Marlin 115 tried to burn my house down during PID tuning #8103
Comments
|
AUTOTUNE does not respect MAXTEMP. If you set 500C ... KABUUUM !!!! |
|
I did:
M303 E0 S205 C10
The issue isn't that I set a high number, it's that the thermistor went offline (it read -14C) so the firmware kept pushing amps to the heater even though it was probably close to 400C by the time I could run downstairs and manually switch off the power. The bottom ~20mm of PTFE in my throat was in a smoking puddle on the build plate....
This is a serious safety concern.
… On Oct 26, 2017, at 12:36 PM, Guilherme S. Blanco ***@***.***> wrote:
AUTOTUNE does not respect MAXTEMP. If you set 500C ... KABUUUM !!!!
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Say ' THANK YOU ' to marlin Buy thermocouple type k or pt100 ; i suggest PT100 because all hotend have the 3 mm holes for it , volcano , or normal , and no limits of temp I love thermistors but not on a machine that can shake while 30 hours a pair of nude wires !!!!!!!!!! |
|
I agree, just get a thermocouple, or PT100, these are sensors designed for high temperature, and are resilient to vibration. Or get an E3D V6 hotend, which have temperature cartridges |
|
Thanks for responding. This is a SERIOUS safety concern. I prefer infinity min temp crashes to one fire.
I do agree about the thermocouple - I already have parts on the way to build M3 cartridge type thermistors for this application. But even when I get those, it does not mean I want to run the risk of a fire under any circumstance just to avoid losing a few prints here and there....
This is a severe issue that needs to be addressed by both hardware and software fixes.
… On Oct 26, 2017, at 5:54 PM, studiodyne ***@***.***> wrote:
Say ' THANK YOU ' to marlin
Now you know , what is the maker life that have thermistors and these eternal wiring short circuit between nude wires or on the hotend
Buy thermocouple type k or pt100 ; i suggest PT100 because all hotend have the 3 mm holes for it , volcano , or normal , and no limits of temp
and ....... NO PROBLEMS OF ETERNAL MIN TEMP BRUTAL CRASHES
I love thermistors but not on a machine that can shake while 30 hours a pair of nude wires !!!!!!!!!!
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
On Oct 26, 2017, at 6:13 PM, Jarrod Smith ***@***.***> wrote:
But even when I get those, it does not mean I want to run the risk of a fire under any circumstance just to avoid losing a few prints here and there....
In hindsight what I wrote there doesn't even make sense. But I think you get my point. The good news is that we all appear to be in agreement that the thermistor issue is very real and annoying. I think that adds more weight to the argument for a software fix. Otherwise you are going to have some newbies like me buy a $300 printer off aliexpress with those crap thermistors and burn half his house down with it.
|
|
So... It is impossible to please everybody. The thermal protections numbers for THERMAL_PROTECTION_HOTENDS and THERMAL_PROTECTION_BED are much too tight for me. I always end up relaxing them significantly for my printers. But the people that did the work want them that tight and I respect their opinions. They know a lot more about the subject than I do. If you are worried about burning your house down, you probably should only use the printer while you are watching it. And you should probably only use it with the printer placed outside in the middle of a concrete driveway. Maybe that isn't enough. Maybe you should stack some concrete blocks around it to make sure that if anything catches fire it is isolated and the fire can't spread. Also, it might be wise to put a 5 amp fuse on the circuit where you are plugging in the printer to help limit what can go wrong. While you are doing those things we can continue to discuss other ways the firmware can help you have confidence in the safety of using your new AliExpress printer! |
|
or use dual thermistors on the hotend... not sure Marlin has support for it ? |
|
Why the hostility? I think a request to obey Max Temp and Min Temp during PID tuning is reasonable.
… On Oct 26, 2017, at 6:32 PM, Roxy-3D ***@***.***> wrote:
So... It is impossible to please everybody. The thermal protections numbers for THERMAL_PROTECTION_HOTENDS and THERMAL_PROTECTION_BED are much too tight for me. I always end up relaxing them significantly for my printers. But the people that did the work want them that tight and I respect their opinions.
If you are worried about burning your house down, you probably should only use the printer while you are watching it. And you should probably only use it with the printer placed outside in the middle of a concrete driveway. Maybe that isn't enough. Maybe you should stack some concrete blocks around it to make sure anything that catches fire is isolated and the fire can't spread. Also, it might be wise to put a 5 amp fuse on the circuit where you are plugging in the printer to help limit what can go wrong.
