Dry-Draw concept

[Generic Default]

Hey everyone,

I've had this idea for a while now that I think might solve some of our problems with filament diameter and water absorption. I call it the Dry-Draw, but it's pretty simple.

So basically it would be a small metal tube with a hole down the middle that the filament gets pulled through. The hole would be the exact diameter of what you want the filament to be when it's printed. This part could be about 1/2 and inch long and would be mounted somewhere between the spool and the cold end extruder. The metal tube would be heated to a point below the plastic's melting temperature, but hot enough to dry any moisture out of the filament immediately before being fed into the hot end (especially nylon and hygroscopic plastics). It would be passive; the stepper pulls it and a small nichrome wire with a low voltage could heat up the tube to a reasonable temperature until the dissipated power is equal to thermal losses. Like a tiny toaster on your printer using only a few watts at most (it could be powered by the 5v output on your printer's power supply). The raised temperature would reduce the force needed to draw it through the metal tube.

The whole point of this is to draw the filament to a consistent diameter to avoid over or under extrusion regardless of variations in the filament diameter, as well as to dry out the filament prior to printing. I think it's a simple solution that would eliminate the need to bake spooled plastic prior to printing. It would also allow the use of homemade filament extruders without the need for an extremely consistent filament diameter.

Does this sound viable, or am I stupid for thinking of something like this in the first place? Any experts on wire drawing or thermodynamics please tell me if this is legitimate!

[Captain Starfish]

It's a neat idea. The devil will be in the details.

Just thinking "out loud":

If it's hot enough to ensure drying at high feed rates and to set the correct filament size, it will be hot enough that the plastic will be soft. Soft enough that if you pull on it the filament will be drawn thinner. And pushing on it will blob.

Sadly, on most prints there's plenty of push, pull and all manner of acceleration in between meaning you're going to end up with a mess of plastic if you put something like this in the live filament path. What you could maybe do, though, is have a steady rate extruder pushing filament through this device at the "average" rate into a big loop into the air which would buffer the extra filament, allowing the main extruder/hot end combo to back and forth as much as required.

Another point - just a bit of nichrome wire won't give you the fine temperature control you need to ride the very fine line between too cold to do anything useful vs melted puddles of plastic. You'll probably need a full PID loop.

All in all it might end up looking like a continual rate 1.75mm nozzle extruder sitting between the spool and the "business end" extruder.

Or

Maybe it doesn't need to be that hot. Maybe 120º or so to boil off the moisture without affecting filament size is all it would take. In which case why not make one up and do some tests? I'm kinda tempted, have the nichrome wire sitting in the bunky bin anyway...

[Eaglezsoar]

I think that this has potential. You need to get one made and experiment with it.
The concerns raised by the Captain are valid but all inventions have problems that
need to be overcome and I still think it is a cool concept and needs to tested and developed.

[Captain Starfish]

Yep, the concept is neat, just need to try it and tweak until it works

[mhackney]

Yes, it is an interesting idea. Dies for drawing wire to size work by pulling the wire lengthwise, thus decreasing their diameters. Our filaments are ductile enough to do the same thing I think. These wire dies are actually conical - and I think your's would need to be too. Think about it, the fatter diameter filament would need to feed through the input and then be reduced in diameter to come out the output. Your idea of using heat would increase the ductility so the stepper could draw the filament, a good idea.

[img]http://upload.wikimedia.org/wikipedia/e ... rawing.svg[/img]

These wire draw plates have a very short length and not a lot of reduction since it takes a lot of force to pull wire. Our plastic should be much easier, especially when warmed, so a longer draw length would be fine and provide the necessary residence time to warm the plastic.

It's time to build one and try it out!

[ccavanaugh]

AND, if close proximity next to the extruder with extrusions and retracts becomes too much force for one extruder to deal with, you could always preprocess in front with a secondary extruder and leave a small amount of filament length as a buffer. The secondary extruders only job would be to push it through the die which should be less force than extrusion at the hot end.

The secondary extruder would simply be wired in parallel with the primary extruder on the same driver and stay synchronized the extrusions and retracts of the primary exturder. A small loop of a meter or two would not be impacted enough by humidity. This would also allow for some cooling to occur as to prevent crushed filament, etc because to is too soft going into the primary. It would also accommodate small mismatches over a large print job if the primary extruder was to feed faster than the secondary.

I've been toying around with installing a second print head, but your dry-draw concept has me rather excited and I believe could be a game changer for print consistency.

[neurascenic]

AND, if close proximity next to the extruder with extrusions and retracts becomes too much force for one extruder to deal with, you could always preprocess in front with a secondary extruder and leave a small amount of filament length as a buffer. The secondary extruders only job would be to push it through the die which should be less force than extrusion at the hot end.

The secondary extruder would simply be wired in parallel with the primary extruder on the same driver and stay synchronized the extrusions and retracts of the primary exturder. A small loop of a meter or two would not be impacted enough by humidity. This would also allow for some cooling to occur as to prevent crushed filament, etc because to is too soft going into the primary. It would also accommodate small mismatches over a large print job if the primary extruder was to feed faster than the secondary.

I've been toying around with installing a second print head, but your dry-draw concept has me rather excited and I believe could be a game changer for print consistency.

Would it? with a wire draw, the extrusion is longer than the original. Even if slight, it will add up. wouldn't this be a problem? I think you would need some kind of condition sensor to see if the feed out of the Dry-Draw has exceeded a certain amount, and kill the feed until it is under some other prescribed amount.

[ccavanaugh]

I was assuming sizing a 1.75 filament for 1.75. You are correct that too much variation would be an issue.

The min limit inputs on the Rambo could be reallocated for a min / max sensor for the filament loop and the 2nd extruder connection on the Rambo could be the driver. It would obviously take some firmware changes, but I think the end result would be very beneficial. I've seen a take up reel system of some sort for the filament extruder that may be adaptable as well.

For an independent solution, a software teak to this http://solidoodletips.wordpress.com/201 ... der-intro/ may work as well.

[geneb]

If you were to add a little gauge (a spring, a lever with a tiny roller bearing on it and an A1302 hall effect sensor) to the filament just before it enters the extruder itself, you could adjust the flow in real time based on diameter changes in the filament. It wouldn't be totally effective until the first measured point entered the hot end though. (the firmware would think, "in Xmm, the filament diameter is now Y, so adjust flow rate accordingly")

g.

[bubbasnow]

who says you have to do this right before printing.. maybe have a reel to reel set up, and when you get a new shipment of filament in just throw it on and use it later.

[ccavanaugh]

If you were to add a little gauge (a spring, a lever with a tiny roller bearing on it and an A1302 hall effect sensor) to the filament just before it enters the extruder itself, you could adjust the flow in real time based on diameter changes in the filament. It wouldn't be totally effective until the first measured point entered the hot end though. (the firmware would think, "in Xmm, the filament diameter is now Y, so adjust flow rate accordingly")

g.

That would be better than trying to mess with the limit switches... It might also tie in nicely with the firmware work in Marlin to add volumetric extrusion control as well.

[Generic Default]

I like your feedback guys.
geneb made the point of adjusting the flow rate based on real time changes in filament diameter; I've heard of this as a concept before. It could be done fairly simply but there would need to be major changes to firmware to support it, as well as extra inputs on the electronics to detect the sensor (potentiometer probably). It wouldn't actually modify the filament in any way, so moisture problems wouldn't get fixed.

Just to make it clear to everyone since my concept picture was vague, the dry-draw would be mounted between the free-spinning spool and the cold end pinch wheel. This would mean that there doesn't have to be any special "buffer length", since the plastic would only be warmed while inside the draw tube and immediately after it exits it.

I also realized that the draw tube itself should be extremely short with fillets and chamfers to prevent jams and stuff. It would probably look more like a plate.

The hole diameter in the plate would probably be ~1.65mm for 1.75 filament, which would slightly reduce the size without putting too much strain on the filament. The point of heating it while drawing is to evaporate some moisture and to make it easier to pull. After being pulled through the plate, the filament would be inside the PTFE tube so it wouldn't re-absorb humidity.

More pictures and stuff coming tonight and in the next few days.

[mhackney]

That looks like it, much more like a wire draw plate. I think that should work nicely.

