Machined Extruder- Thorp's build log

Greetings,

I have been working on the design for my first extruder. I've done my homework and shamelessly borrowed some ideas from others. I think that I am off to a good start but still, I wanted to post up my work and get some feedback.

I've provided an animation for the two revisions and I hope they will do better in conveying the design than the typical single cross sectioned view. The 2nd design was a progression from the first but both are aimed to keep the unit modular. The 2nd concept body is a bit more of a complex part to machine but I'm looking forward to it. All of the pins/bolts/gears are mcmaster parts and then their are three machined parts in each assembly. I will try to post up some work in progress pictures if you guys would enjoy them- probably will only be machining the second version and not both.

I sometimes wonder how much of the conveyed design is lost through communication over the internet. Maybe I will annotate a screen shot or two to specify materials and such?

I leave for a two week vacation in a couple days- won't be getting my hands dirty with this project until I return.

Thanks guys,
Thorp

Rev1:
Perspective view:
[a.imageshack.us]

Sectioned view:
[a.imageshack.us]

Animation:
[www.youtube.com]


Rev2:
Perspective view:
[a.imageshack.us]

Sectioned view:
[a.imageshack.us]

Gear view:
[a.imageshack.us]

Animation:
[www.youtube.com]

Edited 1 time(s). Last edit at 08/12/2010 01:56PM by Thorp.

Very nice designs. I too was thinking I wanted to mount my motor vertically and was wondering how I would get the filament past it. You have shown me two ways, thanks.

I also want to box in the gears so I don't have to keep re-applying grease.

I like the reduction to a single spring as well.

You will probably need a small fan near the heatsink.

Thanks for the inspiration.

Edited 2 time(s). Last edit at 08/12/2010 04:27PM by nophead.

---

[www.hydraraptor.blogspot.com]

Thank you for the kind words, the respect is mutual- really enjoyed reading through your blogs and the analytical approach you take.

I have tossed around a design or two with a cpu fan integrated into the heatsink- all one milled part, instead of machining a turned passive heatsink as it is in these models. If you don't think the passive heatsink in the model will cut it, then I'm going to go with active cooling from the get go.

I'm starting to look into DIY injection molding and stumbled upon these heaters.
[www.backyardengineer.com]
A bit big for our application but a scaled down version of that would be the perfect package for a heating element. Nophead's solution seems straight forward and easy but I hate the non-cyclindrical nature of it. Do these in the link looks like something he is putting together or is their a name, application, and various sizes of these do hickys?

They are called band heaters and you can get various sizes but I have never found any smaller than 1". They are used for heating injection moulding nozzles. Similarly cartridge heaters would be good to use but they are all too big as well, generally mains powered also.

---

[www.hydraraptor.blogspot.com]

Interesting.. if only!

Here is a mock up with a cpu fan.
[a.imageshack.us]

The connection between feeder and heatsink made the assembly big and bulky and so I decided to combine them.
[a.imageshack.us]

The assembly is still rough. And looking at this thing again, I think I could break the two parts up (feeder and heatsink) again and mate them with bolts in the open section above one side of the fan. The PTFE rod would align with a pocket in the feeder to insure alignment.

You can get away with a much smaller fan than that. I use a 25mm one.

---

[www.hydraraptor.blogspot.com]

Excuse my ignorance, but what is the advantage of vertically mounting the motor?

Is it in order to clear the overhead gantry and so give you a greater vertical print area?

Thanks

---

-------------------------------------------------------------------------------------------------
capolight.wordpress.com

It's not for Mendel, its for a moving table machine with multiple heads, so I want to get them close together.

---

[www.hydraraptor.blogspot.com]

Very nice design Thorp.

I also would like a vertical motor design.
How is the nozzle thermally insulated from the heatsink? the light green area?
I'm stuck with my nozzle design heating the main barrel and heatsink as I wanted to remove the need for PTFE/PEEK barrier, but my design is not looking very viable at the moment, you have inspired me to have another try.

I guess there is no reason why you could not still print most of this (motor mount and housing gears etc), you just need to make sure the heatsink stays cool.

Like nophead, I intend to make a multihead machine- not on this build, but down the road and I am designing with that in mind. That is why I did the tensioning unit as I did- so that I can stick another extruder close to the first one.

Richrap, I plan on using a peek/ptfe barrier that mimics hophead's design. The only real difference there is that I may use an o-ring between heated barrel and ptfe.

And in regards to the design-
Maybe I should direct the hot air coming off the sink downward and into the printed part?

Another worry of mine is the intertia of the head. Could that be problematic? I could see it being so with belts. If that is the case, what are my options?

Ballscrews would be one option and if I went that route I would be tempted to go all out, increasing travels too as my Mendel is aluminum.

A wider belt or a lighter head... Anything else?

Edited 1 time(s). Last edit at 08/17/2010 10:40PM by Thorp.

I have switched directions slightly but have been making steady strides.

I need to know the approximate force required on the filament against the hobbed wheel using a bearing.

Looking at the wiki page outlining wades extruder, I found this: "For a given filament drive force, you'll need about twice the spring force - ie, if you want 100 N of filament drive, your springs need to push with about 50 N each" (using 4 springs)

And referencing Nophead's blog, [hydraraptor.blogspot.com], it looks like the maximum force he pulls is 12.5kg which is 122.5 newtons. Double that and I need 245 newtons or 55 pounds.

I'm extrapolating from a couple sources and from dated work, so I just wanted to run these numbers by you guys and see if they sound reasonable. 55 pounds sounds like a lot to me.... I'm using one spring per filament and don't have a lot of wiggle room so I may be forced to use belleville washers if the above assumptions are true.

I don't know how Wade arrived at the twice figure. I don't think I use that much spring force, but I have never measured it. I only use two small springs though. I needed a lot more spring force with a threaded rod drive, less with a pinch wheel.

I think the force you need to get maximum traction is the force that pushes the teeth of the pinch wheel fully into the plastic to get maximum grip. Any more force just squashes the filament. So PLA will need a lot more spring force than ABS to get maximum grip because it is very hard. But on the other hand, because it is hard, it gets better grip with less tooth penetration.

Edited 1 time(s). Last edit at 09/10/2010 05:13AM by nophead.

---

[www.hydraraptor.blogspot.com]

I don't have a drive wheel to test or an accurate way to test force for that matter, but I am not too worried about it. I've found a spring that provides a suggested maximum of 85 newtons (19 lbs) that will fit nicely. If that doesn't suffice I will replace it with a stack of belleville washers.

I am close to finalizing the design and the solid model is all but complete except for a couple pins and bolts. I will finish up the BOM and order parts, tooling, and material in the next day or two.

Nophead and others, could you please check over my barrel and filament path. I am a bit worried that the heated barrel is too long and I will have a problem with oozing. You have mentioned that that can be a problem but also that commercial machines tend have a larger heated section. I'm curious as to what orifice size you use and if the problem is alleviated with a smaller orifice, like they tend to be on the commercial machines. Any other problems you foresee or things that I may have overlooked?

Thanks guys!

Perspective:
[img192.imageshack.us]

Sectioned:
[img36.imageshack.us]

Annotated:
[img163.imageshack.us]

All nozzles will ooze to some extent but reversing just the right amount pretty much cures it, so long nozzles will certainly work. You do get a bit of ooze during the time it takes to fast forward and reverse so the less molten plastic there is the better.

I think with a smaller nozzle it will ooze slower and for longer because the plastic will be under more pressure, so will compress more.

I use a 0.4mm nozzle on HydraRaptor with a very short hot zone and a 38:1 gear ratio. I use a 0.5mm nozzle on Mendel with a much longer hot zone and much lower gear ratio. HydraRaptor has more accurate control of the filament but Mendel is certainly adequate. I don't have any strings when building with ABS. I start the filament path on the inside of the object like Adrian does which helps to hide any ooze.

Your annotated diagram shows Acetal around the hot end, but that melts at 163C according to the wiki?

