set temp. I would think it would be fairly easy to monitor this and now with the all metal hotends and higher temp materials coming into play and almost everyone using thermistor which are rated at only 300C I think it would be a great feature to have.

It would be a lot more safe to mount a termo fuse with the extruder. But of course you could have both.

Repetier, can we have a reply to this please. It is important for safety.
Can you suggest a couple of lines of code that would catch a thermistor being pulled out.

Quote
justcurious
It would be a lot more safe to mount a termo fuse with the extruder. But of course you could have both.

Thermal fuses have been discussed in another thread. First, I can not find one with a tripping temp about 280. The one that trips at 280 is rated to sustain 230 so anything in between is a gamble. As has been discussed there, yes you can mount it not right on the hotend but then you are either waiting for it to heat the whole hotend up and maybe get the heater area up to melting temp of peek or whatever before it trips. Plus I vary my temps over 100c for different materials, I don't have much to worry about with the E3D but even with a j-head you can change temps nearly 70c by going from PLA to ABS so how can you possibly fuse that reliably?

With this firmware feature, this should be caught immediately and kill the printer and not raise the temp any higher than what it was printing at which is why this would be good.

Quote
tjb1

Quote
justcurious
It would be a lot more safe to mount a termo fuse with the extruder. But of course you could have both.

Thermal fuses have been discussed in another thread. First, I can not find one with a tripping temp about 280. The one that trips at 280 is rated to sustain 230 so anything in between is a gamble. As has been discussed there, yes you can mount it not right on the hotend but then you are either waiting for it to heat the whole hotend up and maybe get the heater area up to melting temp of peek or whatever before it trips. Plus I vary my temps over 100c for different materials, I don't have much to worry about with the E3D but even with a j-head you can change temps nearly 70c by going from PLA to ABS so how can you possibly fuse that reliably?

With this firmware feature, this should be caught immediately and kill the printer and not raise the temp any higher than what it was printing at which is why this would be good.

I am not in any way against any FW features but many things are to be considered if this is to be a reliable solution.
1) how can you be sure that fw is running ok when this situation is met, it is not only thermistors that fail. If cpu is runawayy/hang no action is taken.
2)it is a common mistake that a "couple of lines" could solve any problem. Sw is a tool that has a limit with humans way of using it and can be very complex and unreliable and do "mysterious things" and have influence on functionality that was not supposed to be related due to the way functionality is sometimes added seqiential.

It is correct that it is not easy to find a fuse with extreme temperatures, but maybe that is not a real problem? Fuse would normally be mounted on surface of extruder and here temperature is probably far from the internal thermistor temperature. Fuse is never supposed to trigger so you do not need to worry of daily use of different temperatures but only runaways and if fuse trigger you are probably close to dangerous temperatures anyway.

Quote
justcurious

Quote
tjb1

Quote
justcurious
It would be a lot more safe to mount a termo fuse with the extruder. But of course you could have both.

Thermal fuses have been discussed in another thread. First, I can not find one with a tripping temp about 280. The one that trips at 280 is rated to sustain 230 so anything in between is a gamble. As has been discussed there, yes you can mount it not right on the hotend but then you are either waiting for it to heat the whole hotend up and maybe get the heater area up to melting temp of peek or whatever before it trips. Plus I vary my temps over 100c for different materials, I don't have much to worry about with the E3D but even with a j-head you can change temps nearly 70c by going from PLA to ABS so how can you possibly fuse that reliably?

With this firmware feature, this should be caught immediately and kill the printer and not raise the temp any higher than what it was printing at which is why this would be good.

I am not in any way against any FW features but many things are to be considered if this is to be a reliable solution.
1) how can you be sure that fw is running ok when this situation is met, it is not only thermistors that fail. If cpu is runawayy/hang no action is taken.
2)it is a common mistake that a "couple of lines" could solve any problem. Sw is a tool that has a limit with humans way of using it and can be very complex and unreliable and do "mysterious things" and have influence on functionality that was not supposed to be related due to the way functionality is sometimes added seqiential.

It is correct that it is not easy to find a fuse with extreme temperatures, but maybe that is not a real problem? Fuse would normally be mounted on surface of extruder and here temperature is probably far from the internal thermistor temperature. Fuse is never supposed to trigger so you do not need to worry of daily use of different temperatures but only runaways and if fuse trigger you are probably close to dangerous temperatures anyway.

From my search, there are no thermal fuses that have a trigger above 280. The 280c trigger can only sustain 230c, just like the other thread this is a bad idea. If you can't mount it in the BLOCK that cartridge/resistor is in then its a waste of time. There will be too many variables mounting the fuse away from the cartridge in an area where it will be safe from the temps.

Regardless, I don't think I have ever had the arduino hang (where this code would be running...not on the computer). I've had computer lockups but the computer does not monitor and control the temps...at least with Ramps it doesn't. If you want to trust a fuse in a bad location, go for it. I would like a firmware monitor.

If it would be only a couple of lines I would have added it long ago.

How do we know temperature should rise or fall? The firmware has no ram to store the temperature history. It only knows the current temperature. And even with a history you would need some analysis due to the fact that turning heaters takes some time to reach thermistors. And if the read temperature changes it does not make finding a simple (should take no time, we need time for nice path plannings etc) and reliable. We have some tests for high and low temperatures - that is quite easy to implement.

The only solution I see that probably would work is running 20-60 (extruder/bed) second tests. Store time and temperature and integrate power over time with some unknown factors that should result in the reached temperature and test it against last temperature. But here we would also need large confidence intervals to make sure print does not stop from false signals which also gives room for undetected failures. And the extra parameter would need to be found for each extruder as delay, cooling, room temperature etc. plays a role. We could make it a bit easier in that we only look for rising temperatures. So if we know, including safety margin, the temperature should have raised and it went down the thermistor must went defect. I think this is the only dangerous case we need to consider - heating up endlessly.

