Mike's Feeder Development

[Michael Anton]

So, I have started developing a fully automatic feeder, based on some of the various small motors that I have found.  I'm trying to keep the cost of these as low as possible, as long as the end goals I have can be met.  I certainly don't want to see $100 feeders, even if these would be considered inexpensive in the industry, but I would like them to be reliable.  What I like about this concept, is that it could be dropped onto any machine, including the ECM 93 I use all of the time that doesn't have enough feeders.

I thought I would try to come up with something similar to what is used on the Quad machines, that uses the pick of a part to trigger feeding the next part.  This way each individual feeder is self contained.  Likely I'll need to develop a small inexpensive controller board for each feeder, and there will need to be a couple of buttons to set the advance distance after a pick.  I might plan on designing in an RS485 link to link all feeders together in a multidrop fashion, so that they could be triggered at will, and to pass back status information, so that the user could be notified if a feeder jammed.

I'm hoping that I can use an idler wheel running on top of the tape to drive the cover tape pickup spool, like I've seen in another design.

For the first design, I am attempting to use a small DC gearhead motor that is mounted inside the hub of the sprocket, which unfortunately makes the sprocket quite large.  This might be a problem, so I'm looking at a gear or belt drive that would allow the motor to be mounted outside the hub.  This also would open the door to using some of the tiny stepper motors that I have found.  I plan to use a small opto sensor to detect when the sprocket has advanced to the next tooth, so at least there will be some closed loop control, so that hopefully it will be possible to account for slippage, or missed steps.

As a result of this, I have a few questions for the group:

1) What is the maximum height a feeder could be above the 20mm extrusions that they are mounted on?

2) How far outside of the edge of the frame at the standard feeder mounting locations can the head reach?

3) Is #2 dependant on height?

I would like to have the feeder banks mount using Neil's feeder bases if possible.  This might be difficult though, depending on the results of the questions above, but I'll try my best to come up with something.

I managed to get a sprocket printed out of PLA that looks like it will be strong enough to drive the tape well.  I didn't have much luck with the conductive ABS I have, as it is very brittle, and not stiff enough.  I'm not using the same ABS as Neil however, so it might be that other suppliers have better options, and I would be curious to hear back from others regarding this.

It would be great to hear suggestions and thoughts from the other minds in this group.  

Mike

[Neil Jansen]

On Sun, Nov 2, 2014 at 7:27 PM, Michael Anton <3d.m...@gmail.com> wrote:

So, I have started developing a fully automatic feeder, based on some of the various small motors that I have found.  I'm trying to keep the cost of these as low as possible, as long as the end goals I have can be met.  I certainly don't want to see $100 feeders, even if these would be considered inexpensive in the industry, but I would like them to be reliable.  What I like about this concept, is that it could be dropped onto any machine, including the ECM 93 I use all of the time that doesn't have enough feeders.

Mike, that's great to hear, can't wait to see some pics.  Reliable SMT component feeders are easily one of the toughest problems faced in the construction or retrofit of a PnP machine.  Re: Cost.  For those not as familiar with feeder cost, here's a quick comparison: A Manncorp feeder for an FVX is ~$500, a Panasonic or Juki would be several thousand or more each.  My current drag-strip feeders are about $5 in parts cost, and the full-auto feeder I'm using now is probably around $10-15 each in parts cost.  Anyway, having cheap feeders is an absolute necessity for a cheap prototype machine like what I'm trying to build.  However, we've sort of done that already with the $5 drag-strip feeders.  IMO people that want to use a machine in a business setting probably don't care about cost too much.  So if you're doing full-auto feeders that are designed to be reliable, I say make them out of whatever is needed to make them reliable.  People that can afford them will buy them, and everyone else can use $5 drag strip feeders, or a cheap full-auto feeder like what I have now, even as limiting as they are.  I'd say parts cost in the $50 range, that could be built from common tools like a 3D printer and maybe a laser cutter or waterjet, would be absolutely great.  Anyway, I can't wait to see what you come up with.

To answer your questions:

 

1) What is the maximum height a feeder could be above the 20mm extrusions that they are mounted on?

I've thought about this a lot myself, because to get any sort of a full-auto or semi-auto feeder onto the current design would require a lot of room for the sprocket, which I would assume would be around 40mm best-case.  So basically as-is, it won't work, but you can easily modify the frame dimensions to make it work.  Make the machine a bit taller, but keep the bed the same distance to the top plate, which would give you more room to mount the full-auto feeders.   I've no idea if you're looking for a tabletop machine or something a bit more self-contained, but you could also make the vertical columns 1200mm tall or so, put some wheels on it, maybe attach a few FPD's together in a line format, and put the feeders down both sides, where they could tuck underneath the bed area a bit.  That's what I planned on doing when I had the time and resources to offer a business-grade machine.

2) How far outside of the edge of the frame at the standard feeder mounting locations can the head reach?

The end effector can reach a circular pattern from the center 0,0 of the machine.  So it can reach well outside the boundaries for the middle feeders, but getting to the corner feeders are considerably harder.  Right now the current pick spots for the 3x modular 60mm feeders is about the limit of the current design.  In other words, the pick spot should be somewhere around the center of the extrusions, and shouldn't be more than about 80mm from the center of the extrusion towards the vertical corner pieces.

 

3) Is #2 dependant on height?

Yes.  You can see this in the delta simulations that I posted to the Hackaday projects page.  However with the current setup, I've got the feeders mounted to where it shouldn't matter.  There are a lot of factors at play in getting a goeometry to meet all your requirements, and the ones I chose may not meet all of yours.  It's not trivial to come up with a new set of dimensions for everything, but it can be done. I've wrote up a bit on the hackaday page.

 

I would like to have the feeder banks mount using Neil's feeder bases if possible.  This might be difficult though, depending on the results of the questions above, but I'll try my best to come up with something.

Feel free to use them as-is, or modify them.  Or come up with something better.  I think they work in the case of a small prototyping machine but they're limited in that they're 60mm wide and really only work with my current feeders that are ~15mm high.  I think a re-design would be needed to do what it sounds like you want to do. 

 

I managed to get a sprocket printed out of PLA that looks like it will be strong enough to drive the tape well.  I didn't have much luck with the conductive ABS I have, as it is very brittle, and not stiff enough.  I'm not using the same ABS as Neil however, so it might be that other suppliers have better options, and I would be curious to hear back from others regarding this.

The cheap stuff that's more common is made from carbon dust, and is brittle like you say.  The stuff from PushPlastic / 3dxTech is carbon nanotube which should in theory make it even a bit more stronger.  I'd definitely try the 3dxtech stuff.  As for printing the sprockets, it can certainly be done, I would recommend the smallest nozzle that doesn't clog up, like 0.35mm or 0.4mm worst case.  I don't know that 0.5mm would print well, which is what I use most of the time.

 

It would be great to hear suggestions and thoughts from the other minds in this group. 

As I said, I think having different types of feeders supported is great, and I'll do whatever I can do to help.  My only real recommendation would be in how to interface them to FirePick Delta and/or OpenPnP.  There are obviously lots of different ways to do this, and none are really right or wrong, but I will list a few things to consider.   I'm going to assume you're going to use OpenPnP to drive whatever system you come up with.  Jason has designed OpenPnP to support custom Java modules that can be added onto the system, to support adding cool new stuff to the machine.  So you would have something like a MichaelAntonTapeFeeder.java, and you would implement the functions like getPickLocation(), isEnabled(), feed(), getPart().....  Here's the javadoc for the Feeder SPI and the reference tape feeder:

http://openpnp.org/doc/javadoc/org/openpnp/spi/Feeder.html
http://openpnp.org/doc/javadoc/org/openpnp/machine/reference/feeder/ReferenceTapeFeeder.html

The great thing here is that you can put whatever code you need in your java class to get the job done.  I don't really think there's anything holding you back from using asynchonous serial, network sockets, or sending the commands to the motion controller.  If you did serial, you could buy a cheap USB-to-RS-485 converter, and that would make a nice clean system that would work on any machine out there, and would support any number of feeders.  You would be free to use an existing protocol, make up a protocol, or hopefully collaborate with OpenPnP and FirePick on something that would hopefully be more futureproof and standardized down the road. 