While you are doing those things we can continue to discuss other ways the firmware can help you have a confidence in the safety of your using your new AliExpress printer!
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
Did M303 REALLY try to burn your house down? You are basically accusing the author (and everybody that has contributed to the M303 code) of intentionally trying to cause you harm. Perhaps it would make sense to change the title to something like "Suggested improvement for M303's thermal behavior" |
|
STOP ! Yes , we must work on several wiring problems and your issue will be seriously explored Give us , all , the circonstances , to make this crash M303 is a child of alarm function and use the temp alarm that all use in print and preheating The only possibility is , the wiring diagram , you have connected the heater right to the left and the left to the right , and all without THERMAL PROTECTION No war here , the only war , is to have the first perfect layer with UBL loll |
|
Thank you. If I was impolite in my wording of the subject line, it was because this incident freaked me out. I hope all you reasonable and generous people can understand that. That does not in any way diminish the otherwise extremely clever bit of software you've managed to put together so don't read any more into it than that.
… On Oct 26, 2017, at 7:40 PM, studiodyne ***@***.***> wrote:
STOP !
The only thing we want to say , is our experience of marlin and the brutality of the temperature alarm algorithms
If temps make shit , print stop
And all here knows enough marlin to say , that it's possible there is an issue about this , but , we are all sure that , before to bug , we have a hardware that don't give enough stable value
Here , we are all sure that if you haven't theses thermistors 'sorry ' these nude wires that touch the hotend and receive your 3G/4G cell phone waves and more lolllllllll , you never had this issue
Yes , we must work on several wiring problems and your issue will be seriously explored
Give us , all , the circonstances , to make this crash
In a perfect order , because , we all want to try to make this at home , without your hardware , but to have the same exact situation , and after we will say if it's an issue
No war here , the only war , is to have the first perfect layer with UBL loll
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
From another perspective.... Everybody here has spent countless hours trying to make things better for the generic user. My biggest contribution is the UBL Bed Leveling. I did EVERYTHING possible to make it SAFE so it doesn't needlessly hurt anybody's machine. There are 1000 different ways the bed leveling and mesh editing software can drive a nozzle into the bed and damage the user's machine. That still does happen occasionally. But I've done everything in my power to make it not happen. The people that work on the thermal code are at least that contentious. If people are volunteering their time (and expertise)... It is best to proceed gently and make them want to help you. With that understanding... I suspect somebody is going to find the problem interesting and be willing to add some safety checks to M303. |
|
I know that everyone who has contributed has put in many hours to make Marlin better and safe and I appreciate their effort. However, I found the pid auto-tuning code and it seems to be a little rough (one confusing 200+ line function). I'm interested in contributing and will look around and see if I can find the issue. |
|
If people are volunteering their time (and expertise)... It is best to proceed gently and make them want to help you.
Roxy-3D, thank you for your contributions (seriously). We all want to be recognized for our hard work, that is human nature.
Since something I said seems to have really rubbed you the wrong way, I'm going to get very specific so you and others who I may have put off will understand where I'm coming from. I am a person who watched their printer nearly go up in flames. Politeness was not at the top of mind when I came here to alert you all to an issue that could potentially cause loss of property or even life. Getting your attention was. My frank tone is not coming from a place of anger, or ego, or spite. It's out of concern that the next person who uncovers this issue may not be in a position to intervene as quickly as I was.
If someone reading this has the expertise to diagnose and fix this, don't do it for me - I know how to avoid it. Do it for the next guy or gal, who doesn't.
And thank you all for an amazing piece of software. I authored some open source software back in the '90s, so I fully understand the commitment. I once had someone at a LUG meeting tell me to my face that my software was a piece of shit. And later I got threatened with a lawsuit by a publicly traded company who decided I was infringing on their trademark by supporting their products in Linux.
This is not that.
|
|
Thank you Rowan. If I can provide any info from my end, let me know.