---

[www.hydraraptor.blogspot.com]

Thank you for your input. In what size do you microstep your extruder? 1/4? 1/8? 1/16? I read in your blog that torque is effected by microsteping by a factor of .7- is that .707 and from the RMS of the sinwave? Should motor torque vary with different levels of microstepping? What is the downside to 1/16 microstepping vs 1/8 besides max speed, or is there any?

After taking a more detailed look at extruded filament vs step size I have found that I have overlooked the knurled diameter. Mine is currently 25mm (machined into worm gear) vs your 13mm and Wade's 8mm. That parameter is pretty important and even though my gear ratio is 50:1 compared to your 39:1, my extrusion per step is close to double yours. On the other hand, a bigger diameter means better traction. . . I am having a hard time balancing these parameters.

The other thing that I have overlooked, is that a set of these gears that I have picked out cost ~$65. With two sets, I'm looking at $130 just in gears. I think that a better alternative might be a nylon worm gear, or possible 2 or 3 spur gear sets.

But then it also comes back to if I intend to make a one off extruder or if I am designing to possibly commercialize a product. If for a one off then I will spend more time redesigning than I would by saving a few bucks on gears. If the later, then price is a huge concern.

Nice consideration on the acetal I hadn't looked into that. What reprap page are you referencing?

Thorp Wrote:
-------------------------------------------------------
> Thank you for your input. In what size do you
> microstep your extruder? 1/4? 1/8? 1/16? I read in
> your blog that torque is effected by microsteping
> by a factor of .7- is that .707 and from the RMS
> of the sinwave? Should motor torque vary with
> different levels of microstepping? What is the
> downside to 1/16 microstepping vs 1/8 besides max
> speed, or is there any?
>
I think actually with a chopper drive you can simply increase the current by 1.414 to get back to the same torque and the same power as full step. I use 1/8th micro stepping because that it the highest the chip I am using does. I can't see any downside of using 1/16th. You don't necessarily get more precision but it certainly gets smoother.

You do have to chose the component values more carefully to be able to achieve the wider current range. See [hydraraptor.blogspot.com]

> After taking a more detailed look at extruded
> filament vs step size I have found that I have
> overlooked the knurled diameter. Mine is currently
> 25mm (machined into worm gear) vs your 13mm and
> Wade's 8mm. That parameter is pretty important and
> even though my gear ratio is 50:1 compared to your
> 39:1, my extrusion per step is close to double
> yours. On the other hand, a bigger diameter means
> better traction. . . I am having a hard time
> balancing these parameters.

I wouldn't say mine was optimised in any way, just what I tried first. It does seem to have more grip on softer plastics than Wades though. Actually you are comparing the ID of mine with the OD of Wade's which is only about 5mm.
Wades: OD 8, ID ~5
Mine OD 15, ID ~13


>
> The other thing that I have overlooked, is that a
> set of these gears that I have picked out cost
> ~$65. With two sets, I'm looking at $130 just in
> gears. I think that a better alternative might be
> a nylon worm gear, or possible 2 or 3 spur gear
> sets.
Yes engineering gears are very expensive which is why I use Meccano. I think Nylon might not be up to the force.

>
> But then it also comes back to if I intend to make
> a one off extruder or if I am designing to
> possibly commercialize a product. If for a one off
> then I will spend more time redesigning than I
> would by saving a few bucks on gears. If the
> later, then price is a huge concern.
>
> Nice consideration on the acetal I hadn't
> looked into that. What reprap page are you
> referencing?

Sorry I meant Wikapedia when I said wiki.

---

[www.hydraraptor.blogspot.com]

Hi Thorp

If eventually you made this a single extruder unit but with the ability to bolt 2 or 3 together I would be very interesting in buying some if you can get the price down to a resonable level.

A modular, stackable design is an interesting concept but at what cost? I don't see a solution to get the two or three nozzles close together in that way. Something to think about though.