[Michael Anton]

On Sunday, November 2, 2014 6:19:34 PM UTC-7, Neil Jansen wrote:

On Sun, Nov 2, 2014 at 7:27 PM, Michael Anton <3d.m...@gmail.com> wrote:

So, I have started developing a fully automatic feeder, based on some of the various small motors that I have found.  I'm trying to keep the cost of these as low as possible, as long as the end goals I have can be met.  I certainly don't want to see $100 feeders, even if these would be considered inexpensive in the industry, but I would like them to be reliable.  What I like about this concept, is that it could be dropped onto any machine, including the ECM 93 I use all of the time that doesn't have enough feeders.

Mike, that's great to hear, can't wait to see some pics.  Reliable SMT component feeders are easily one of the toughest problems faced in the construction or retrofit of a PnP machine.  Re: Cost.  For those not as familiar with feeder cost, here's a quick comparison: A Manncorp feeder for an FVX is ~$500, a Panasonic or Juki would be several thousand or more each.  My current drag-strip feeders are about $5 in parts cost, and the full-auto feeder I'm using now is probably around $10-15 each in parts cost.  Anyway, having cheap feeders is an absolute necessity for a cheap prototype machine like what I'm trying to build.  However, we've sort of done that already with the $5 drag-strip feeders.  IMO people that want to use a machine in a business setting probably don't care about cost too much.  So if you're doing full-auto feeders that are designed to be reliable, I say make them out of whatever is needed to make them reliable.  People that can afford them will buy them, and everyone else can use $5 drag strip feeders, or a cheap full-auto feeder like what I have now, even as limiting as they are.  I'd say parts cost in the $50 range, that could be built from common tools like a 3D printer and maybe a laser cutter or waterjet, would be absolutely great.  Anyway, I can't wait to see what you come up with.

I'm trying to do as much as possible with 3D printing, but given some of the limitations with ESD rated plastic, I'm thinking that making say the sprocket out of laser or waterjet cut aluminum might be the way to go in the long run.  So, I haven't ruled other options out at this point, but it was nice to find that I could 3D print a workable sprocket, which I wasn't sure was possible.

I have a fully parametric sprocket model now, which might be handy for others.  One just enters the number of teeth, the tape width, and a few other tooth and tape related parameters, and it updates the part.  I did it as a snap fit design, so there is no hardware required to fasten the two pieces together.  I plan to do the same for the cover tape pickup spool as well.  Tonight I've been working on implementing all of the parameters in a spreadsheet, so that it can drive the entire feeder design, and the spreadsheet can be common to all of the parts in the assembly.  It is not as clean an implementation as I'd hoped it might be, but it is still better than changing a bunch of parameters in multiple files, that all need to be set the same.

 

To answer your questions:

 

1) What is the maximum height a feeder could be above the 20mm extrusions that they are mounted on?

I've thought about this a lot myself, because to get any sort of a full-auto or semi-auto feeder onto the current design would require a lot of room for the sprocket, which I would assume would be around 40mm best-case.  So basically as-is, it won't work, but you can easily modify the frame dimensions to make it work.  Make the machine a bit taller, but keep the bed the same distance to the top plate, which would give you more room to mount the full-auto feeders.   I've no idea if you're looking for a tabletop machine or something a bit more self-contained, but you could also make the vertical columns 1200mm tall or so, put some wheels on it, maybe attach a few FPD's together in a line format, and put the feeders down both sides, where they could tuck underneath the bed area a bit.  That's what I planned on doing when I had the time and resources to offer a business-grade machine.

