Monday, August 6, 2012

Road to better paste extrusion, episode 2: Auger Valves

After the previous lengthy post (and the time it represents) the short conclusion was that a paste extruder merely working by pressurizing a syringe full of material in order for the material to extrude is not a reliable method for use in RepRap machines. It's simple, it's straight forward, served us very well but impossible to predict and control the flow rate.

The next episode of chronological brain dumping, an auger valve extruder prototyped and tested in the first halve of 2011.

Convinced that one doesn't need to reinvent the wheel, I started to search for industrial areas and applications where they have similar needs: precise extrusion of paste materials. The field I found the most closely related is the fluid dispensing technology used for the precise deposition of dots or lines of sealant, adhesive, silicone, epoxy, glue, resin, soldering paste etc.

The basic setup of most fluid dispensing implementations looks like this:

There is a material reservoir which holds the material under the necessary conditions (stirring, heating etc). Next there is the feeding system, to transport material from reservoir to dispensing, for low viscosity material this can be gravity feed but higher viscosities need a pressurized syringe or large pressurized container. High viscosity material feeding can be done by using a feed pump (not in drawing). Right before the nozzle is a device called the dispensing valve.

Lets skip the first two for now and jump straight to the dispensing valves, the operating end of this system. I was immediately drawn to them because they sure look like badass extruders!

source: http://www.biztrumpet.com/en-us/products-70247/TH_Dispensing_Valves.aspx

There is a whole range of valves used in these industrial applications. After researching many suppliers websites and catalogs (with often confusing and contradicting use of terminology), it becomes clear that they fall into only a few categories and most of these devices are, as their name implies, simply valves.

As a non native speaker I had to remind myself what a exactly a valve is, its like a faucet in your kitchen. A device that can switch a flow of material on and off. Nothing more but also nothing less.


Valve
a device for controlling the passage of fluid through a pipe or duct.
New Oxford American Dictionary 



In this basic valve category we can find fine apparatuses like the Poppet, Needle, Diaphragm, Spool and Pinch Tube valve. They differ in working principle and therefore are better suited for different applications. The Poppet and Spool valve seem the most suitable for high viscosity paste materials, even abrasive and filled pastes (what clay is). They are all operated by air pressure and follow digital logic, open or closed. One can regulate these valves only with set screws which increase or decrease the diameter but this is a fixed setting, not something you control during operation. So it boils down again to two other 'controlling' parameters, time of opening and feed pressure and we know those two from the last episode. This types of valves would make a nice addition to the previously discussed pressurized syringe systems by switching the on/off action to the nozzle side of your system instead of pushing the whole body of material. Think (in faucet analogy) of the pressurized syringe time/pressure system as trying to control your kitchen faucet water flow all the way up at the water tower... No, it makes sense to add the faucet or valve right where you need it, just like a water tower, a syringe is constantly pressurized and will only feed material when there is a demand. 

But I was not interested in these 'valves' because while a nice potential improvement for the starting and stopping of your extrusion, there is still no way to predict and control the flow rate. We still have to visually verify how much material is extruded through the nozzle at any given moment or use complicated flow meters and regulating gear.

(To be complete, in many industrial applications previous episode's simple time/pressue syringe is used but I would rather not even call that a valve, and we dismissed this principle for an extruder in that episode)

So first conclusion is that what we're after is not a valve so much but a pump.

There are various other speciality valves that I will look at in later posts like Peristaltic or Progressive Cavity valves. But there was one type at this point in time (feb. 2011) when initially researching industrial dispensing valves that was very different from the simple valves and immediately caught my eye: the Auger Valve. Some promo snippets:
"The XXX series of precision auger valves is suitable for dispensing medium to high viscosity pastes, epoxies, solder pastes and other filled materials."
"The valve uses a feed screw (auger) to dispense fluid with a rotary displacement action, allowing ultra-precise control of the dispensing process."
"The XXX Auger Valve is a precision dispense valve specifically designed for metering controlled deposits of solder pastes, thick sealants and other particle-filled materials."
"Fluids can also be accurately dispensed in continuous beads in widths that range from 0.010” – 0.050” (0.25mm - 1.27mm) at rates up to 4” (102mm) per second."
"The unique design ensures that material is constantly present at the feed screw inlet, while the controlled rotation of the feed screw moves the material from the feed to the discharge point. Discrete control of the forward and reverse rotations of the feed screw controls the amount of material discharged."
"With a feed screw made of Delrin®, the valve is designed for use with two-compound and abrasive materials."
These quotes gathered from the various vendors of auger valves sounded like a sales pitch aimed straight at me:

-high viscosity paste: check
-abrasive filled materials: check
-ultra-precise discrete control and metering: check
-continuous lines: check

Auger valves works like this: The material is fed from a continuously pressurized syringe or external reservoir into the top of the valve. The valve consists of an auger screw fitting perfectly in a cylindrical housing which at the bottom ends in the nozzle. The pressure on the syringe is just enough to feed the material in the valve where it will hit the auger and stops there because of the increased friction caused by the narrower size of the fluid path along the screw thread and ultimately at the even narrower nozzle end. The screw is actuated by a motor and this rotation forces the material down the screw thread out of the nozzle. The flow rate is controlled by the RPM of the motor.


source: dispensingtips.com

These valves are used a lot for precise dispensing of solder paste in SMD electronics and are offered by various brands like Nordson EFD (794 Auger Valve), Nordson Asymtek (DV-7000, DV-8000), Techcon Systems (TS5000, TS5000DMPTS7000), Fisnar (PDV-1000), IntelliSpense (Auger Valve), GPD Global (HyFlo, MicroDot).

These things are rather expensive (upwards from 2500$) but when you look at them, they consist of a DC motor with optional gears/encoder and on the other end the screw housing with fluid inlet, nozzle end and the screw. If we substitute the motor end with 'standard' RepRap Nema steppers then what we need is only the screw assembly, the extruder's 'business end'. Thats why my attention was immediately drawn to Techcon's TS5000DMP because it has a 'Disposable Material Path' hence the DMP in the name. The DMP is the whole screw assembly in small and affordable delrin plastic package. The cheapest that I have found is around 25€ at dosieren.de (and up to 125€ for the exact same bit, industrial premium prices...).

source: adhesivedispensing.co.uk
source: adhesivedispensing.co.uk

The inlet side has a male luerlock, the nozzle side a female luerlock, the auger is sealed with an o-ring and has a square slot for the motor shaft, you can have the augers with different pitches.

Ordered a couple of the medium 8pitch DMP's and designed a very simple extruder head that mounts on the bfb Rapman for doing the experiments. The extruder is basically two halves that hold the DMP and both bolt together on a stepper with the extruder carriage plate in-between. An new carriage plate was needed because both the syringe and the extruder had to pass through. Its not designed for easy replacement of the DMP, was just to get the job done. The files can be found on Thingiverse: thing:28018

extruder

extruder half open in the middle and custom carriage plate

stepper and DMP parts

Extruder mounted on the Rapman

The first tests with clay were fantastic, this was exactly what we needed. It behaved very much like your standard plastic filament extruder. Turning the stepper on resulted (after priming) in an immediate flow of clay proportional to the speed. So you could easily speed up and down the flow of the clay. Stopping the stepper resulted in an immediate flow stop and no material passed the auger form the continuously pressurized syringe, when that pressure was set just high enough to feed the clay in the auger. We did some initial test with clay, chocolate and potato mash and those looked all promising. The following movie was shot using my original iPhone so looks a bit crap but you get an idea how it prints. This test object has three separate single wall shapes to test the on/off capabilities.



Tests to reduce ooze which is much more annoying in clay because it drags the whole print with it. Prime and reverse were used and jitter to make the point each layer starts at random


But after some further extensive testing with porcelain clay the system became less and less reliable and odd symptoms started to occur: flow rate became unreliable during a print and adjustment was needed, material leaked past the auger when the motor was not turning. Closer inspection of the auger confirmed a suspicion, that the delrin plastic augers were being grinded down by the ceramic material. The auger showed already degradation after a couple of minutes, therefore the auger was less capable at transporting the clay and increasingly the air pressure started feeding the material instead of the degraded screw thread. Vieweg (dosieren.de) conveniently also sells stainless steel augers for use in the DMP, they come at a more hefty 125€ a piece. After installing one of these augers another problem arose, the steel auger and the abrasive clay were eating away the housing. The problem being that it is rather difficult to get the auger to perfectly align with the housing so when it sits at the slightest angle it will grind the housing. I have printed a couple of extruder heads and made adjustments to about every part between stepper and DMP so that the auger would nicely sit in the housing but that proved very hard with RepRapped plastic parts. Maybe one of the reasons that original auger valve was 2500€, precision engineered... I also found complete stainless and even ceramic augers + housings but they were not of the 'disposable' category and cost 600€ and up.

left stainless steel, right delrin plastic (different pitch)

hole in the inner tube, you see the auger through


So the promising DMP part is not really up the task, I also had the feeling that it was underdimensioned for the purpose of extruding continuous and with a rather large flow rate. The stepper was turning at 100-200RPM to get to the speeds needed. But its not because this DMP bit is not up to the task that the whole auger principle is down the drain for our purpose is it?

Imagining that the rapid wear and tear problem would be solvable than there was another observation that was against what we need in a reliable predictable paste extruder. I did a couple of tests to see what happens when you changed the air pressure on the syringe that feeds the material into the auger and, obviously, below a certain point the pressure would drop too low to feed the auger but above that point the output of a fixed speed turning auger would vary in relation to the input pressure. Definitely not as much as when you would use a direct air 'frostruder' system but still more than enough. So that means that after all a change in material viscosity would still result in a changing flow rate at the nozzle and there would be no way to say that at a certain RPM you would get the right width of your extruded line, you would still need to fiddle with the air pressure. DAMN.

I know some people tried building auger valves using drill bits like these guys from Bauhaus University Weimar: GMU:RAPMAN/CLAYSTRUDER Before writing this post I checked if they got any further success with this system but unfortunately not.

While revisiting all my notes bookmarks on the topic I stumbled upon the following pdf: Archimedes Pumps. This is by far the best technical (but very readable) description of how an auger valve works, its operating principles and strong and weak points. In two different paragraphs the relation between viscosity and flow rate (in a previous post I described the relation between viscosity and air pressure, a small change in material viscosity means a change in flow rate under a constant pressure).


"Any changes in material flow characteristics affect the volume of material dispensed"    
"Positive displacement implies that a specific volume of material is displaced within a specific mechanical actuation. Positive displacement is not influenced by changes in temperature or viscosity. The Archimedes screw pump is very consistent, but this consistency depends on the viscosity and flow characteristics of the material. Archimedes screws have been marketed as positive displacement pumps, but “viscosity metering pumps” is a better definition."


So the auger pump is not a true positive displacement pump (sometimes referred to as a metering pump) and the conclusion is that that is what we're after!


Positive Displacement Pumps, unlike a Centrifugal or Roto-dynamic Pumps, will produce the same flow at a given speed (RPM) no matter the discharge pressure. Positive Displacement Pumps are "constant flow machines"

Next episode: Positive displacement pumps aka "constant flow machines" (I like that, sounds good)

ps. If there is anyone (or you know someone) with day job experience in the dispensing industry please chime in with your thoughts on these observations.


Some extra links and references:

General descriptions of various valves with pro and contra can be found on several supplier and manufacturer websites:
http://www.sealantequipment.com/metering.htm
http://www.fisnar.com/media/technical/0407MeD-Valves_pd.pdf
http://www.fisnar.com/media/technical/Valve_Chart.pdf
http://www.adhesivedispensing.co.uk/valveguide.htm

more links to various auger valves:
http://www.nordson.com/en-us/divisions/efd/Literature/Brochures-Data-Sheets/Valve-Systems/Valves/Nordson-EFD-794-Data-Sheet.pdf
http://www.nordson.com/en-us/divisions/efd/Literature/White-Papers/Solder/Nordson-EFD-Auger-Valve-Dispensing.pdf
http://www.dispensetips.com/pages/auger.html

23 comments:

  1. Thanks for sharing all that research :)

    ReplyDelete
  2. welcome, let me know if you use it.

    ReplyDelete
  3. You mentioned that you made mods to "Modular Moineau Extruder" that made it water and air tight.

    I was wondering where I could find them?

    Thanks,

    Will...

    ReplyDelete
  4. You mentioned that you made mods to "Modular Moineau Extruder" that made it water and air tight.

    I was wondering where I could find them?

    Thanks,

    Will...

    ReplyDelete
  5. Hi Will,

    Unfortunately did not yet have the time to work further on this. Some parts where modified 'analog' (read hacksaw) and I have not yet modeled those modifications for a new version. If you want the old files, shoot me an email, you find contact details at unfold.be

    ReplyDelete
  6. What are the pressure ranges used in this device?

    ReplyDelete
  7. low, 1-3 bar otherwise you push material past the auger

    ReplyDelete
  8. I am unsure if the clay extruder is still on your mind. I did a bit of research, and it seems as if internal gear pumps are typically used in this application. They involve a small driven gear, a crescent, and an outer gear ring with cogs facing center. A gerotor pump might also be appropriate, depending on viscosity, specific gravity, and pressure. Are you essentially running porcelain slip at 120 viscosity, 1.85 specific gravity, and 3 bar pressure? What is your maximum flow rate in liters per hour? I assume the use of a high pressure hydraulic hose would allow the extruder gear to be decoupled from the deposition nozzle, reducing mass.

    ReplyDelete
    Replies
    1. Hi Russel,

      Gear pumps have been tested and they clog easily with clay. There is a lot of contact surface area between the moving gear and the static housing and the clay gets into that space and grinds away there. The mechanism stalls or you need to apply lots of torque but you'll damage the pump, clay is very abrasive. The porcelain is less of a slip, more a paste and I use around 3-5 bar indeed. Not sure what unit your '120 viscosity' is or how to measure viscosity easily on a thick paste, also don't know how to measure specific gravity. Max flow rate is about 15ml/min.
      The moineau which I have been working on lately is kind of a 3d version of the gerotor with very limited contact areas, the problem is that these need to be sealed very well and that is hard to archive with DIY printed parts.

      Delete
  9. This comment has been removed by the author.

    ReplyDelete
  10. I tested your Moineau pump with a thick saw dust water solution and due to the high viscosity of the material it did not work (my guess). Asking for advice I was told the stator should be flexible (rubbery) to make the contact with the rotor leak-proof.

    I am now testing whether this helps or not. My paste is not abrasive but so think it is giving me hell to be pumped.

    Are you still using your pump successfully?

    ReplyDelete
  11. Hello Miguel, could you specify exactly which model pump you are referring to that you have build?

    ReplyDelete
  12. It's based on your design, but I was looking for the pump to be attached to the bottom of my paste container. http://www.thingiverse.com/thing:204649

    ReplyDelete
  13. This is your design that I had trouble with. But I am not sure why. http://www.thingiverse.com/thing:16741

    ReplyDelete
  14. ok, that one was a quick n dirty update of the moineau design. I have done some further work on it few months ago but not yet time to update everything and also not everything solved. I did write an article for the upcoming RepRap magazine that goes in a bit of more detail on the moineau. I am on the verge of moving to the USA from EU for 6 months to work on new stuff with ceramic 3d printing. I am planning to publish a better moineau in the coming months (but not much now).
    Coul you elaborate on the issues? I guess its mechanical, most of the issues I stil have are of that nature. Possibly your stepper is not powerful enough and it is skipping steps, or the coupling is lost its grip on the stepper shaft, or the claws break of. Those are the problems I still have with thick clay with high visco. If its te stepper, use a larger size or a geared one. And yes, definitely print everything in ABS with vapor treatment.

    ReplyDelete
  15. Apparently the problem I am having is due to the fact that not having a good seal between the rotor and stator it allows my solution to be dried up (water is squeezed out) while solids remain filling and blocking the cavity till the stepper cannot move any longer.

    ReplyDelete
  16. ok, probably because the material is behaving a lot like a suspension and the water flows back up the stator. That doest happen with clay so much because of the better bonding of water and clay particles I guess. One thing you can try is adjusting the original moineau parameters. In the OpenSCAD files you can change the tolerance between Stator and Rotor, you can play with that to get a better fit between the two. I noticed the default one is very loose. Second you can extend the rotor-stator to get extra windings and therefore extra cavities and seals.

    I changed:

    R1=2; // radius of rolling circle
    H=60; // height
    c3=0.1; // rotor clearance
    phi=960; // degrees of rotation of stator (>360)

    To keep the same geometry but add windings I believe you just have to add H and phi proportionally. So 16 degrees per mm.

    This also lowers the cavity volumes to get less pumped volume per rotation. (mine was extruding to fast al low RPM)

    Indeed a professional moineau will use a steel rotor and rubber (something durable like POM) for the stator. I assume you will need a very durable rubber so it will not be worn down by the not so smooth rotor.

    Keep me posted, as said more on my own experiences soon....

    ReplyDelete
    Replies
    1. My attempt with compressed air container plus an auger valve I made myself did seemed to work during tests but it failed when I wanted to use it.

      So I moved one to use an industrial Moineau pump I borrowed from Food Technology department as the source of pressure for my fluid joined with a T-join that diverted the flow to a feedback flow back to the pump and the other path went to my auger valve.

      Again, the failure was due to the auger pump not doing a great job and a clump eventually developing and blocking that flow completely.

      The idea of a direct drive from direct drive from the industrial pump sounds promising but a big nema34 would be required to replace the 1CW motor the pump uses. The AC motor with a variable frequency drive cannot be controlled accurately enough to do a controller extrusion.

      So my two new attempts ended up with failure at the end, though both seemed to work at small scale :-(

      Delete
    2. What commercial Moineau pump did you use?
      Clever approach to create a feedback loop. I already started typing a "you should not use a positive displacement pump as feed for the auger" until I read it again and understood what you did.

      Before I comment in greater length, when you say "auger" are you referring to the Moineau design on thin givers which you've been working on, or the auger pump described in the above post? They look similar but operate very different :)

      Delete
  17. Hi I'm a HS CAD teacher and we are looking to add this to the 3d printer that my students built from scratch. Do you have any suggestions or concerns before going ahead and starting this project?

    Thanks,
    Shane

    ReplyDelete
  18. I used a pump like these http://www.inoxpa.com/sectors/product/progressive-cavity-pump-kiber-ksksf/description/food-processing-sector couple with an auger valve I designed http://www.thingiverse.com/thing:224503 which did not prevented water from separating from the other matter in my solution which caused jams.

    ReplyDelete
  19. Thats a massive moineau pump Miguel!

    Thanks for the link to your auger. We did some thing very similar here before summer. Together with an engineering student we tried to solve some of the issues in the auger design. First we scaled the auger up at least ten fold since the one discussed in the above post was so seriously under-dimensioned, and than started to play with various parameters. Long story short (I will post a follow up on this in the next two three weeks, it will also appear in the next RepRap magazine) we never got very consistent results. The original observations still stand. But another issue we came buy very regularly seems to be mirrored in your experiences (and I am actually really happy that someone else has observed the exact same behavior): a clump formed at the end of the auger, right before the nozzle. We have no clear understanding why this happens but one theory was that there was water back flowing in the pump, up the auger. This leads to the clay -which is a suspension- to be separated into more solid clay and water. This was backed up by some of the clay up in the auger feeling slightly softer. I think I need to talk to an auger expert to ask for input on this. I haven't had this with the moineaus...

    Using your size moineau will probably be challenging because of the size, but who knows...
    I'll try to upload my further Moineau changes together with the new post. I still think they give the best results, even without a rubber seal.

    ReplyDelete