I would like to think that this extruder is the head on a body and will eventually go on a machine that I have also designed. I am sometimes amazed at the success that makerbot and the like have received because in my opinion (and quite frankly), it is all ass backwards. It is a derivative of darwin, which was not designed with intentions to be mass produced- it was designed to be self replicating, and they started mass producing them. It makes more sense to me to take a step backwards and look at the machine from the ground up instead of being stuck with some of the weaknesses that darwin has, only because it was designed to be self replicating. But... power to them, them seem to be doing very well.

I have attached some pictures of a new direction that I may pursue. And also a picture of a compound gear. I am looking into prices on a gear such as this but have had a very hard time finding sources. Berg and SDP do not carry a single compound gear. How can that be? I even talked with an engineer at SDP. He wanted me to couple together a couple of his cluster gears. It didn't sound like a very clean solution.....

I personally liked the earlier design - motor vertically and using the worm gear, nice slim tall package.

What was wrong with the worm gear?

I made the blunder of using two gears that are incompatible:
A 1Y 5MYK08RA
A 1B 6MYH08R050

They have different lead angles. SDP only stocks one worm that fits the 5mm stepper shaft. The only compatible worm gears with that worm are nylon or acetal, which means I can't machine the filament gripper onto the worm gear itself. Probably still doable, but I'd have to rework the arrangement a little bit or find a different supplier. Toss in the extreme cost and I started to look into other avenues.

I am still alive, chasing that cake.

I have a job now. Or maybe a time sucker I should say. Also a 1964 Cal28 sailboat and a girl I'm genuinely interested in. None of which interest you at all and I bring it up only to use as an excuse for my insignificant progress on the extruder...

But I went ahead and ordered a set of the following *matching* gears
A 1P 6MYK08R050
A 1Y 5MYK08RA

And stopped by industrial metal supply today for the 2.5"x2.5" aluminum block. I will cut the aluminum block up on the horizontal ban saw tomorrow at work.

The ever changing latest:
[img836.imageshack.us]

A lot can happen in a month...

It's looking interesting, do keep on going with it.

I'm having a go at my own, but there are limits with a drill-press and a hacksaw.

---

[richrap.blogspot.com]

Thorp Wrote:
-------------------------------------------------------
> I made the blunder of using two gears that are
> incompatible:
> A 1Y 5MYK08RA
> A 1B 6MYH08R050
>
> They have different lead angles. SDP only stocks
> one worm that fits the 5mm stepper shaft. The only
> compatible worm gears with that worm are nylon or
> acetal, which means I can't machine the filament
> gripper onto the worm gear itself. Probably still
> doable, but I'd have to rework the arrangement a
> little bit or find a different supplier. Toss in
> the extreme cost and I started to look into other
> avenues.

I made a custom 1:45 worm/wheel arrangement made for our current head mounted extruder. It cost ~ £60GBP

The worm is st/steel and the wheel is brass if you are interested?

---

Cheers,

Paul

Paul, I have already purchased a set of gears but would be interested to see the work you are doing or any more information you have.

Thorp Wrote:
-------------------------------------------------------
> Paul, I have already purchased a set of gears but
> would be interested to see the work you are doing
> or any more information you have.

Thorp, if you care to PM me with your email I will certainly send photos of what we have produced.

It seems you have figured out the design anyway for your commerical product so perhaps you dont need it anyway

---

Cheers,

Paul

[img813.imageshack.us]

Any New News Thorp ?

I have been test-building a few dual extruder hot and cold ends over the last few weeks, but I'm still really interested in your nice machined design.

---

[richrap.blogspot.com]

Plugging away. Here and there...



Will check your blog out with interest.

It's looking interesting.

I have two extruders working now, but I need a better way of mounting them together, at the moment it's just 2 x WADE design's back to back (one reversed) - I'm also getting side-tracked with rebuilding my machine and new electronics it's all good fun.

I'm contemplating switching my new machine over to a dual Bowden extruder system so I can mount the heavy motors and Filament drives off the X/Z platform and get both the hot-ends nice and close together.

Rich.

---

[richrap.blogspot.com]