Jump to content
Ultimaker Community of 3D Printing Experts
Sign in to follow this  
burki

no-jam-extruder idea

Recommended Posts

Fiddling with clogging and such, i was wonderning...

If one took the existing extruder, and changed the extruder side of the bowden tube so that there would be a spring loaded fixture...

With such a setup, if the extruder feeds too much material, the spring loaded mechanism would be pushed out of the extruder setup.

If now we measured how far it is being pushed out, for example with a hall efect sensor, or even a cheap sharp infrared distance sensor or anything of the likes, it should be possible to tell the extruder to extrude less.

when the spring is not compressed at all, while the extruder pushes material, we might tell it to extrude more.

Thus we could ignore the filament diameter, and even handle varying diameters fairly easily.

Even the problem with the first layer being too low, forcing too much material against the build plate and thus building up a filament jam should be a problem of the past.

Obviously the firmware would need some tweaking for such a setup.

What do you guys think?

 

Share this post


Link to post
Share on other sites

The problem with using only nozzle pressure ( ~ bowden pressure) as a proxy for flow rate is that it varies greatly with printing speed, printing temperature, and material selection.

There is absolutely an improvement to be made by measuring nozzle pressure, but it isn't quite as simple of a setup as you describe.

 

Share this post


Link to post
Share on other sites

Good point...

Yet: Speed and temperatures are known factors, at any given time, for the software.

So whats missing is some sort of material constant.

Compared to the filament diameter, that should be easier to acquire. So i assume that if we had a means to measure the pressure in the bowden tube, that should indeed improve matters ?

 

Share this post


Link to post
Share on other sites

Yeah. I was looking at adding a pressure sensitive pad in line with the Bowden mount - sparkfun has several that look good - but don't really have the time to do that kind of tinkering these days. I like a pressure sensitive resistor better than a spring with measured travel, because the compression would make retraction more challenging, or at least less effective.

Share this post


Link to post
Share on other sites

I was thinking that the springloaded mechanism should compress the spring by something like maybe 2 mm when operating correctly.. That would not affect retraction unduly, those 2 mm can easily be added to the retraction length.

I still have the impression that such a mechanism would offer a fairly inexpensive and robust way of handling issues with irregular or badly measured filament diameters and clogged nozzles.

I don't think i really understand your thought about a pressure sensitive pad. Could you explain that a bit more?

 

Share this post


Link to post
Share on other sites

Here is a strain gauge that, once calibrated, can keep track of the pressure accurately.

 

 

kraftmess_gebastel.jpg

In case you don't recognize this - this is a UM Original feeder where the bowden would normally attach and the strain gauge is in the gap and the bowden is instead attached with black tape.

This is not my picture or design but very interesting. I found this on ultimaker google group but can't find it now.

Share this post


Link to post
Share on other sites

Here is a strain gauge that, once calibrated, can keep track of the pressure accurately.

 

 

kraftmess_gebastel.jpg

In case you don't recognize this - this is a UM Original feeder where the bowden would normally attach and the strain gauge is in the gap and the bowden is instead attached with black tape.

This is not my picture or design but very interesting. I found this on ultimaker google group but can't find it now.

 

I would be very interested in any further information you may find on this. I too thought of a strain gauge element to apply to the filament somehow, unfortunately one needs a really great bond to get good readings (source: applied over 1000 strain elements) and of course you can't do that to the consumed filament. It looks like the creator of this idea bypassed that by putting some sort of light clamping mechanism to the filament and then reading the corresponding strain of the flexing joint. I would like to ask the creator of this design a couple of questions, like if they saw stuttering in the strain signal, and if they could accurately measure retraction distance?

 

Share this post


Link to post
Share on other sites

Well the posting including the above photo came from someone called "Martin Renold" over on google groups:

https://groups.google.com/forum/#!msg/ultimaker/rx4t5ADCc8U/NKAqvUYoSt0J

He has a few posts near the bottom. I recommend you ask him there. The people over on ultimaker google groups are really smart.

 

George, how would that open design work with nylon or flex PLA?

 

You would definitely have to reduce that gap! For example you could extend the bowden partly across the gap. I don't know the answer. I'm not even sure how useful this is. I think this is only useful if you also measure the filament position.

 

Share this post


Link to post
Share on other sites

Hmmm.. admittedly, the spring loaded mechanism, or even that strain gauge setup has the downside of too much travel.

How about this: a fixture where the bowden tube goes in. this is loosely fitted in some sort of bearing, such that when the material is fed into the bowden, the fixture is pushed against the bearing. between the fixture and the bearing there's a pressure sensor. For example a flexiForce sensor.

This would eliminate the travel / retract issue and still be fairly inexpensive, while the flexiForce sensor would provide the presure information.

Is there a free analogue input on the arduino board? if so, total costs would be something like 20-30 Euros, since i assume that fixture and bearing should be easily printable.

Now all we'd need would be one Daid :-) and some experimenting / calibrating.

 

Share this post


Link to post
Share on other sites

Thank you for the link, Johan.

It seems this research does somehow at least support the iddea that measuring the force of the feeder should go in the right direction.

While i am aware that i am of course in a case of confirmation bias, what do you folks think? Is this an idea worth pursuing?

 

Share this post


Link to post
Share on other sites

Working on an feeder mod that uses his idea. Being the lazy person I am, I am trying to reuse as much of the existing ultimaker feeder frame design as I can.

For the middle layer, I am reusing the original plywood lasercut piece, which should at least reinforce the motor mount.