… On Oct 27, 2017, at 1:09 AM, Rowan Meara ***@***.***> wrote:
I know that everyone who has contributed has put in many hours to make Marlin better and safe and I appreciate their effort. However, I found the pid auto-tuning code and it seems to be a little rough (one confusing 200+ line function). I'm interested in contributing and will look around and see if I can find the issue.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
@Roxy-3D Note : ThinkyHead added my FwRetract revision PR 😄 , if you know how to make an UBL/FW issue , try it now please , and keep the gcode file , and see , if there are some unwanted G10/G11/G10S1 somewhere |
|
I have tried M303 with bed ' STATERELAY and 600w bed ' , i have tried with thermocouple 1 et 2 , with my ' turbo hotends = Volcano + E3V6 + Flexxion holder + 60w cartridge ' , and all works Note : with my cartridge , It's not easy to reach 400 degrees , it takes time , and risk of fire is very long on my config and i'm sure , it's not your chinese 40w hotend that can reach more than 300 degrees lolllllllllllll Works perfect Contributing to marlin ,or other sofware is a big donation of time , every line of code generate 1500 mistakes and errors |
|
This might be wrong because I've never really looked at this code base before. I believe your problem occurred because pid_autotune directly sets the PWM values while setting everything else to indicate that the heaters are off (it calls disable_all_heaters when it starts so no outside error checking occurs). Also, pid_autotune has a 2 minute timeout but just returns from the function when that timeout is hit which means that the heaters are not disabled and an error is not thrown. I believe that the thermal runaway issues could be partially fixed by adding a line to pid_autotune's timeout behavior that throws a temperature error. The downside of this approach is that it only takes effect after two minutes. Edit: After looking at the code some more, it seems that fixing the sanity checking errors would be difficult because it would require significantly changing the sanity checking in Temperature::ISR (interrupt method). The changes would break the conditional compilation. Another option is essentially to have duplicated error checking code in the pid_autotune but that would increase the binary size and separate the error checking logic. |
|
I run my printer near an open can of gasoline... is that an issue? |
|
Oh !!! No temp alarm on M303 😲 .... |
|
If you're right .. this must be quickly fixed |
|
I think I found a pretty noninvasive fix so I will probably submit a pull request tomorrow. |
If the problem is reproducible in this order, it's not M303s fault. Min temp error should kill the machine in the moment the sensor is disconnected. Are you sure this downloaded binary is produced with the appended config? There is one way to disable min temp errors - setting mintemp to a value lower than the lowest value in the thermistor table. |
|
hex ? Or marlin 1.1.5 ? |
|
@RowanMeara , do you really have identified M303 is not under thermal prot ? I think , the only pr to make is to show a warning message about the needed presence of human during this tuning , and it's all |
|
Think i got it. |
|
On Oct 27, 2017, at 6:50 AM, AnHardt ***@***.***> wrote:
Are you sure this downloaded binary is produced with the appended config?
I have no way of knowing without asking Timothy. It came from the source folder of the binary download, but he may only provide that to comply with licensing.
I have never built Marlin from source.
|
|
I did this: #define HEATER_0_MAXTEMP 275 After that, I ran the autotune: M303 E0 S2750 C10 Autotune does not respect HEATER_0_MAXTEMP and will attempt to reach 2750 C. |
|
IMHO, this sounds like a real problem. So why are all the developers defending themselves instead of looking for a solution. It only makes a long thread that gets nothing done. |
|
My low temp is set for -15. I don't need an error when the ambient temp is that low. But, I'm not usually starting a print when it's that cold. |
|
I really hope your thermistor never fails open because you have effectively defeated an important safety feature.
… On Oct 28, 2017, at 12:45 AM, Tannoo ***@***.***> wrote:
My low temp is set for -15. I don't need an error when the ambient temp is that low.