[geneb]

You don't want to use a pot - you'll eventually get noise and other issues. The idea is to use a hall effect sensor as a rotating position sensor. Here's one I've built in the past (made Hackaday on this one! *laughs*): http://www.simpits.org/geneb/?p=299

If there's an available analog input on the RAMBo, that could be used. I wouldn't bother with Marlin on a bet.

g.

[JohnStack]

Fascinating idea.

A couple of questions:

1. Would you have to put a thermistor and associated logic on it to control its temp? Calculate the right temp based on ambient temp and humidy?
2. Would the filament stretch a bit? Requiring the device to be placed right after the extruder gear or perhaps right before?

[Generic Default]

The filament would definitely stretch a little bit if it was larger in diameter than the hole. If the filament were smaller than the hole, it would just be pulled through and heated up a little bit. If it were larger than the hole, it would be drawn to size and heated up. The hole would just reduce the diameter variations of the filament to get the exact flow volume at the nozzle.


There wouldn't have to be a thermistor on the metal plate (probably); a few watts of power from the power supply 3v output would be enough to heat the metal plate until the input wattage would be canceled out by thermal loss to the surrounding air. This would have to be fine tuned to the plate material and geometry. Of course this draw plate wouldn't need a very precise temperature control anyway. It just has to keep the temperature in the 100 degree Celsius range to evaporate some of the water in the filament and make the filament easier to pull through the hole.


I'll keep working on the design of this thing but I won't be able to prototype it for a few more months. The goal is to make it as simple as possible, safe, cheap, and without using any valuable inputs or outputs on the RAMBO.

[JohnStack]

I would almost fab something using a hot plate (setting it on extremely low and leaving it there - and attaching a reel spool on it. My Max is maxed for power. I'm loath to put anything else on it - plus I'm kind of reserving for hall effect sensors some day...

[dtgriscom]

I'm an engineer with very little experience in 3D printing. That said, I don't think that this will work. It takes time to air the moisture out of nylon; it isn't just a "get it to the correct temp and the moisture pops right out" situation. You need to leave it in (relatively) dry air at the right temperature for hours, so that the moisture that has slowly diffused into the filament will slowly diffuse out again. Heating it up enough to make the water come out quickly will make it melt, sizzle and spit, just like it does in the hot end.

The second problem is the confined tube. Water will diffuse out of the nylon is if there's less of it outside (e.g. the dry air in your oven). But in a tube, where does the water go? It would only have a place to escape once the filament had left the heated tube, when the filament would already be cooling down.

Perhaps you could heat it up enough to actually boil out the water, re-extrude it, and then cool it enough for the Bowden drive to not tear it apart. Sounds complex, though.

[Generic Default]

Good observations with your last post! I realized shortly after posting the tube diagram that a small tube wouldn't be the best way to do this, since, even if the plastic is dried out and pulled through, the thermal expansion might cause a jam inside the tube. I think the heated draw plate would be better for getting the filament down to size (which is half of the point of this device). The filament would be heated before going into the draw plate, possibly through a small tube that's larger than the diameter of the filament. This would dry the filament a little bit and make it more pliable to be pulled through the hole. The filament wouldn't be as hot after leaving the hole, which would prevent it from being stretched too thin (hopefully). The filament would still be hotter than the surrounding air, though, and it would continue to dry.


I think the huge industrial plastic drying systems used for injection molding and just plastic processing in general have a completely different situation than what we have to deal with. The large industrial dryers have massive hoppers full of metric tons of plastic, which is why they need so many hours to get the moisture out of it. Our filament has a much higher surface area to volume ratio (assuming surface area in this case means plastic surface surrounded by dry air and not more plastic pellets). A 1.75mm or 3mm rod of plastic can be dried much faster than a pile of pellets.

What gave me hope that this would would was that I noticed that leaving the fan off on my E3D hotend made the aluminum heat sink get hot, and I could slowly print saturated nylon without having as many bubbles as when the heat sink fan was turned on. I figured that the heat sink was preheating the plastic and removing some of the moisture. Not all, but some. It made a noticeable difference in the number of bubbles in my prints.

Maybe you're right about this, though. It's still just an idea until I get the right tools to make it.