Working on an new Auto feeder

[sa...@wetronic.nl]

After an lot of struggle with different other automatic feeder designs.
And the fact that yamaha cl feeders where going to be an big unit with an lot of weight to my machine.

I've thought well why not step up and try to design my own feeders.
I liked the design of neoden 4 due the size, and the essemtec because of the easy way to operate that feeder.

Well it already cost me an lot of hours thinking and re thinking. Several bad attempts.
But I think i'm on to something at the moment.

It's an dual feeder 2x8mm or 2x12mm possible in same housing the dual feeder is 30mm wide in total. And only 110x75mm in dimensions.
The cover is pulled in an 45 degree angle with almost no spacing for the component to reduce any parts falling out of there pockets.
The cover tape is pulled by an small low cost N20 motor with 2 gears fitted.
The reel itself is pulled by an low cost stepper motor with an sprocket wheel. And driven by an timing belt from the stepper to the sprocket gear.
An photo interrupter is used to detect the positon of the stepper gear for calibration / start up process. It looks at the sprocket gear so no issues with the clear tape options.

The top of the unit is based on the essemtec design, it lifts up with an spring and also pushing down on the tape when closed.
It guides the tape and is flexible in the different tape thickness.
Only an small pocket is open for the parts to be picked up. To minimize the change for parts to fall out of their pockets.

*right lane has to be designed more mostly working on the left lane first.
right lane must be at the same pickup point as the left lane 

Step file: www.winius.net/auto_feeder.stp

[Jason von Nieda]

Looks interesting. I'm curious: why does it need to be dual lane? Are some parts shared?

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/98e70065-2ee1-4e26-bad3-a783703edc2e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Veenie]

Hello Jason,

The dual lane is due the fact the hardware wont fit in an 15mm housing.
The stepper motor for example is 24mm without the pulley.

But also some items will be shared like the controller (arduino) and some axles like the one used for the sprocket gear.
The tape is feeded in from the side like the essemtec feeders.

https://www.youtube.com/watch?v=2VhuwemMqcE&t=18s

So the design should be able to be picked up from the machine.
Not all details are 100% clear for me at this moment.
Like which kind of communication shall we use to talk to the arduino inside of the feeder.
Which functions do we need on the feeder itself?

At the moment I'm thinking for several functions like:
-End of tape detection (error warnings?)
-Homing the sprocket gear / perhaps calibration check every 10+ components?
-Feed button for manual control
-Indication LED's.

Op zondag 21 mei 2017 19:14:40 UTC+2 schreef Jason von Nieda:

[Jason von Nieda]

Sounds very good Veenie. I look forward to your progress!

Jason

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/fcfc507f-987a-4afa-aeef-4989d5a06ffa%40googlegroups.com.

[Veenie]

Well here is the first update already :).

Worked out the drive gear more, perhaps if somebody knows an gt2 idler pulley with about 30 teeth this will be an nice change to the design to give the stepper more torque.

Now it's only an 16/20 gearing an 16/30 would be better. Speed won't be the issue.

Hope the pictures make more clear of the design I'm having in my head.
It's hard to explain in 2D. The step file explains an lot more there are some clever tricks in the design making loading reels an easy job.
And also keeping the costs low but the tolerance pretty neat (I Hope).
Also you can see here why it's an double lane feeder the 2 steppers are mounted in an way they both keep within the 30mm width of the feeder.

The yellow part in the drawing is spring loaded (pulled down back on the feeder).
This trick will give me the ability to run 0.2mm - 2.3mm thick tapes.Most plastic tapes are about 0.2 - 0.3mm thick and jumping around in most feeders due to the 1.2 - 1.6mm tape thickness that most feeders are designed for.

The tape will be fed through the green part and held down by the yellow part.
The side walls will keep the yellow part in place not allowing it to rotate / flex anyway when in the down position. (missing on the detail drawing here below)

Op zondag 21 mei 2017 21:59:10 UTC+2 schreef Jason von Nieda:

[Veenie]

Well another update already.

Still in the drawing process.