IPC docs seem to indicate that a minimum tape bend radius is 25mm, and that is for 8mm tape, so a 40mm sprocket, is still outside of this, but I'm sure it would still work.  One wouldn't have to wrap the tape around the sprocket either, but it probably helps to have more teeth engaged if possible.

I never thought about lowering the the bottom extrusions to make room for the feeders.  I guess then the baseplate would be too low, so the board holder would need to be mounted on taller standoffs.  I always planned on making my vertical extrusions longer, with the intention of making it easier to enclose the electronics.  Maybe it is worthwhile starting out with extrusions that are around a meter long, and then to shorten them later once all of the details are worked out.

I've been waiting on building a real machine for some of the development to settle down, but I figured that working on some fancier feeders would be a safe thing to start with, since it doesn't really depend on anyone else.

 

2) How far outside of the edge of the frame at the standard feeder mounting locations can the head reach?

The end effector can reach a circular pattern from the center 0,0 of the machine.  So it can reach well outside the boundaries for the middle feeders, but getting to the corner feeders are considerably harder.  Right now the current pick spots for the 3x modular 60mm feeders is about the limit of the current design.  In other words, the pick spot should be somewhere around the center of the extrusions, and shouldn't be more than about 80mm from the center of the extrusion towards the vertical corner pieces.

OK, thanks, I'll try to keep that in mind.  I would like to try to fit in with what you have already accomplished if possible.  I suppose I could put the sprocket outside the edge of the frame, and pick in the location that is required, but that means that I have to push further on the tape, and keep it supported.  It might be possible, but it is probably less than ideal, but I think maybe I've seen other feeders like this.

 

3) Is #2 dependant on height?

Yes.  You can see this in the delta simulations that I posted to the Hackaday projects page.  However with the current setup, I've got the feeders mounted to where it shouldn't matter.  There are a lot of factors at play in getting a goeometry to meet all your requirements, and the ones I chose may not meet all of yours.  It's not trivial to come up with a new set of dimensions for everything, but it can be done. I've wrote up a bit on the hackaday page.

 

I would like to have the feeder banks mount using Neil's feeder bases if possible.  This might be difficult though, depending on the results of the questions above, but I'll try my best to come up with something.

Feel free to use them as-is, or modify them.  Or come up with something better.  I think they work in the case of a small prototyping machine but they're limited in that they're 60mm wide and really only work with my current feeders that are ~15mm high.  I think a re-design would be needed to do what it sounds like you want to do. 

I had hoped to keep 8mm feeders to a 10mm width, but that is not looking trivial, so I'm planning to compromise at 11.5mm, which allows for 5 feeders in a 60mm block.  We can never have too many feeders after all, so the more compact they can be made the better.  The 8mm feeders will be the most difficult of course, and larger feeders should be almost trivial, as the motors can be larger etc.

 

I managed to get a sprocket printed out of PLA that looks like it will be strong enough to drive the tape well.  I didn't have much luck with the conductive ABS I have, as it is very brittle, and not stiff enough.  I'm not using the same ABS as Neil however, so it might be that other suppliers have better options, and I would be curious to hear back from others regarding this.

The cheap stuff that's more common is made from carbon dust, and is brittle like you say.  The stuff from PushPlastic / 3dxTech is carbon nanotube which should in theory make it even a bit more stronger.  I'd definitely try the 3dxtech stuff.  As for printing the sprockets, it can certainly be done, I would recommend the smallest nozzle that doesn't clog up, like 0.35mm or 0.4mm worst case.  I don't know that 0.5mm would print well, which is what I use most of the time.

I initially figured that it probably didn't really matter who made it, so I found the cheapest stuff I could, which likely came from Reprap Walmart.  For some things it works well, like the dump cup for the ECM machine that I printed the other day.

 

 

It would be great to hear suggestions and thoughts from the other minds in this group. 

