Another madcap idea for the mix. If we know what a full spool of filament weighs, what the empty spool weighs and how many meters are on it, we should be able to calculate roughly how many meters we have left at any given moment by checking its weight. Possibly refuse to start a print if the weight is too low.
You could also have an arm, gravity holding it against the filament on the spool. Detect the angle of the arm (or use a micro-switch at the right place). That measures the diameter of remaining filament, so when it gets close to the spools core, it sets off a buzzer, or emails you etc.
Looking at my spool right now the diameter at one side is far greater than the other, so a mechanical check could give you false alarms unless you had several arms. Optical detection may be easier to engineer.
Have a loop at the end of the arm with the filament going thru the loop, allow the arm to move sideways, then it will measure the diameter at the spot the arm is at, keep a running average, reset the average when the arm rises.
Or have a flap/bar the width of the spool, measures the max diameter.
Use load cell to continuously monitor weight of spool. It would be very accurate and simpler to develop. Should be robust and reliable compared to pure mechanical solution. Your initial thought of weight being the metric was spot on. The weight of an empty spool is subtracted from the gross weight to derive the tare.This would give a very accurate measure of the length of the filament available.
User just chooses spool/filament type from simple menu, or the system could recognize spool using RFID or barcode and display how much filament is available automatically.
Load cells are available widely and cheaply, you can find them around the house in kitchen/bathroom scales even in luggage handles.
If the length of filament spooling to nozzle is accurately measured and compared to changing weight data, an interesting new data stream emerges.This could be used to verify filament specification, or used for accurate logging of job data and manufacturing costs.
.
I think you’re right. A load cell is < £10, all I’d need is an Amp chip between it and one of the Arduino Mega 2560’s 16 analog inputs (via the RAMPS1.4). Then code added to the Marlin software to calibrate it at zero and say 1Kg. Ordering a spare Arduino and RAMPS to experiment with right now.
That is something similar to what 3d systems do with their cubes. But they run theirs based on a chip and the chip doesn’t know if you have pulled out two meters of filament and cut them off. This sounds like a much better system.
What spool weights do people buy? My current one is 1Kg so this load cell looks about right. http://proto-pic.co.uk/3133_0-micro-load-cell-0-5kg-czl635/