This time the photo interrupters are placed in the design along with some other electronics like the arduino and the stepstick stepper drivers.
But most important I think the tape peelers are pretty much finished now. With spring loaded gears to get enough traction out of it.

Well images only show the 2d part.
Also an new step file is available so most of you can see the files in 3D and seperate parts.
It's an bit messy still cleaning up will be done after all technic aspecs are covered.
Also I think that using the DIN rail clamp units for mounting the feeders to the machine will do an nice trick in mounting the units to the machine.
These clamps can handle an pretty load of weight and are very stable once closed. Perhaps an support pin / position pin will be added to the design as well.
Going to use the chipsetter one design for the feeder mount and power supply / communication. Just an strip with 5 lanes (12V, 5V, RX, TX, GND) which can be mounted inside of the din rail 35mm.

https://de.aliexpress.com/item/35mm-DIN-Rail-fixed-clamp-fasten-clip/32799925198.html

3D step files: www.winius.net/auto_feeder_v8.step

[Veenie]

Bomlist for now:

*is there no edit button for google groups?

[Michael Anton]

There is no way to edit posts, unless you are an admin on the group...

[Veenie]

And another update ;).

This time an more render like image.
Did some work on the feeder frame.
Also worked out the last issues with the tape peeler modules.
Made an easy to mount system for the reels.
Now have to put some work in the front and left over items.

The system will be mounted with an DIN rail clamp.
Hook in the bottom first and then pull up to fix it to the rail.
Removing is the other way around. Pull up the feeder (2-3mm) angle it an bit backwards and pull back.

 

[Malte R.]

Hi.

Thanks for sharing your great work. Looks really promising!

Are you also located in Germany? Saw "Conrad" as supplier on your BOM ;-)

Anyways, I like many things about your design and would like to contribute a few comments - make suggestions and ask questions:

== Driving sprocket wheel ==

Instead of stepper, pulley and belt you could consider a small DC motor and a worm gear.

Given the encoder at the sprocket wheel it should be accurate enough.

I would envision that pretty much the way SIPlace 80 feeders work - see here:
https://youtu.be/ql0o4lEWL2A?t=7m12s

Might be cheaper, save some space and eliminates the need for a stepper driver. Simply H-bridge or even just a MOSFET is enough to drive such motor.

I recently bought this part for my own experiments:
http://www.ebay.de/itm/2pcs-Metal-Worm-Wheel-Plastic-Gear-Reducer-Reduction-Gearset-DIY-Accessories-/291686181590?hash=item43e9d9ead6:g:nMgAAOSwKrhVZD8C

It might work nicely with a N20 geared micro motor like this one:
http://www.robotdigg.com/product/914/N20-micro-geared-stepper

Both parts can be sourced cheaply when buying in bulk from China.

== Cover tape pulley ==

I assume you plan to route the cover tape along the sides of the feeder and through the pulley at the bottom?

How do you plan to control the timing here? I like the ingenious way that Siemens did that; add a spring loaded pulley that eliminates any slack in the cover tape and has a switch that is tripped whenever tension is insufficient.

Unfortunately I do not have a great idea yet on how this could be added to your current design... But I think it would be beneficial.

== Microswitches at top ==

What is the function of these?

== Contacts on DIN rail ==

Great idea to use DIN rail + clamps. Cheap and widely available.

For the contacts, I would use long PCB with connector pads, that fits the middle of the rail. If you get those with thicker copper layer in ENIG finish, they make great contact and are rather durable.

For contacting from the feeder side, I'd recommend "pogo" test pins. Those pins could be embedded in the bottom and make the necessary contact when the feeder is clamped in. The spring-loaded contact ensures connectivity.

Well reknown manufacturer "Feinmetall" happens to produce great quality pins in the next village from where I live:

http://www.feinmetall.de/produkt-finder/kontaktstifte/kontaktstifte/?kid=F697&filter1=2&filter2=&filter3=&filter4=1&tags=

They have a huge line and we can certainly find a solution.


Thanks again and keed up the great work!

Best regards
Malte

[Veenie]

Hello Malte,

Well not in germany :) but we're neighbours i'm from the Netherlands.
So yes an few shops are known shops for both of us :).

Most of your ideas are already on my design note at my desk.

The stepper motor is pretty cheap as well. With driver about 5 usd.
An n20 gear motor is about 3 usd and all the other parts will count an additional 3 usd.
So yes that will be an bit cheaper, but gives some other issues. We will need an lot of gears and making the design an lot more complex and more sensitive.
The belts are pretty flexibel in design and giving me much more reliable drive. Also can control the stepper way better then the n20 motor.
So thats why I went for the stepper motor with stepper driver.

The cover tape is pulled outside the case and fed in to the rotating gears with the n20 gear motor.
The gear motor will be driven by an current limiting driver *just an mosfet with an potentiometer.
So it will pull on the tape with an limit force and perhaps I'll take the arduino to control the motion.
An bit like the essemtec feeders, Just rotating randomly with an puls from time to time.
The second set of gears is spring loaded on the the first set of gears which are driven with the n20 gear motor.

The switches at top are optional, but I'll use them to detect the end of the tape so we can perhaps trigger an message or pause the process when an feeder runs out of stock.
Especially on large production batches i'm not planning in counting the parts on the reel. Perhaps I can save the reel info in the arduino but not sure this will be an good way to solve this.

The din rail idea came from the chipsetter one design:
https://static1.squarespace.com/static/54d42fb5e4b0ca1f1af0f1c1/t/581fd32b6a49638d9518ec03/1478623259225/fast+setup+gif

Also there is the PCB inside for the contacts.
The spring loaded contacts are widely available also at ebay/aliexpress not sure what price feinmetall will ask for theirs?
Those are very cheap from ebay: http://www.ebay.com/itm/100-Pcs-9100B-1mm-30-Degree-Spear-Tip-Dia-Spring-Test-Probes-Pins-S3M1-/172627017703?hash=item28315f3fe7:g:x8kAAOSwlY1ZGr8k

Op woensdag 24 mei 2017 19:17:23 UTC+2 schreef Malte R.:

[Veenie]

Well another update.

Bit late over here.3 o'clock in the night hmmz... well just work better at night hours then day.
Made the mounting bracket / system in the drawing.
Did some minor changes, can you find them :)? All there to improve the design.

Wonder if all items can be printed with an FDM 3d printer. I'm guessing that some items will have to be printed at shapeways or other services like that offering SLS printers.
Well just lets see what my own craftbot+ will do with these items.

 

[Veenie]

forgot one picture.. bit sleepy over here :)

5 pogo pins for 12V, 5V, GND, TX, RX.
And the din rail clamp design in view.

[Anthony Webb]

I am excited to see your progress.  I've been down this road several times before and the real trick here is going to be getting the physical mechanics of it just right.  Things make all the sense in the world until you actually print/build them, then the real work begins.   Lots of great ideas here, I'm rooting for you.

One point that we ran into was trying to make that tight radius as you pull the tap and bend it around to eject it.  Although we were totally in spec and is seemed like it should work, it was a big fail.  You'll probably find you start slipping belts as you try to make that radius.  But I'd certainly love to be proven wrong.

--

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+unsubscribe@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/cb6160da-4f89-4cda-b042-0ab6a006514d%40googlegroups.com.

[Michael Anton]

If that outside piece at the front is a guide for the tape, you will want another one behind it as well, so the tape rides in a channel.  If you are feeding plastic tape, it will bind up on your inside mechanism without the second guide.  Even with a second guide, it can still bind up.  Ask me how I know... :-)

[Veenie]

The radius of the tape it self is 30mm. This should be pretty much the same as what essemtec does with their feeders.


Op donderdag 25 mei 2017 03:36:04 UTC+2 schreef Anthony Webb:

I am excited to see your progress.  I've been down this road several times before and the real trick here is going to be getting the physical mechanics of it just right.  Things make all the sense in the world until you actually print/build them, then the real work begins.   Lots of great ideas here, I'm rooting for you.

One point that we ran into was trying to make that tight radius as you pull the tap and bend it around to eject it.  Although we were totally in spec and is seemed like it should work, it was a big fail.  You'll probably find you start slipping belts as you try to make that radius.  But I'd certainly love to be proven wrong.

 

@Michael Anton

If that outside piece at the front is a guide for the tape, you will want another one behind it as well, so the tape rides in a channel.  If you are feeding plastic tape, it will bind up on your inside mechanism without the second guide.  Even with a second guide, it can still bind up.  Ask me how I know... :-)

Yes that's correct that part is missing in the current design since I'm working on that region at the moment.
The Tape will be locked up between two plates and not only the outside plate seen now.
Also the sidecover will be closing the last part so the tape will only have one way to go and that is out! we don;t want the tape getting back in to our gears and ruining the drive of our feeder.

[Malte R.]

Hoi Veenie,

groetjes van de buren ;)

I have worked in the NL for almost a year but sadly my Dutch language skills are lacking...

The stepper motor is pretty cheap as well. With driver about 5 usd.
An n20 gear motor is about 3 usd and all the other parts will count an additional 3 usd.

However the stepper still requires pulleys + belt; probably the same cost as my gears? But I agree - the difference in cost does not seem significant, and stepper certainly gives you more exact control. How many steps per revolution / what stepping angle does your model feature? Have you calculated the "resolution" i.e. movement of tape per step? For 0402 reliably positioning the tape can be a tricky task.

Which also brings me to my next point: Just indexing by sprocket teeth (using opto) might not be sufficient for 2mm or 1mm component pitch. You could still advance by a certain number of steps and just use the sprocket wheel for indexing at the holes but it would no longer be closed loop.
 

The cover tape is pulled outside the case and fed in to the rotating gears with the n20 gear motor.
The gear motor will be driven by an current limiting driver *just an mosfet with an potentiometer.
So it will pull on the tape with an limit force and perhaps I'll take the arduino to control the motion.

I'd recommend the latter. You should have plenty of pins available. To stay on the cheap side, you could use a shunt resistor and ADC pin to measure motor current and drive your mosfet from another pin. Gives you more control options.

I like the idea of replacing mechanical sensor by current limiting. Now you mention it, I remember working with a electric drill recently, which applied the same technique to limit torque - rather then old-school adjustable clutch ;-)

 

The second set of gears is spring loaded on the the first set of gears which are driven with the n20 gear motor.

Makes sense. This design is widely used in Siemens feeders and - on a bit larger scale - by SmallSMT.

The switches at top are optional, but I'll use them to detect the end of the tape so we can perhaps trigger an message or pause the process when an feeder runs out of stock.

OK. Not my must-have feature but might be helpful.
 

The din rail idea came from the chipsetter one design:
https://static1.squarespace.com/static/54d42fb5e4b0ca1f1af0f1c1/t/581fd32b6a49638d9518ec03/1478623259225/fast+setup+gif

I remember seeing that but failed to realize it was a standard DIN rail + PCB :-)

 

Also there is the PCB inside for the contacts.
The spring loaded contacts are widely available also at ebay/aliexpress not sure what price feinmetall will ask for theirs?
Those are very cheap from ebay: http://www.ebay.com/itm/100-Pcs-9100B-1mm-30-Degree-Spear-Tip-Dia-Spring-Test-Probes-Pins-S3M1-/172627017703?hash=item28315f3fe7:g:x8kAAOSwlY1ZGr8k

High quality often comes with a bigger price-tag. Probably the China-ones will do.


With regards to the driver board do you want to use ready made Arduino or make your own board with a Atmel mcu?


One more thing I have not noticed in your design: You should add two or three tactile switches and maybe a few LEDs in a location which users can access during operation to interact with microcontroller. Might help both debugging and operation.




Are you planning to share the source files at some point? I think a lot of people might be interested to try your design.

So far it looks like most parts could be printed, laser cut or milled - probably your intention :-D


If you could use a hand with design or testing, let me know.

Bye voor nu :-)

Malte

[Veenie]

Hello Malte,

Well I can read german most of the times but speaking or writing german is an whole other story:).

The stepper motor is an 0.9 degree version with micro steppign we should get an decent amount of resolution.
Due to my calculations with an 1/8 micro stepper setting we would get an resolution of 0.0337mm per step. For 0402 we need 2mm.
2 / 0,0337 = 53,07 steps. Only question is will the stepper be strong enough for micro stepping and if yes can it handle 1/16?

Also 0402 and smaller will need vision anyway for the last final correction after pickup.

The sprocket gear sensor is mostly for the starting point of the feeder (start up calibration)
and can be used to check up from time to time after each X component say like 10 or so?

The user interface part with buttons and stuff has to be designed didn't focused on that part yet due the fact it's not the most important part of the design.
I have only 2 buttons in mind (1 left lane, 1 right lane) but perhaps you can tell me why you think we should need more buttons then just those two?
Which function do we want to control from the buttons. As far as I know we only need to push the button to move forward on the feeder. Keep it pushed for longer feeding and short push for one component at the time.
Also one green led for the power (perhaps 2 becuase of the 12 and 5V) and 2 red leds for any error or other messages. If you like other colors well be my guest :) just replace the LED's in the design.

In the messages before I've already placed some step files. From time to time I'll send some new step files of the latest design.
Licensed under the do what ever you like with them but no commercial activity's like selling the design or feeders build by this design.

The production is pretty simple of these, an couple of items will probably need an professional 3D printing service like shapeways due the floating parts and the smoothness.
But most components can be made with an decent 3D FDM printer. The frame it self is just an plain simple 2mm sheet of metal can be handmade or lasercut.

In the case of this feeder my plans are to let the feederframe be cut out of metal and the plastic items printed by shapeways. Just because printing all these items with my own 3D printer will take weeks if I want to build 30pcs of them.
For the prototyping I'll use probably some low cost plastic for the frame and most parts 3D printed my self (kinda worst case scenary if you want to build this feeders completly home made)
Unfortanly some items will need an lathe for construction, it are only an few axles nothing very special. An small lathe will do the trick. Or you could use brass and solder the items together.

Op donderdag 25 mei 2017 13:46:50 UTC+2 schreef Malte R.:

[Michael Anton]

If your frame is just a flat plate, you might want to consider making it out of a PCB.  These are pretty cheap to have fabricated out of China now, and you would have a hard time getting anything else made as cheaply.  You need a PCB anyhow for the electronics so using at as the frame too would just be a bonus.  I'm looking at doing this on my feeder design.  You can even get aluminum PCBs made now, though they are more expensive than FR4, but still might be cheaper than a machined piece of aluminum.

[Veenie]

Hello Michael,

Well an metal plate laser cut at 30pcs is about 3.30 usd per piece it's not that expensive as long as you take an greater amount.
Prototype is always the most expensive version.

But I like your idea it would solve some hard bits in the design due to wires.
Do you know an address where we can get the pcb costs down to around 4 / 5 usd total at 30pcs?
The seeedstudio I use as reference is about 8 usd.

Op donderdag 25 mei 2017 22:54:43 UTC+2 schreef Michael Anton:

[Michael Anton]

Try http://www.smart-prototyping.com/, or https://www.pcbway.com/.  Both of these companies do an excellent job, though shipping might be more expensive, but they will have boards to your door in about 2 weeks.  I don't know if either of these will be in your price range, since I don't know how big your feeder is, but it is easy enough to do an online quote to find out.

[Michael Anton]

If you can live with stainless steel for your plate, you might try http://lasergist.com/ for your prototypes.  They have odd requirements for hole sizes though, so I'm not sure what you would get if you used small holes.

On Thursday, May 25, 2017 at 3:08:14 PM UTC-6, Veenie wrote:

[Mike Harrison]

>> Do you know an address where we can get the pcb costs down to around 4 / 5
>> usd total at 30pcs?

This site can search PCB quotes from multiple suppliers.
http://pcbshopper.com/

[Veenie]

Thanks for the pcb suppliers.

