Nozzle Clog Detection
I'm working on a college team to design a mod that detecting nozzle clogs, and is focused on hobby printers using the MK3s. The mod would automatically stop the print. What kind of things would be important to people who would potentially use this mod? What kind of resources would people have to make it themselves? What would you be willing to spend making it yourself? And any general comments on this kind of mod if you have experience with doing stuff yourself.
Updates to come as we progress.
The original MK3 had a laser motion sensor to detect filament stalls. The primary issue was the sensor is designed to see metal, not filament, and some filaments were invisible to the sensor so it would fail and stop prints for no reason.
There is a user mod to add a metallic bearing that rolls on the filament and presents a solid surface to the sensor. This mod to a MK3 accomplishes what you are trying to do.
The filament sensor in MK3S detects filament run out. However, it doesn't detect a clogged nozzle. When a nozzle gets clogged or filament stuck at the beginning of PTFE tube, the sensor will keep sensing the filament in and will not trigger.
One way I see to detect both filament run out and stuck filament is using a roller connected to a potentiometer. As the filament feeds, it will turn the roller+potentiometer. The FW will compare this movement (In/Out/Stop) to the Extruder movement. It should trigger if there is a difference. Basically, a feedback system on the filament movement.
It would be worth measuring the stall current of the extruder motor when there is a nozzle clog / jam. Since the motor will continue to push filament in, right before the gear skipping (clicking noise) one should see a slightly increase in the motor current draw before it drops again and repeats the process. I am not sure if the clogged nozzle would be enough to reach the stall current, but there would definitely be a current spike. The question is "how much". An oscilloscope might come handy here.
you would need a rotary encoder ( continuous rotation) not a pot
I know this is an old topic, but i was just kind of brainstorming and thought about how AI is being used for things like spaghetti detective to determine failed prints visually.What if a simple microphone or cheap and sensitive accelerometer could be used and train the AI to recognize the extruder skipping noise or vibration "signature" of the sensor? I know virtually nothing in terms of how feasible something like this might be, just thought i'd throw the idea out there for smarter people than me to ponder lol.. I also like the idea of monitoring and referencing the current draw of extruder motor which would surely throw out a clear anomaly for some clever software to recognize and either shut down the print entirely or send gcode to lift z, move out of the way and pause the print ant notify you to come take care of it .. then if its late in the print, theres a chance you could clear the jam and salvage the print. .
I have something along the same lines set up for my air filter using a smart plug with power monitoring to automatically shut it off if the current goes higher than normal basically letting me know the filter is clogged when there's a spike in the fan's power draw avoiding a burned out motor.
For quality control, a miniature sensor mounted near the orifice of a nozzle can optically detect a spray pattern emerging from the spray tip and send a “spray present” signal to the controller. Unconfirmed spray cycles – often the result of clogging – can activate alarms or stop production.