---

Repetier-Software - the home of Repetier-Host (Windows, Linux and Mac OS X) and Repetier-Firmware.
Repetier-Server - the solution to control your printer from everywhere.
Visit us on Facebook and Twitter!

Quote
repetier
So if we know, including safety margin, the temperature should have raised and it went down the thermistor must went defect. I think this is the only dangerous case we need to consider - heating up endlessly.

I agree that the most important scenarie is to have a registration of non raising/falling temperature if full power is switched to the heater over time, it could be a challenge to foresee this both outside and inside PID-regulation but with some care probably possible. however, thermistor could be defect, shortcircuit or disconnected and show different states of temperature scenarie.

Edited 1 time(s). Last edit at 01/31/2014 10:30AM by justcurious.

Quote
repetier
If it would be only a couple of lines I would have added it long ago.

How do we know temperature should rise or fall? The firmware has no ram to store the temperature history. It only knows the current temperature. And even with a history you would need some analysis due to the fact that turning heaters takes some time to reach thermistors. And if the read temperature changes it does not make finding a simple (should take no time, we need time for nice path plannings etc) and reliable. We have some tests for high and low temperatures - that is quite easy to implement.

The only solution I see that probably would work is running 20-60 (extruder/bed) second tests. Store time and temperature and integrate power over time with some unknown factors that should result in the reached temperature and test it against last temperature. But here we would also need large confidence intervals to make sure print does not stop from false signals which also gives room for undetected failures. And the extra parameter would need to be found for each extruder as delay, cooling, room temperature etc. plays a role. We could make it a bit easier in that we only look for rising temperatures. So if we know, including safety margin, the temperature should have raised and it went down the thermistor must went defect. I think this is the only dangerous case we need to consider - heating up endlessly.

Attached is a picture of the issue I had, you can see swings of 130c in under 5 seconds in the reported temp from the thermistor. I don't think any hotend is going to raise 130c in under 5 seconds but these variables could be left to the user. Since it thinks it dropped well below set temp, it continued to heat.

My thoughts are to maybe store temp readings for 5 seconds and if any points in that has changed greater than X value, kill printer like an e-stop from the host would act.

I don't think you should be worrying about thermistor reporting higher temps now because say if it reported 400c, well the logic won't power the hotend.

We need a solution for just one case. It is when the hardware is functioning correctly but the thermistor is pulled out of the hot end. The thermistor temperature suddenly drops and full power is kept on the hot end.

This happened to me when a print became detached and the ball of plastic spaghetti going round with the hot end pulled the thermistor out. It charred the plastic and filled my living room with smoke. There could have been a fire but fortunately I caught it in time.

Other checks are already in the firware, e.g. thermistor open circuit / exceeding max temp / min extrude temp. What we want is something else.

IF the heater is on and IF the hotend temperature drops more than a certain rate over a certain period of time, then it disables the heater.

It is not meant to be failsafe or replace other solutions to prevent fires but it would be a big improvement to what we have now.

Edited 1 time(s). Last edit at 01/31/2014 12:44PM by dave3d.

For the thermistor falling out scenario, i think here are 2 parts:
1. A "guard" temperature could be implemented and very easy to monitor. A simple way is to have a user defined parameter which can guard for +/- X degrees from the set temperature, and this would become active only once the initial target temperature is reached and started printing. This could be say a 3 strikes and your out style, if during 3 consecutive X duration of samplings and it is still under then stop/pause/whatever. If thermistor falls out mid print, the read temperature would drop and be caught. This could also serve as an extra safeguard for those occasions when something changes in the environment and it drops too far and would start to strip filament.

2. The rise/fall as Repetier above mentioned when the heater is meant to be heating or cooling down is hard. Possibly the important scenario is where the termistor fell out before the initial heating (i.e. fall out during homing). For those of us now using deadtime (which i really like btw) we already know the delay before we expect something should start to happen, at least initially when starting a print, and a prev temp and current temp could easily be compared after that time + margin. Would a thermistor fall out during heatup?


Or course, this all assumes the termistor is functioning perfectly to start with. I always watch extruder heat up and make sure the first layer gets started ok.

I think we have 2 important cases.

1. Thermistor dropped out during print
2. Thermistor was already dropped before print

Of course a attentive user would see 2. at the start and act on his own, but some may simply start and leave when they trust the printer. For my part I wait at least for the first layer as it is the most frequent failure point.

1. would cause the heater to go on for ever. After having temperature reached we could limit the full power time to x seconds. Longer periods are then a defect. When target temperature was reached that time can be reduced to 30 seconds. Works especially good with dead time/bang bang as it has only power on and off. PID changes power permanently, here we could only safely detect full power periods happening when control range is left, which is 20 degrees deeper.
For case 2. we need also to guard the time until temp. is reached. That could be done with a second full power period for temp. change which is long enough to cover the first heating up range.

Thinking of it, we have already 2 modes of heating up. Outside control range where we only have full power and off. So we could take one time for that and control range where we can narrow down the period. The only assumption here is that thermistor and heater are decoupled. The example of tjb1 was a bit different. Also it might be the high temperatures he used (310°C) could have damaged the thermistor. But even that example showed nearly 2 minutes of full power and no extrude should need 2 minutes when temperature was reached.

---

Repetier-Software - the home of Repetier-Host (Windows, Linux and Mac OS X) and Repetier-Firmware.
Repetier-Server - the solution to control your printer from everywhere.
Visit us on Facebook and Twitter!

is there a thing instead to check for software block? like if the computer crashes or it disconnects? i don't think it's good for the printer to be idle at 215 °C for 10 hours without extruding either...