Unfortanly I think the plan is not going to work or we need 4 layer of pcb.
The 12mm tape is rubbing along the side of the pcb.
On long term this will mean that the pcb will wear out during the tapes cutting in to it.
This is something I dont want.

In the netherlands I already found some good and affordable metal laser cuttign companies.
Even stainless steel comes at about 4.5 usd an piece when bought at 30pcs.
Aluminium is about 3.7 usd and normal steel 3.5.

Well steel needs coating so I'm not sure if I go for the aluminium or stainless steel option.
Something to think about. I think the stainless will bring much more strength to the body.
The other side of the story is that 2mm aluminium isn't that weak as well. 

[Malte R.]

247Tailorsteel is a Dutch company; I order from them here in Germany all the time and an happy with both quality and prices. Checked them out?

If you deburr the part edges yourself you can usually save a lot of money.

2mm aluminium might flex too much. As weight is probably not an issue, I would go for stainless steel.

[Veenie]

Hello Malte,

Didn;t had an account at their site, just created one and received my login information this morning.

So have to check their prices to see if they can do it for even less.

Also made some new progress on the design of the auto feeder and most of it is almost complete / done.
Just an few minor things left (yes the buttons and leds :P) and some mounting points for the side panel.
Have changed the mounting of the din rail bracket due the fact it was mounted only to the plastic and didn't liked that,
Now it's mounted to the metal frame it self with an old trick (step files are attached so perhaps somebody finds out what I;ve done to solve that issue :).

Well most information / detail is in the step file it self. it went from 1mb to 10mb already.
So an lot of work has done only have to solve the final steps of the puzzle to finish and start ordering prototype parts.

www.winius.net/auto_feeder_v34.step

Op vrijdag 26 mei 2017 11:09:07 UTC+2 schreef Malte R.:

[Michael Anton]

One thing I would suggest, is on your inside tape guide, that you make it tangent to the sprocket.  I have seen plastic tape continue to wind around the sprocket, and it may jamb with your current configuration.  I certainly have seen this happen even on feeders that try to peel the tape off the sprocket sooner.

[Arie Lashansky]

Hi Veenie

I like your design a lot as it is uses a proven simple method.I got my first view of a neoden 4 at a Trade Fare last week.and the small compact feeders are a win.
The best part of your design for me is the DIN rail mounting.It allows to do a setup not connected to the tool and swap setups without breaking the setup.

Lets say I have 5 din rails that can be mounted and changed out I do not need to move reels off the din rails between production runs Just swap the DIN rails.
That will save lots of time between Jobs. It also allows me to have a setup of five times the feeders.

Please update the progress
Thanks Arie

[Veenie]

@Michael Anton,

Changed some parts to even get better care of that issue.
tried with the 3d printed parts several issues and couldn't find any issues with hand driven.

@Arie Lashansky,

Thanks that is the way I normally like to go in my designs.
And not sure if we live in the same country but the neoden was displayed here in the netherlands as well.
Not sure what to think of it, I like most of it but the price is still pretty high with about 12.000 for an complete unit with only chinese support.

The din rail can be used in several ways, just do what ever you like with them :)
I was an few days out for other business so that's why the last update was an bit ago.
But the first prototype was printed to see if all dimensions suit op.
still think the best way to go with this is to let i be printed by shapeways or another manufactory that can print SLS parts.
FDM is not up to the job to get the surfaces smooth enough and the overhang without any issues.
Perhaps an water soluable support material will help out but my 3d printer only has one head.

Well here are some pictures of the prototype print from my own craftbot+

[Arie Lashansky]

Hi Veenie

Do you have any progress on your design

thanks
Arie

[Veenie]

Hello Arie,

The frame part is placed on order expected any day here.
Unfortanly all my time is consumed by an rebuild of my shop.
Just got the keys of an new part to extend my business.

Op dinsdag 20 juni 2017 14:59:52 UTC+2 schreef Arie Lashansky:

[Malte R.]

Hey Veenie,

any news on this project?

Don't want to annoy you - just curious about the progress :-)

Dank je wel!

Malte