What I am wondering is if the part printed in PLA will hold up to the gripping mechanism. I haven't made hardly any changes to this area of the design, besides removing the pieces that grip the bowden tube retainer thingy.

 

Share this post


Link to post
Share on other sites

Hey guys,

just a wild idea here. Wouldn't it be possible to measure the voltage of the stepper and extrapolate the actual torque? If it is possible, that no additional hardware is required. I found this discussion on the Arduino forums. Maybe it is useful (http://forum.arduino.cc/index.php?PHPSESSID=hthtsj0ddum9ceufng9j53ddr3&topic=136261.msg1024483#msg1024483).

 

Share this post


Link to post
Share on other sites

Hey guys,

just a wild idea here. Wouldn't it be possible to measure the voltage of the stepper and extrapolate the actual torque? If it is possible, that no additional hardware is required. I found this discussion on the Arduino forums. Maybe it is useful (link).

You could, but it might be noisy. One issue I see is that the gearing heavily favors the stepper motor, so there might not be a substantial voltage change. Maybe on a direct drive, the voltage difference would be large enough to measure.

The other issue I see is that I am guessing the stepper controllers use PWM to control current, so you will have to be able to integrate and average the signal. Also, the inductive load of the motor can cause some funny business. I guess you could use a low pass filter, but it is still something you need to think about. You would need some sort of filtering to guard against voltage spikes that occur when power is shut off to the motor.

Finally, you still have to the build the electronics to give you the voltage feedback. When I last built a stepper controller from scratch, the chips I used did everything internally, so I would have had to add some electronics to measure the voltage. I would guess that you couldn't connect the output directly to the Arduino, but maybe using a voltage divider.

My point isn't that you couldn't do it, but that it wouldn't be simple and it might not work very well. From an engineering perspective, the load cell seems easier to design and probably more precise.

 

Share this post


Link to post
Share on other sites

I share aviphysics' view: My original idea was trying to have one isolated way of determining the current pressure.

Measuring Volts or Amperes the motor is consuming adds potential errors.

A Load cell, or a flexiforce sensor, are fairly inexpensive, so i think that approach is fairly direct. It would even continue working when you change the extruder, or the gearing, or whatnot.

 

Share this post


Link to post
Share on other sites

I like this thread.

I was having the same idea as i was designing an adapter for my E3D-Hotend (I had bought the direct-extrusion type, not the bowden type one).

It sits where the bowden tube goes in (as Burki said). One end grips into the upper part of the E3D-heatsink and the other holds a M6-nut where the tube is attached. One could easily redesign it to hold a strain gauge. Maybe cut it into two halves and glue over the gap using two gauges.

I also have uploaded the .scad-Files on thingiverse (yeah, still not really into youmagine, mea culpa).

http://www.thingiverse.com/thing:163856

 

Share this post


Link to post
Share on other sites

There are probably a couple of issues to overcome, either with the strain gauge at the drive end, or the hot end.

With the gauge at the drive end, you might have to compensate for variations in bowden friction, as the bend radius varies with the x/y position of the head. This would also be filament dependent; stiffer filaments would cause a greater effect. A solution would be to calculate the bowden bend radius based on the head position, and to have some sort of stiffness factor for each filament type, with which to calculate a modifier for the measured force.

With the gauge at the hot end, there are fewer factors, but the lateral acceleration on rapid x/y movement will probably introduce some noise. Some sort of clipping filter applied to the data and/or a deep set probe housing to minimise lateral slip across the force sensor might work.

With either design, any gap in the bowden path would have to be the minimum necessary, or else very soft filaments could bend or deform in the gap.

 

Share this post


Link to post
Share on other sites

This is a rough drawing of what I am thinking. Obviously, elegant design is not exactly my forte. The thin bits attached to the loadcell are metal brackets that came with it.

I think it would mount to the enclosure by the thicker arm on the same side as the gears. The reason to mount it off of this side of the load cell is so that it is always loaded the same way and isolate the force reading from the movement of the head.

Not sure what to do about disruptions from despooling. I am mostly interested in using it for qualitative diagnostics, so I don't think this part will be a problem for me. If someone wanted to use it for controlling feed, they would need to address this issue. To me, it seems like trying to control feed by measuring force won't work very well.

sGhKcW6.jpg

 

Share this post


Link to post
Share on other sites

Hello Aviphysics,

i assume the grey bit on the bottom of the drawing is the load cell?

If so, the design looks feasible for me.

Have you taken a look at flexiForce-Sensors? -> http://www.phidgets.com/products.php?category=33&product_id=3101_0

They might allow for an even smaller setup.

I don't understand what you eman by "disruptions from despooling". could you explain that?

 

Share this post


Link to post
Share on other sites

Hello Aviphysics,

i assume the grey bit on the bottom of the drawing is the load cell?

If so, the design looks feasible for me.

Have you taken a look at flexiForce-Sensors? -> http://www.phidgets.com/products.php?category=33&product_id=3101_0

They might allow for an even smaller setup.

I don't understand what you eman by "disruptions from despooling". could you explain that?

 

I already have that load cell, so that is why I am using it.

The issue with despooling disrupting the force measurement is that the side of the force cell connected to the boden tube is fixed to the body of the machine. This helps prevent the motion of the hot end from affecting the force measurement. However, the side of the force cell connected to the feeder drive mechanism can be tugged on by the spool as more filament is needed. This tugging would affect the force measurement, because the side of the force cell with the feeder mechanism is not fixed.

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×

Important Information

Terms of Use Privacy Policy