But, I'm not usually starting a print when it's that cold.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
My safety thermal settings are 120 seconds and 10 degrees for bed , and 120 seconds and 4 degree for hotends |
|
You should never rely on software for safety. You have heated surfaces, and plastic that burns really well, a recipe for disaster. Commercial plastic products must use flame retardant additives by law*, filament used for 3DP does not contain retardants. (* depending on national laws I guess). The safety of home 3d printers is very poor, and although a known issue is often overlooked. Personally, I always regard printing as a "attendance required" activity. Now I know a lot of folks do long unattended prints and say "its fine", but those are all accidents waiting to happen. Unfortunately people take a binary approach to risk:
Fire services here are concerned about a rise in house fires due to domestic appliances, such as fridges and washing machines. Those are appliances that are supposed to have passed safety regs. Home builds or cheap machines imported from China won't even have been tested against safety standards. We can add all the features we like to Marlin, but it will never be "safe" to an official standard - it requires hardware support. I would rather people recognize that 3d printing is inherently risky, rather than be complacent and falsely rely on "safety features" in the software. If you are concerned about your house burning down, do not print unattended. If you are not concerned, you should be! |
|
If anything this would make a great clickbait YouTube video title… Guy starts PID autotune with a defective thermistor. |
* Fixed M303 thermal protection The temperature sanity checking logic was not being applied during M303 (pid autotuning) because instead of setting a target temperature, it directly manipulated the pwm values. When PIDTEMP/PIDTEMPBED is enabled, PWM values rather than the target temperature determine whether the heater is on. I changed this to look directly at the PWM amount when pid is enabled. * Turn off heaters on M303 error Currently, PID autotuning stops if it overshoots the temperature by 20C or if if the temperature does not change for 20 minutes and it times out. I added calls to disable the heaters in these scenarios. * Removed unnecessary if statement. Added changes suggested by GMagician. * Update temperature.cpp * Update temperature.cpp * Update temperature.cpp
Very much appreciate the addition! Safety is vital, especially around the heaters. I think up to now this hasn't been looked at because |
… into bugfix-1.1.x-HD-CR10 * 'bugfix-1.1.x' of https://github.com/MarlinFirmware/Marlin: (94 commits) [1.1.x] Fix M303 thermal protection MarlinFirmware#8103 (MarlinFirmware#8126) Implement support for Dual X and Y endstops Add Dual Steppers / Endstops to configs Cleanup for DIGIPOTS settings (1.1.x) serious bug G33 Add scripts for .travis.yml to append config options Revert default BABYSTEP_MULTIPLICATOR to 1 Added MKS GEN L board (MarlinFirmware#8088) Concise SD_REPRINT_LAST_SELECTED_FILE description Clean up trailing whitespace (1.1.x) auto tune calibration parameters (MarlinFirmware#8031) Tweak neopixel self-test Add some Polish translations Fix FWRETRACT logic, apply common sense give this language an unique name Initial conflict resolution of SD_REPRINT_LAST_SELECTED_FILE (MarlinFirmware#8104) 1.1.x - save 1400 bytes of FLASH by using reduced font for some languages (MarlinFirmware#8095) More specific M100 description Fix spacing, use single instances of similar pins Encourage users to re-examine their configs ... # Conflicts: # Marlin/Configuration.h # Marlin/Configuration_adv.h
…re#8126) * Fixed M303 thermal protection The temperature sanity checking logic was not being applied during M303 (pid autotuning) because instead of setting a target temperature, it directly manipulated the pwm values. When PIDTEMP/PIDTEMPBED is enabled, PWM values rather than the target temperature determine whether the heater is on. I changed this to look directly at the PWM amount when pid is enabled. * Turn off heaters on M303 error Currently, PID autotuning stops if it overshoots the temperature by 20C or if if the temperature does not change for 20 minutes and it times out. I added calls to disable the heaters in these scenarios. * Removed unnecessary if statement. Added changes suggested by GMagician. * Update temperature.cpp * Update temperature.cpp * Update temperature.cpp
|
There is only so much software can do. Has anyone considered using a thermal cutoff fuse in the hotend and/or heated bed? |
…re#8126) * Fixed M303 thermal protection The temperature sanity checking logic was not being applied during M303 (pid autotuning) because instead of setting a target temperature, it directly manipulated the pwm values. When PIDTEMP/PIDTEMPBED is enabled, PWM values rather than the target temperature determine whether the heater is on. I changed this to look directly at the PWM amount when pid is enabled. * Turn off heaters on M303 error Currently, PID autotuning stops if it overshoots the temperature by 20C or if if the temperature does not change for 20 minutes and it times out. I added calls to disable the heaters in these scenarios. * Removed unnecessary if statement. Added changes suggested by GMagician. * Update temperature.cpp * Update temperature.cpp * Update temperature.cpp
|
This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
Bug Report
Description: I started a PID tune from Octoprint's terminal. My thermistor went out during the PID tune (or maybe before I started it, not sure on this detail). Heater turned on and just never shut off until I intervened.
Expected behaviour: Firmware should detect mintemp before/during PID tuning and shut the heat off (it was reading -14C at the time).
Actual behaviour: Smoking PLA and scorched hotend parts.
Steps to reproduce:
Attach a ZIP of
Configuration.handConfiguration_adv.hby dropping here.config.zip
The text was updated successfully, but these errors were encountered: