P1 - Buddha Tape Feeder

[Daren Schwenke]

I had a dream three days ago.  It showed me this...  I built it.

I had the itch to design a semi-auto tape feeder the week prior, and was trying to use ratchets turning toothed sprockets.  The previous attempts used gravity for the ratchet return, and sucked.

This one is positive acting, and checks all my boxes, minus one.  It's gotta have that Buddha belly sized cover tape reel to maintain the proper cover tape speed/tension.  Then again it's a catchy name...

• Consumes 3.5mm of width, so 8mm tapes use 11.5mm.
  
• Advances the tape on release, and has a built in adjustable clutch driving the cover tape removal.
  
• Supports tapes from both bottom edges, pushed up and centered with respect to the feed edge.  Pickup point is on the sprocket so the bent tape means it's always squared up.
  
• Tapes from .2mm-.8mm edge thickness supported.  Print a new cover window for the oddball thicker tape.
  
• Tapes from 8mm to 24mm width supported, print 2 parametric parts in about 30 minutes to change the width.
  
• Tapes with dual feed holes drive both sets.
  
• Bearings, screws, and pins where it matters.
  
• Only one small, easily replaceable component experiences sliding of the tape.
  
• Everything prints flat with no bridging and minor overhangs.
  
• Stackable.  Subtract out the center post and make a whole row of them with some 5/16in threaded rod, and the bearing center will take all the compression load.  I need to rework the cover tape clutch for this to be viable though.

I'm just to the point where I'm happy with it, and have done absolutely zero assembled testing... but all the bits work by themselves.  :)

I still have to design the tape cover/peel slot, but it's printed flat and bent into place so that will require some iteration, and the rest of it printed first.

The largest part is printing now. Solid with a 0.3mm nozzle, 0.1mm layers and 30mm/s speed should print in 2.2 hours.

Yes, you need a .3mm nozzle and a well tuned 3D printer to print the ratchets and drive teeth.

Let me know what you think, and what you would change.  

I'll release the source here once it's 'done'.

Project home: https://hackaday.io/project/159792-p1-buddha-tape-feeder

Ratchet video: https://youtu.be/3F3Mts_Vd04

[Олег Перевышин]

Very cool!

пятница, 27 июля 2018 г., 7:55:13 UTC+3 пользователь Daren Schwenke написал:

[Michael Anton]

Impressive.  To support 24mm wide tapes, you need to be able to handle up to 12mm deep pockets (according to the EIA 481-D spec, though the Onsemi spec you used specifies 11.5mm).  Do you have the clearance for that?

You may want to rotate the cover tape pickup down, so that there is more room for larger pickup heads.  At the moment, it looks like it sticks up quite high, and some machines won't have clearance for this.  Also, making the cover tape spool deeper so it can collect more tape would be a good idea.  It is not uncommon for the cover tape to be 20m long, if using a full reel.

On Thursday, July 26, 2018 at 10:55:13 PM UTC-6, Daren Schwenke wrote:

I had a dream three days ago.  It showed me this...  I built it.

[Daren Schwenke]

On Friday, July 27, 2018 at 2:01:18 AM UTC-4, Michael Anton wrote:

Impressive.  To support 24mm wide tapes, you need to be able to handle up to 12mm deep pockets (according to the EIA 481-D spec, though the Onsemi spec you used specifies 11.5mm).  Do you have the clearance for that?

To have clearance for 12mm tape depth, the spool area would have to be 24+22+1.5+4 tall, so you would be picking up those parts at least 54mm up.  Is that too tall?  If that works, then it should work fine. I figured that was getting too tall, so I just said 8mm max depth, but 24 width.

You may want to rotate the cover tape pickup down, so that there is more room for larger pickup heads.  At the moment, it looks like it sticks up quite high, and some machines won't have clearance for this.  Also, making the cover tape spool deeper so it can collect more tape would be a good idea.  It is not uncommon for the cover tape to be 20m long, if using a full reel.

K.  That moves the feed arm back and down as well as they are coaxial.  Putting the feed arm on the cover tape sprocket gave me back almost 2mm of width so I would like to keep it there.  

Also, the tape entering and exiting both have to clear underneath it.  But that was one of the numbers I got the parametric stuff right for, so I can mess with it to implement suggestions.

I screwed up some other math though I'm still fixing on.

Thanks for the tips! 

[Michael Anton]

On Friday, July 27, 2018 at 12:15:24 AM UTC-6, Daren Schwenke wrote:

On Friday, July 27, 2018 at 2:01:18 AM UTC-4, Michael Anton wrote:

Impressive.  To support 24mm wide tapes, you need to be able to handle up to 12mm deep pockets (according to the EIA 481-D spec, though the Onsemi spec you used specifies 11.5mm).  Do you have the clearance for that?

To have clearance for 12mm tape depth, the spool area would have to be 24+22+1.5+4 tall, so you would be picking up those parts at least 54mm up.  Is that too tall?  If that works, then it should work fine. I figured that was getting too tall, so I just said 8mm max depth, but 24 width.

That can probably be dealt with, by lowering the mounting surface for the feeder.

It looks like the bend radius of the tape may need to be 30+12mm, so a sprocket diameter of 84mm may be needed to support tapes that deep anyhow.  I have feeders that don't obey the bend radius for 8mm wide tapes, but for deep pocket tape, that radius may be required to allow the tape to bend, and not have the pockets hit one another.  Alas, I don't have any sample tape to check this...

You may want to rotate the cover tape pickup down, so that there is more room for larger pickup heads.  At the moment, it looks like it sticks up quite high, and some machines won't have clearance for this.  Also, making the cover tape spool deeper so it can collect more tape would be a good idea.  It is not uncommon for the cover tape to be 20m long, if using a full reel.

K.  That moves the feed arm back and down as well as they are coaxial.  Putting the feed arm on the cover tape sprocket gave me back almost 2mm of width so I would like to keep it there.  

Also, the tape entering and exiting both have to clear underneath it.  But that was one of the numbers I got the parametric stuff right for, so I can mess with it to implement suggestions.

I screwed up some other math though I'm still fixing on.

Perhaps putting a bend in the feed arm would potentially help move the feed location to some place more desirable. 

[Daren Schwenke]

One more question.  When is the best time to pull the cover tape?  

I assumed it was during the feed move so the tape is pullled while the tape advances. 

Do some machines pull on the opposite stroke? Does that work?  I could see the tape going up under the slide and being pulled back to the peel point and it working fine as long as the tape itself isn't allowed to retract.  Perhaps even better as the first time the components are exposed then would be at the pick window.

Thoughts?

[Michael Anton]

Pulled while the tape advances is common, but only expose the part to be picked, as other exposed parts are likely to jump out of their pockets on the next advance.  Some feeders have a sliding cover that keeps the parts covered, until it is time to pick them at which point the cover slides back to expose the part.

[Daren Schwenke]

So, i suppose I could have the cover tape still over the ready to be picked part... and engaging the lever arm is what pulls the tape back.

This would presume that moving to that station is what engages the lever arm, and not a separate operation... because as soon as you leave the next part will advance, but still be covered.

That won't work for my application as the whole head goes up and down, but would be easier to implement than what I currently have.

As for my operation, I could just move the cover tape peel post back 1 section, so it would peel before.

I have another idea for my application though, so I probably won't.

The force required to both retract the tape and advance the part on release is requiring additional structural thickness to handle the spring load required.  If I can split the operations into either motion.. that would be better and my cover tape spool can then turn in the other direction.. which will reduce the vertical space I need considerably.

Thank you for your input.  I appreciate it.  I'm still building something here I've never used yet from specifications and community experience.

[Michael Anton]

The auto feeders on my machine advance under spring pressure, that way a clutch is not needed on the sprocket.  You do want to limit the force on the sprocket as well, as if it jambs, which is not uncommon, you will tear out the tape holes, or in the case of plastic sprockets, potentially damage the teeth.  The advance of the cover tape spool is also done under spring pressure.  The feed lever cocks the cover tape advance, but it doesn't happen until the lever is released, and only then if there is tape to take up as well.  This way a clutch is not required.  It is a bit more complex to implement this way than what you have though, but there is nothing to ever adjust using this method, and you are not relying on friction to set the required force.

The feeders I have also have pawls to stop the sprocket, and cover tape spool from rotating in reverse.  You might find that this is required once there is tension on the cover tape, and you try to get the ratchet to return to the start position.  The friction in the ratchet will try to drive it backwards, especially with tension on the cover tape trying to pull it in the same direction.

I'll try to take some pictures at some point so you have a reference.

[Daren Schwenke]

I have implemented ratchets to prevent reverse rotation on the cover tape peeling and the tape feeder, but I do not have a clutch on the tape.   I can implement one but it will cost me one millimeter of width then and you currently need to adjust the clutch from the side.

The current design does have both the cover tape and the feed advancing under spring pressure, but that is proving to be a lot of spring pressure hence my reasoning for trying to split the operation.

There is a difference between the advancement motion and the pulling the tape back to latch on the ratchet though so having the operations be separate maybe advantageous anyway.

--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/7DpGipBIRvE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b9ed1cb4-83d9-4dc9-9280-1996289e16fd%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Daren Schwenke]

Reworked the model so everything is parametric now.

You can change the offsets, gear ratios, sprocket and ratchet tooth count, thickness, and even the cover spool direction on the fly and everything scales.

[Daren Schwenke]

Source is here: https://cdn.hackaday.io/files/1597926836540928/Arcus-3D-P1-Automatic_tape_feeder.scad

Still missing the tape cover, but you can play with the model now if you like.

Let me know what you think.

[Daren Schwenke]

Reworked it this weekend.

Moved the cover spool down to flush with the drive spool.

Went from using a pushrod to a pull, which I'm much happier about.

Changed from a binder clip to a pen spring, and it's now on the drive pulley ratchet instead.

Tweaked the gear ratios and ratchet positions so each lever push locks at the next ratchet position at 2/3rds throw, so anything more than 2/3rds is still one advancement.

Added servo support.  You should be able to run two of these with one servo.

Maintained the ability to change the cover spool direction in code, and you only need to reprint 2 parts.

I settled on 12mm width.  It was a lot easier to make it strong enough supported from one side that way.  If you are stacking them, you can go back to 11.5 or even 11mm.

I'll update the source repo later today.

On Friday, July 27, 2018 at 12:55:13 AM UTC-4, Daren Schwenke wrote:

[Anthony Webb]

This is really great work!  Are you planning to post the STL files? I'd like to learn some more about the design by putting one together and testing it.  Where are you anticipating attaching the servo?  In the case no servo is used are you thinking these are a bump feed?

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/edd0bb62-97a3-4f25-926d-c63d14b491a3%40googlegroups.com.

[Daren Schwenke]

On Monday, July 30, 2018 at 10:36:28 AM UTC-4, Anthony Webb wrote:

This is really great work!  Are you planning to post the STL files? I'd like to learn some more about the design by putting one together and testing it.  Where are you anticipating attaching the servo?  In the case no servo is used are you thinking these are a bump feed?

Yes, bump feed.  The notch on the back is for the servo. 

The source I keep updated up on hackaday.  OpenSCAD source... render it yourself.  It's getting a little slow as I've started 'cheating', but hey..  it works.

https://hackaday.io/project/159792-p1-buddha-tape-feeder, under files.

On Mon, Jul 30, 2018 at 8:10 AM Daren Schwenke <darens...@gmail.com> wrote:

Reworked it this weekend.

Moved the cover spool down to flush with the drive spool.

Went from using a pushrod to a pull, which I'm much happier about.

Changed from a binder clip to a pen spring, and it's now on the drive pulley ratchet instead.

Tweaked the gear ratios and ratchet positions so each lever push locks at the next ratchet position at 2/3rds throw, so anything more than 2/3rds is still one advancement.

Added servo support.  You should be able to run two of these with one servo.

Maintained the ability to change the cover spool direction in code, and you only need to reprint 2 parts.

I settled on 12mm width.  It was a lot easier to make it strong enough supported from one side that way.  If you are stacking them, you can go back to 11.5 or even 11mm.

https://hackaday.io/project/159792-p1-buddha-tape-feeder

[Daren Schwenke]

Animation of the disparate drive ratios due to the connecting rod placement.

https://youtu.be/9I4Ee7dsj3g

On Friday, July 27, 2018 at 12:55:13 AM UTC-4, Daren Schwenke wrote:

[Daren Schwenke]

Finished tuning the tape cover dmensions, and did a video.

Pardon the orientation.  I swear I didn't record it 'sideways.

https://youtu.be/DwUKRlJsiOY

On Friday, July 27, 2018 at 12:55:13 AM UTC-4, Daren Schwenke wrote:

[Marek T.]

Great job but understand it requires zyz to advance the tape then yzz to pick the part. A lot of moves means time losing. In my opinion it would be more nice if advancing just with single zz like normal machine feeders have.

Second, test it with lightweight parts like 0402 to check if the advancing is not to rapid and parts don't jump out. I don't tell it is but it's for sure problem that cannot to occur.

[Mike Menci]

Hello - yes very interesting
There is no need to trigger it mechanically with Z - if you add a bracket X or Y could trigger this Buddha and
Marek it has un option to trigger it with servo (1 servo for two feeders) which might work better. See the video.
- are there .stl files avalible - I could not find those on Hackaday web?

Mike

[Marek T.]

If there is a servo then ok. If the intention is also to use it on machine without servo then the fastest operation is single Z like in Yamaha. Sorry, maybe I'm fixed to much on yamaha but really think it's really reasonable solution they made.

[Mike Menci]

Hello Daren, 

From the video I think I see sprocket holes as indicated in "Yellow" on enclosed sketch. 

Note that SMT carrier tape spools have sprocket holes on "left" (as indicated in 2nd enclosure) and it seams that location of sprocket holes should be on opposite side on your Buddha feeder. ??? 

I might be wrong ... because I do not know how will it will operate !?  Is the cover tape spool orientated towards PnP machine or away from it ? - where is the charging tape location and direction ? 

Please clarify

Mike 

[Brynn Rogers]

He is okay with the sprockets, the feed direction is to the left in the 'P1 Buddha tape feeder - YouTube.jpg'   photo.

One question I have is what size components are used in the demo video?    I am guessing 1210 resistors.  They look pretty square.

I also thought the 'pick window' maybe should be a little bigger for this size component, and maybe smaller with smaller components.

What is the life of the 3d printed ratcheting parts?     Seems like those ratchet and pawl parts should be steel.

Good job, clearly much less money to build than an off the shelf feeder.

Brynn

[Daren Schwenke]

I don't know how long they will last.  I have the first one I printed still though and it has no detectable wear.  

They are designed that on engagement they expand out a tiny bit so they are contacting the teeth over their entire length.  

In other words the force gets transmitted along the length and is supported, so the ratchets can be long and thin enough that they can bend without deforming. 

[Daren Schwenke]

Yeah, basically, but you could use the carrier instead of the nozzle to engage them.  That would reduce it to a Y move.

The advancement speed will be equal to the retract speed you program on the Y move, so you should be able to control that.

I don't have any 0402 spools here.  I only have 3 spools in 2 widths for testing, all with huge parts.  Someone else will likely need to give it a try.  :)

[Mike Menci]

Daren 

Please load all .STL files somewhere, and your bearing size please,...  so I could try it out. 

Thanks Mike 

[Daren Schwenke]

Made a repo and rendered the STL files into it.

https://github.com/Arcus-3d/P1-Buddha-tape-feeder

It's a standard skateboard bearing, aka 608-2RS.  Dimensions are 22x8x7mm.  

[Daren Schwenke]

Been playing around with it... I think I may need to move the cover tape peel slot back a tiny bit.  How far back it is from the part slot controls how far the cover tape is retracted.

I have also have never tested anything other than the 8mm width, although I generated the STL files for 12mm and 16mm.  I think the 24mm width may be constrained to using two bearings and being mounted on threaded rod to avoid excessive wear, and I have not generated that as such.

Print it out, try it out, and let me know what you think.  I (or you) can generate new dimensions for the slot cover opening as needed.  A different one for each part dimension/offset is relatively easy.

The parts that are designed to be slip fit (the lever arm, the cover tape drive, and the tape drive), need to be slip fit, or the whole design will fail.  No play, but also low friction.  You may need to play with over/under extruding to make that work since you are starting with the STL.  In the source, you can adjust a variable.

If you don't feel like diving into the world of OpenSCAD, just tell me what direction and how much, and I'll generate a new STL file for custom slot covers for each part.

Thank you.

On Saturday, August 4, 2018 at 4:18:36 PM UTC-4, Mike Menci wrote:

[Marek T.]

Still when you get mispick and need to pick again you must do y+, y- and then pick with z. So two y redundant moves required to perform.

[Marek T.]

And one more thing. Generally rolling the covering tape onto the roller is popular solution but not very convenient in use. Much better is using two sprockets with the tape between them. Faster to place there the tape and to remove it when finished.
But it's only my opinion ;-).

[Daren Schwenke]

Yeah, what's worse is you currently have to adjust my cover tape tension from the side.  I imagine that will be pretty inconvenient as well if you have a stack of these.  

That's part of the reason the slot cover is removable now, and the reason I have another enhancement rolling around in my head to make the cover tape spool 'drop in'.

Eventually.  Right now I just need it to work so I can switch back to my main project, and I can come back to that later.

I appreciate the opinions.  Thank you.

As for having the pick move do the advancement.. That was the plan, but I ran into an issue.  

I was trying to keep the force required to do the advancement as low as I can as my target platform can't generate a whole lot of force in the Y dimension.  This meant using a smaller spring, which meant I no longer had enough force to do the cover tape pull move also on the release stroke.  I am using tape that has been 'expired' for 10 years, so perhaps it's harder than it should be to pull now, but that's the amount of force I designed for.

Not doing the cover tape pull on the release stroke now means that the tape reel moves back a little bit before the ratchet engages to stop it.  Only when the lever is released again are you guaranteed to be at a known position.

You can still change the direction of the cover tape removal to be on the release stroke again by switching a 1 to a 0 in the code, and reprinting three parts should others want to experiment with that solution.

[Marek T.]

Yeap, the feeder seems be simple device but it only looks like this. In real it's quite complicated when you start to dig in it :-).

[Олег Перевышин]

Good job! Thank you for your work!

Is this the final version or will there be a change?

Please make the model in STEP format.

[Daren Schwenke]

The cover tape needs a little more extra advancement I think.  

I changed it from pulling 125% to 112.5% at the last revision, which is not quite enough it seems.  I need to put it back.

Busy atm.  I'll make the changes and have the new files up here by Saturday.

[Marek T.]

Haven't you thought to make optional change to replace plastik sprocet with laser cut?

0.5-1mm cutting is cheap and usualy no problem to do it in companies who makes laser paste stencils.

What the servo you have used there?

br

Marek

[Daren Schwenke]

Yes, but now the design has evolved to use the inside for the ratchets, the left for stability, the right would extend into the 'part space', and the outside is the teeth.  

I would need to add a way to mount it and all surfaces are 'consumed'.  So.. It would probably make it .5-1mm wider to do it.  I think I'm just going to keep it all 3D printed now.

The teeth are actually leaning into the feed direction by ~5 degrees to account for the minimum 3D printable radius at 0.3mm also.

It's the standard, knockoff, cheap analog servo - MG90S.

[Marek T.]

I thought about to remove your plastic sprocket (then your main roll "connected" to it would be thinner for these 0.5-1) and place there metal sprocket instead. Then same widthness together like now. Mounting could be easy, just two small 2mm holes in metal sprocket and tho plastic pins on your main roll.

Advantages of metal no sense to tell, it's obviously better except that cannot be 3D printed :-).

Sure if you find plastic is solid enough and leading good enough - it's ok.

Found today nice video showing philips feeder how works and designed. Maybe useful for you as idea/technology recognition. It's old but still good.

https://www.youtube.com/watch?v=qUeRnI4IwLo

br

Marek

[Daren Schwenke]

The inner ratchet teeth go all the way out to the outer rim, minus 2*nozzle_dia.  The center hole which centers/connects it to the spool idler is the thickness of the teeth.  So to remove the plastic which is the teeth and replace it, I would need to either make the ratchet smaller (it works a bit better when it's larger), steal the thickness from the ratchet teeth which are already only 1mm deep, or add some thickness.

Thanks for the video.  I see what you mean for the spool and the vertical advancement mechanism.  

If I have to time to rework a vertical version, I will.  Since my pivot point is back on the cover tape spool, changing it to vertical might not be that hard actually.

[Marek T.]

If your plastic sprocket works well and position is stable then no sense to combine with steel.

If I will find, saw somewhere, will give you some another nice video about the rolling tape.

This vertical advancement has one advantage - no need to perform any extra moves to advance the tape but pick the part.

Generaly I like your project and try to print it to test :-)

[Marek T.]

See here. Left-down corner two black sprockets to advance covering tape between them. Very fast to mount the tape between them and remove from the feeder when finished - in fact it's always staying removed :-).

https://www.youtube.com/watch?v=nBnfRhdnPks

[Daren Schwenke]

I think I can do that.  The existing cover tape spool could be that outer shape pretty easily and the same drive ratio would work.

I'll need to add that idler spool, another pivot point, and another spring somewhere though.  

The bottom and right are currently consumed by clearance for deep parts for the incoming tape path, putting it up top will add height (unless I can significantly tweak the gear ratio for a smaller cover spool), and left is servo right now though.  

Some stuff would need to change.  I would probably go for up top.

--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/7DpGipBIRvE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/2a13dd7b-7eb7-43dc-a86a-f7c94fcba49f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Marek T.]

Sure this is your project and your decissions only. But from my point of view you go to ideal direction ;).

Very nice is also that total width of your feeder is only 12mm. It allows to mount them many without the raw space wasting.

[Daren Schwenke]

FYI, files are not updated yet.  

I blew through most of today battling my printer trying to do tiny single wall parts in TPU, so I didn't have the the time to make the changes I want to yet.  

Tomorrow.

[Marek T.]

Thx for keeping updated.
Easy, if it's going about me I'll try your final art with pleasure but I'm not in hurry.

[Daren Schwenke]

Changes are done, but I still have yet to re-tune the stationary ratchet positions to match the new drive ratio.  

Where they are and exactly when they engage is important, and I'm done for today.

Tomorrow, probably.

[Олег Перевышин]

Tried to use smaller bearings, so you can skip the tape thicker or are there any special features?

[Олег Перевышин]

As an example 624 (4x13x5mm) - 22$ 100pcs on AliExpress

[Олег Перевышин]

and 603 (3x9x5mm)

четверг, 16 августа 2018 г., 11:25:31 UTC+3 пользователь Олег Перевышин написал:

[Олег Перевышин]

And just a good option HK081210

[Daren Schwenke]

Only issue would be you would probably have to make it a little thicker so you could have a surface to bolt to. 

Right now I'm using the larger ID to let me put the head of the bolt inside the bearing.  (and more importantly, I have ~100 of those bearings laying around)

I have a few of the other sizes here also, but only a couple of each.

Yes, that would let you have a smaller spool diameter and get more tape depth clearance.  

The bearing dimensions are a variable you can play with in the OpenSCAD model if you like.

[Marek T.]

Daren when do you plan to release your final (or almost final) version to the files?

This week, this month? Not any pressing just only to know when may try to print it out :-)

[Daren Schwenke]

I'm on the other task atm as the machine it's going in is getting assembled today.  

After changing my dimensions, the design fails to advance properly.  It's 'sticky' now, so it still needs a bit of tweaking.

Also, a friend of mine suggested I implement DIN rails as a good way to secure these down without needing an array of holes in my platform.  

I liked the idea, so I'll be implementing that as well first. 

[Marek T.]

But goods mounted on the DIN rail have not fixed position like using holes array.

So how will you find X-center position of the feeder if it will be not constant, fiducials on the feeders???

[Олег Перевышин]

Pay attention to the bearings EWC0608 (6x10x8mm) (they are one-sided, they rotate only in one direction, in the opposite direction they are blocked).

There may be some ideas on their application))

[Daren Schwenke]

That's just about perfect.  Larger ID with a relatively small OD, same depth, and no reverse ratchet needed then.

[Daren Schwenke]

That is a good point.  Don't you have to do alignment now anyway?  Yes, being on the DIN rail though we only need to find the X position. 

I hadn't thought of using ficiduals...  I may have to steal that idea.

I was looking at mounting this the normal way, and I would need holes every 4mm for every combination of widths, unless I provided offset mounting holes on the feeder, then it was 8mm.

Still, that was too many holes I thought and would limit adoption to those with a laser cutter. 

I'm trying to keep the tools required for the entire project to be hand tools, maybe a drill press, and a 3D printer.

I suppose I could print a DIN rail with raised notches every 4mm, and then if you use a regular DIN rail it slides, whereas using the printed rail, it locks at 4mm intervals.

I think I'll do that.

[Marek T.]

But with typical DIN rail you have to seek this x position after each feeder changing, even when you take it off for a while to replace the reel.

Holes array keeps every feeder always in the same position. So you program it only one time till you don't take off the plate (however plates also have positioning pins and holes). If feeders are calibrated well, you even don't need find their zero if replace ie feeder 7 with 12 etc - they are every the same.

Usually feeder plates just have holes in raster like 13mm, same like thickness of the 8mm-tape feeder + some clearance (ie 1mm). If you mount 12mm-tape feeder that is wider - you just loose one feeder place that's all.

But no problem I can cut your din holder off from the feeder and add just plate with pin :-)

[Daren Schwenke]

Did a mockup of what I was thinking.  DIN rail dimensions, but with raised teeth at 4mm spacing.  Corresponding teeth inset in the model.

Parametric, so you can pick your length and print it, and it always renders out as multiples of 8 so you can chain them.
I did a hollow one too where I control the inner ribs, but I think I'm just going to let the infill handle it.

So now regular DIN rail works, but no index.. use this DIN rail, you have index.

I suppose I could print an index to just fit inside the DIN rail as well... better locking as that surface will be steel/AL then, and then I still have the indexing.

Thoughts?

[Marek T.]

So you want to print an index fitting to typical metal DIN rail?

[Daren Schwenke]

Yep... something like this.  

Fits inside normal rail.  Bottom of the notches match top of rail so the teeth extend up above the rail into the feeder. 

Matching index inset into bottom of feeders, gives you 4mm spacing (which is what I need for no wasted space and all widths).

[Daren Schwenke]

Only issue with doing it this way, is those teeth rely on layer adhesion to keep from breaking off.  The previous one the teeth would be printed on the bed and so would be much stronger. (like my drive teeth)

[Daren Schwenke]

Oh..  two bad things.  

1. I use the outer surface of those bearing as a sliding bearing for my drive ratchets, and they are not round/smooth like regular bearings.
2. The stop function of those would rely on passing the force on as compression/twisting to the center post, which is currently plastic.

I think regular bearings would end up being more reliable in the long run.  Good idea though...

[Trampas Stern]

Consider clutch bearings

[Marek T.]

Sounds good...

[Daren Schwenke]

Even better idea. Lego. Post spacing is 8 mm so I'd have to design it to work at half spacings as well but then the bass is done for me. I could use the same system to hold the boards as well.

[Marek T.]

But don't you think that feeder holding is less important than the main feeder mechanism finished?:-)

[Trampas Stern]

The best way I found for the take up cover is to have it spring loaded. This way the take up cover constantly has tension applied. 

I personally like the way the Yamaha feeders do the cover tape, that is they do not wrap on a pulley but use pinch rollers and discard out bottom. This way when you change reels it is less of a mess. 

Most designs fail when it comes time to make the toothed sprocket for the tape feeding, ones that get past that often fail when it comes to keeping parts from jumping.  usually they find that a slit has to be made for the cover tape to come out and then leave the parts covered for a part or two before opening for pick. This keeps the tape from jumping as cover tape is ripped off.   Another failure point is when they realize they need different advancement distances for 0402 verses 0603 verses SOT23 or microprocessors. 

As a boss once told me "Any engineer can design our product and get them to work, I hire engineers that can design the products and have them work for our 20 years." (Which meant working 20 years from a non replaceable battery.)  The point is anyone can design a feeder that works, designing to work reliably over all operating conditions is the hard part.  The only way you get there is to build, measure, learn, and repeat until you beat it into submission. 

 

[Daren Schwenke]

On Friday, August 17, 2018 at 6:40:06 AM UTC-4, Trampas Stern wrote:

The best way I found for the take up cover is to have it spring loaded. This way the take up cover constantly has tension applied. - before I kept track.  3 obvious versions. Constant tension via a clutch.

I personally like the way the Yamaha feeders do the cover tape, that is they do not wrap on a pulley but use pinch rollers and discard out bottom. This way when you change reels it is less of a mess. - Suggested, planned.

Most designs fail when it comes time to make the toothed sprocket for the tape feeding - 7 versions.

 

ones that get past that often fail when it comes to keeping parts from jumping.  usually they find that a slit has to be made for the cover tape to come out and then leave the parts covered for a part or two before opening for pick. This keeps the tape from jumping as cover tape is ripped off. - 22 versions 

Another failure point is when they realize they need different advancement distances for 0402 verses 0603 verses SOT23 or microprocessors.  Part of the 22, but not tested.

 The ones I kept.  Some stuff from the P1/C1 in there too.  The past is paved in PLA.

As a boss once told me "Any engineer can design our product and get them to work, I hire engineers that can design the products and have them work for our 20 years." (Which meant working 20 years from a non replaceable battery.)  The point is anyone can design a feeder that works, designing to work reliably over all operating conditions is the hard part.  The only way you get there is to build, measure, learn, and repeat until you beat it into submission. 

Beating in progress.  The current issue is balancing the force required for the above with what my pen spring can provide.  I think I still need a little bit more.  

 

[Brynn Rogers]

So if I want to print one out, are the STL files the latest and greatest?    I should I always render my own from SCAD?   (I haven't messed with SCAD much)

Brynn

[Daren Schwenke]

The rendered STL files are for 12.5% extra pull on the cover tape. I noticed I needed to either pull it slightly more, or move the cover tape slot back a little.  I tried the former. 

I adjusted that extra pull up to be 25%, but that proved to be too much for my current spring (position) and so the return stroke sometimes doesn't pull quite enough to put the tape drive sprocket back onto the hard stop, which means the tape position can vary.  Sucky.

So... what that means is what is currently rendered out to STL is the best I got.  I haven't moved the slot back.  That might work too and is a tiny part.

The scad that's up is the old one, as the new one has more issues... 

Yes, you should render from the scad if you need to change anything.  Otherwise, use the stl's.

I'll render a new cover slot if you want to try it like it is, but I'll be making other changes as well for mounting, and getting a little more torque out of my spring, which will likely change 60% of the parts.

[Brynn Rogers]

Okay, thanks.  

   I just figured there was a bunch of printing to do, and for my machine I will likely want to have an air cylinder advance the tape (because I have a bunch, with solenoids) , so while I am waiting on other parts I could heat up the printer and let it make something.

Brynn 

[Daren Schwenke]

By all means then, download the scad and muck around with the servo mount to adapt your cylinder.  I'll give you a hand if you need help interpreting my gibberish.

Add a tiny bit more spring tension there (if your cylinder has a spring return... which it probably does) and everything will probably work as intended.

I was trying to keep the required force on the lever as low as possible because I was needed to push it with an end effector supported by strings...  That has proven to be pushing the limits.

[Brynn Rogers]

Hmm,    I'm going to need some other changes, too.    I have very little clearance above where the feeders are, like 4mm above the top of the components.

So the bump arm will need to end up below the ratchet wheels on mine.  Actually it will turn into an actuator arm or something.

I'll also need some way to mount, with like a pin at the front that fits in a hole on a rail.  And maybe another pin down under to keep it perpindicular (sp).

 

Brynn

[Daren Schwenke]

There's a short version of the arm in there already called lever arm servo or something like that. The rest I will actually have to understand to implement.

[Daren Schwenke]

Hmm.  Well, you will be hard pressed to use it like it is then.  

4mm isn't enough to clear the ratchet which drives the tape feeder spool either.  It sticks up about 6mm, 2mm from the hole edge of the tape.

I needed it that far from the pivot to give the spring enough torque to drive the advancement, and that's the only place it can go where it won't add thickness to the design.

[Daren Schwenke]

First off, skip to the end if you are a mechanical wizard with a better memory than me.  I could use some help with something.

I decided the height of the Buddha feeder was going to cause too many issues with my current application.  

Offset height mounting seems to be fine for most of you all, but will not really work for me for my current application as envisioned.  

I'm trying to go cheap/easy for the masses so a single piece of plywood constitutes my frame.

Hence...  this was born.

Basically it is everything Marek wanted combined with everything I wanted.

• Lego plate mounting, just cause I think it's going to be so cool and more importantly, work well.  :)
  
• It's a vertical feed, push down to compress springs (and pick up the part), and on release it then feeds and retracts the tape.  Or, use a servo and eliminate the feed arm (which is my application).
  
• Cover tape removal is now pinch roller based, and uses the same 'pull more than needed/clutch' as the Buddha, but now the retract is done on the feed move so no possible backwards motion or relying on counter rotation ratchets to engage, during the pick move.
  
• Tape advancement will now be via a bogey with a pin, and the end position has a fine position adjustment via a 3mm screw.
  
• Tape is supported on both sides and held up against the top surface with piano wire 'leaf' springs.
  
• It is now designed with the channel for the far edge of the tape included as the back face of the next feeder.  You HAVE to stack them now (or print out an end part).
• Using the Lego plate combined with this geometry gives me every possible tape width without adjustments.  Lego plate spacing is 8mm between posts, and the feeder mount is designed to securely snap at either half or full spacing giving me the 4mm spacing I need to cover every possible width.
• Width is still <tape_width> + 4mm.  Height is now <max_tape_depth> + 5mm and is now configurable in the model.  What is shown here is 16mm depth support which is actually easier for me.
  

Here is a video of me gushing about the Lego mount.

https://youtu.be/Y1LbbHuZtO8

I'm still searching for a particular mechanical tidbit I pulled from my memory though.  It was an elliptical motion design based on orbiting a pin I saw somewhere.  Kinda like how a cycloidal drive works, but you can initiate the orbit with just linear motion.

It would allow me to have the forward motion for the pin feeder grab the tape, but then the reverse motion move down and away from the tape while retracting instead of dragging across the holes.

If you recall what I'm talking about and have some kind of reference to feed me, that would be way faster than me trying to work it out again from scratch.  :)

If nothing else, I'm not above dragging along the holes, but this would be better if I can keep it simple/reliable.

Thank you all for your input.  It has been invaluable.  (and not in the Three Amigos sense)

[Marek T.]

Hi Daren,

First impression about the lego-click mounting - hopeless idea. Ok easy to mount and fast but:
Distance from base surface to the top of the tape is some 3-4cm (sorry I'm not at pc with scad now). So any minimal change of angle feeder-base due to not extremely solid keeping the feeder by lego-click will effect change of x position of the feeder top for some part of mm, means changing of the pick point position. The position of the top must be extremely stable what seems be not guaranteed by Lego system, maybe I'm wrong as I guess you tested it somehow.
What's your idea of reel storage? On some arm integrated to feeder like in many systems or independently? If on arm then we have large lever and not small weight of the full reel - in effect front v of the feeder (at the machine side) may be pulled up - means changing z pick position and still higher susceptibility for changing x...

Idea with using din rail plus plastic insertion seemed be more stable.

As said, maybe only the impression like that :-)

[Daren Schwenke]

You are right.  If the engagement of the pins did not bottom out, the resulting variance in the top angle would be enough to make this useless.  

So... I have made my Lego interface taller than the actual pins by about 1mm, so that each one bottoms out and is actually resting on the plate.

The lego plate serves just to provide the x,y positioning at a resolution I probably could not achieve on my own, and still provides a decent way to hold things down.

The angle and height is stabilized by spanning at least 3 posts, all of which are bottoming out.

For the reel positioning I was planning on NOT integrating with the feeder  That would be nice, but I honestly don't have the stability here for that.  

The short plan was to drill some holes near the edge and two bends later I have a swing arm for a reel.  However I'm back to drilling a bunch of holes in my build area again then.  

I will probably just hang them off the edge of the build area supported on both ends by a rod.  I didn't like this as changing the reel in the middle means unloading all of them, but it is far easier and requires much less stability than any of the alternatives.

Like most of what I'm doing here, it is untested and currently lives in my engineering nirvana.  But success or failure... then we will know and learn something from it hopefully.

[Marek T.]

Common rod for all the feeders is shity idea. Imagine the mess when you have 30 reels on it. And you must stop the machine for much longer time for as you said unload-reload all of them. Maybe some kind of shelf with the grooves where the reels could rest? Then if you take off the rod the reels are moreless kept on the positions? But reels have different diameters... Some idea but far from being perfect. Definitely the best is individual holder for each tape. I have some idea, will try make some sketch to show it.

Why have you given up an idea with din rail? It provides vertical holding very strong.

I'll try to find and show you how yamaha does feeders holding in their two different systems. Maybe some path to use or to consider.

[Marek T.]

Kind of plate fixed to the feeder working as reel holder.
It's steel but could be printed as 2mm PLA then should be rigid also enough.
But required strong vertical holding the front of the feeder against its pull off (up).

[Marek T.]

Forgot the link...
https://m.youtube.com/watch?v=mbpqPKfTcIg

[Daren Schwenke]

New version, and I finally have everything sorted.  

Smaller overall, smaller bearings, no pins minus the feed pin.

Everything prints flat, no supports, with max 4mm bridging required in one spot, 2mm in one other spot, and 0.8mm everywhere else (the tape feed channel on the back.

Supports the tape from the bottom edges with spring tension at the feed window.

Integrated pinch roller for cover tape removal.

Feed area is 20mm high and is currently configured for 12mm tapes.

Feed lever is 40mm high and vertically actuated.  ~14-18mm of travel will advance it.

The last 2mm of lever travel will disengage the feed pin for tape removal.

60mm total height.

140mm long including the servo. 110mm without it.

11.9mm wide for 12mm spacing.

Printing now.

On Friday, July 27, 2018 at 12:55:13 AM UTC-4, Daren Schwenke wrote:

I had a dream three days ago.  It showed me this...  I built it.

I had the itch to design a semi-auto tape feeder the week prior, and was trying to use ratchets turning toothed sprockets.  The previous attempts used gravity for the ratchet return, and sucked.

This one is positive acting, and checks all my boxes, minus one.  It's gotta have that Buddha belly sized cover tape reel to maintain the proper cover tape speed/tension.  Then again it's a catchy name...

• Consumes 3.5mm of width, so 8mm tapes use 11.5mm.
  
• Advances the tape on release, and has a built in adjustable clutch driving the cover tape removal.
  
• Supports tapes from both bottom edges, pushed up and centered with respect to the feed edge.  Pickup point is on the sprocket so the bent tape means it's always squared up.
  
• Tapes from .2mm-.8mm edge thickness supported.  Print a new cover window for the oddball thicker tape.
  
• Tapes from 8mm to 24mm width supported, print 2 parametric parts in about 30 minutes to change the width.
  
• Tapes with dual feed holes drive both sets.
  
• Bearings, screws, and pins where it matters.
  
• Only one small, easily replaceable component experiences sliding of the tape.
  
• Everything prints flat with no bridging and minor overhangs.
  
• Stackable.  Subtract out the center post and make a whole row of them with some 5/16in threaded rod, and the bearing center will take all the compression load.  I need to rework the cover tape clutch for this to be viable though.

I'm just to the point where I'm happy with it, and have done absolutely zero assembled testing... but all the bits work by themselves.  :)

I still have to design the tape cover/peel slot, but it's printed flat and bent into place so that will require some iteration, and the rest of it printed first.

The largest part is printing now. Solid with a 0.3mm nozzle, 0.1mm layers and 30mm/s speed should print in 2.2 hours.

Yes, you need a .3mm nozzle and a well tuned 3D printer to print the ratchets and drive teeth.

Let me know what you think, and what you would change.  

I'll release the source here once it's 'done'.

Project home: https://hackaday.io/project/159792-p1-buddha-tape-feeder

Ratchet video: https://youtu.be/3F3Mts_Vd04

[oliver jackson]

Looks great! Is there a non printed parts list as I guess the bearings have changed 

On 28 Aug 2018, at 12:27, Daren Schwenke <darens...@gmail.com> wrote:

New version, and I finally have everything sorted.  

<Screenshot at 2018-08-28 07-19-57.png>

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/fd0f364a-2235-4407-9868-caf4084f9e56%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

<Screenshot at 2018-08-28 07-19-57.png>

[Daren Schwenke]

2 bearings 4x13x5mm.  I forget the number atm..

1 M3x5mm bolt

1 M3x10mm bolt and nut

1 M4x10mm bolt and nut (I may need to cut down a 15mm actually, we will see).

A short bit of piano wire between .8mm and 1.2mm dia.  A safety pin would work too.

A handful of silicone hair braiding bands.

I'm guessing about 25g of relatively stiff filament and a .3mm nozzle to print it with.

[Marius Liebenberg]

Great stuff. I can't wait to print one. Will you let us know when you have updated Github please?

[Daren Schwenke]

I have to print it and check that my tolerances held up ok first.  

That's a long span between the servo leg and the main support.  I may need to account for shrinkage.

Also, the tape feed slot may be a bit tight or it may need a 1 nozzle width lead-in where it is on the bed.  

I'll find out in a couple hours.

[Anthony Webb]

This looks great! Would love to see a video of it in action when you get to that point as well. 

Sent from my iPhone

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/c380369a-280a-46de-8ca0-a8ff3ae8371e%40googlegroups.com.

[Daren Schwenke]

If this pans out, the first test of the whole system together is slated for Thursday at the 'local' makerspace.  

However, I'm already sleep deprived and some things still don't work.

Like for example, I can't get bottom and top vison working using the same physical device.  

When I get back to that part, I'll sure I'll make a new post pleading for help.

Also, my internet connection at my current location involves the use of directional antennas, amplifiers, and a 90ft pine tree, so video uploads can be painful.  

Anyway, I should have video Thursday with any luck.  

[Daren Schwenke]

Print done.  Needs longer legs for the Lego bit.  I'm back in the realm of 'interference' fit instead of a bending interface so it goes on really hard.

So hard in fact that I snapped off part of the tape cover window doing it. I need to beef that up too.

The tape slot on the bed is also too narrow due to the first layer bleedover like I feared and is a PITA to clean up.

So... making some revisions.

[SMdude]

Hi Daren,

Re the cameras not working together, you need to hook each camera up to a separate usb host controller.

On my laptop, there was only one usb host controller that all 3 external usb's hooked up to. The other host controller was running the webcam in the screen. So I routed the second host controller to one of my other usb ports and now have a camera hooked to each host controller.

Looking forward to seeing how your feeder goes...

Cheers

[Daren Schwenke]

I'm afraid it's not that simple.  I'm using the same physical camera for both bottom and top vision.  

You can select the same camera for both in Machine Setup, but when you add the second one OpenPNP grinds to a halt and streams Java errors.  Adding it to one or the other works fine.  

I'm fairly sure this is happening because it is expecting to have exclusive access to the device and so when it shows up in use (by the other thread) it pukes.  I'm also fairly sure it would be an easy fix to allow one device to passively obtain the stream provided it can use exactly the same settings for both, so I'll probably end up having to tweak light levels via PWM instead of just using exposure and such to get them balanced.

There was a comment in a thread here about making the vision system more modular, which I'm obviously all for.

Just another thing on the 'to do' list.

[Jason von Nieda]

Hi Daren,

I need to solve this same problem for the CHMT36VA that I am working on, so if you aren't in a major hurry, just hang tight and this will fix itself. I expect it will be in the next 2-3 weeks.

Jason

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/03e3b25e-c69a-4480-a8a6-9a069f09247d%40googlegroups.com.

[Daren Schwenke]

That's good news Jason.  Thank you.

I'm not sure I can wait as the event I'm trying to spool this up for is in 4 weeks, but I have some other stuff to sort out first anyway.  

If nothing else, I may just hack and slash your code to get it done if I can figure it out.  Gut one of the drivers of it's initialization stuff, and see how to get the stream betwen threads.

I thought about perhaps trying to fake it as two devices in the vfl2 or Fuse realm.  I have no idea if v4l2 would allow me to create pseudo devices or not, things would probably get confusing (if it tries to use the kernel driver) and I would probably loose a ton of performance completely consuming a core to do it in Fuse.  But if it let me, the code would not be hard andthen I'm working in a space I know.

Working on it in your space would definitely be better, but I only touch Java when required to fix something, so my out of scope design patterns may not be pleasing to your sensibilities.

If you have a branch that you are working this in already or a tied issue on github that would give me a head start.

Oh, when I implemented proper stepwise resolution output in the native C, the Java side blew up.  I found VLC had some good code directly applying to this issue after I already had most of mine done, so that would probably be a better starting point than what I hacked together.

I'm assuming the reason for the crash was were too many available resolutions for some stack as with stepwise resolutions, you can pick anything from max to min, in increments of 2, for both X and Y independently (according to what the driver was reporting anyway)  I tried some mathematical trickery to instead present a list of common resolutions based on the reported max resolution, but In the end I copped out and just hard coded the common ones and called it a day.

I think perhaps if a stepwise only camera is detected, you should probably just present a text box or a list of common in-range resolutions to fill in, and just enforce the min/max from what the driver gives you. There is actually a use case for allowing fractional resolutions here though, as then the scaling and cropping is done on the hardware and nothing is wasted.  I'll be using that.

Oh, and list the fastest (least transforms/math needed) implemented modes first, and probably don't bother to list video formats which are not currently supported at all.

Well that got long.  I should probably make this an issue.

[Jason von Nieda]

Hi Daren,

Sorry, no progress on this yet. Still working on making the machine move first. I'll post here when I have something.

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/b240031b-3e37-4779-9bbb-86b065dc7584%40googlegroups.com.

[Daren Schwenke]

Well this really sucks.  The original design worked better.

• The tape slot in the neighboring feeder has to be very shallow at around 0.5mm to clear the bottom of the parts, but the tolerance on tape edge width can be as high as 0.2mm. That is too close for comfort for me.  The other feeder with the bending wheel support design took this in stride.
• The top between the peel slot and the part window being supported entirely by layer adhesion is inherently brittle and weak and even doubling the part thickness here, I still broke it after a few times of taking it on and off the Lego sheet.  The other one was printed flat, and if I did break it, was a 10 minuted print and not the entire base.
• Getting the tape lifting force just right is too finicky and didn't have the range to go from my thick tapes to the thin ones.  The spring area needs to be longer, but I don't have the room to make it so.  The other one with the wheel, this tuned itself.

As much as I wanted the lowered output window this design would give me, I think I'm going to call it and go back.

[Anthony Webb]

All part of the learning process! Kudos for taking a chance, your design will be more solid/vetted as a result of finding the weaknesses.  Keep movin!

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/90e36de7-6096-4c6b-8619-82bf3e9d7df5%40googlegroups.com.

[Daren Schwenke]

Mashup complete. Printing.

[Brynn Rogers]

Looking better Daren,

   More like what I need now...   Is there a step file or something that I can import into fusion360 so I can spin it around to view and maybe even modify?

Brynn

[Daren Schwenke]

I'll put the source and rendered files up later tonight here: https://github.com/Arcus-3d/P1-Buddha-tape-feeder

I will still have to iterate on the slot cover again though as I changed a lot of dimensions affecting it, and it's designed to be bent so that's hard to model right.

I have to have everything printed out to do that.  That will be a couple hours.  

Github actually lets you spin around the models right in the interface now, but they will be the individual parts.  

I suppose I could render it all as one too if I don't run out of memory.  :)

The render you see here is 9 separate prints.  Longest one is printing the base at about 2 hours though.