As I said, I think having different types of feeders supported is great, and I'll do whatever I can do to help.  My only real recommendation would be in how to interface them to FirePick Delta and/or OpenPnP.  There are obviously lots of different ways to do this, and none are really right or wrong, but I will list a few things to consider.   I'm going to assume you're going to use OpenPnP to drive whatever system you come up with.  Jason has designed OpenPnP to support custom Java modules that can be added onto the system, to support adding cool new stuff to the machine.  So you would have something like a MichaelAntonTapeFeeder.java, and you would implement the functions like getPickLocation(), isEnabled(), feed(), getPart().....  Here's the javadoc for the Feeder SPI and the reference tape feeder:

http://openpnp.org/doc/javadoc/org/openpnp/spi/Feeder.html
http://openpnp.org/doc/javadoc/org/openpnp/machine/reference/feeder/ReferenceTapeFeeder.html

The great thing here is that you can put whatever code you need in your java class to get the job done.  I don't really think there's anything holding you back from using asynchonous serial, network sockets, or sending the commands to the motion controller.  If you did serial, you could buy a cheap USB-to-RS-485 converter, and that would make a nice clean system that would work on any machine out there, and would support any number of feeders.  You would be free to use an existing protocol, make up a protocol, or hopefully collaborate with OpenPnP and FirePick on something that would hopefully be more futureproof and standardized down the road. 

I'm not really up on modern PC software development anymore.  My development experience goes back to the days of DOS.  I'm hoping that if I can come up with the mechanics and electronics, that someone else will be interested in implementing the software side, if it is even required.  Like I said earlier, a completely automatic feeder could likely be dropped on any machine, without any configuration, other than where to pick the part up.  For example, on an ECM machine it could be treated as either a tape, or linear feeder, with no interface the the machine itself.

Mike

[Neil Jansen]

IPC docs seem to indicate that a minimum tape bend radius is 25mm, and that is for 8mm tape, so a 40mm sprocket, is still outside of this, but I'm sure it would still work.  One wouldn't have to wrap the tape around the sprocket either, but it probably helps to have more teeth engaged if possible.

One thing I've been slacking on is coming up with a master list of all the IPC specs and other documents that we should be looking at when coming up with a machine that's supposed to be taken seriously.  I've had access to these in the past at other jobs, but I working in the engineering dept. these days and aren't allowed anywhere near the SMT machines at my current job.  Could you post a list of any important documents that we should be looking at?  I don't mind buying them if they're not free.

I've been waiting on building a real machine for some of the development to settle down, but I figured that working on some fancier feeders would be a safe thing to start with, since it doesn't really depend on anyone else.

That's not a bad way to go really.  Not all of the jobs we have open are so modularized as an SMT feeder, and you've already got another working machine.  Sounds great to me.  I'm very grateful for any feeder work you're doing that could have future use-cases for FPD.

 

I had hoped to keep 8mm feeders to a 10mm width, but that is not looking trivial, so I'm planning to compromise at 11.5mm, which allows for 5 feeders in a 60mm block.  We can never have too many feeders after all, so the more compact they can be made the better.  The 8mm feeders will be the most difficult of course, and larger feeders should be almost trivial, as the motors can be larger etc.

If you can keep them to even 15mm that's quite a feat.  My current full-auto feeder is about 60mm wide for 1 piece of tape... but it's cheap and works well.  I know blade feeders are popular (if not ubiquitous) on real machines, however there's quite a bit of engineering in them to make them that thin.  What I did on my design was to sacrifice the density for something that could be with parts that were easily attainable.  I think making something like what you're doing is a complete job in itself and can't really be an afterthought, so it's great to have someone like yourself working towards that goal.

I initially figured that it probably didn't really matter who made it, so I found the cheapest stuff I could, which likely came from Reprap Walmart.  For some things it works well, like the dump cup for the ECM machine that I printed the other day.

I bought a roll before I knew better but haven't printed with it yet.  I'll probably keep that advice in mind and only use it for really simple stuff without breakable features.

I'm not really up on modern PC software development anymore.  My development experience goes back to the days of DOS.  I'm hoping that if I can come up with the mechanics and electronics, that someone else will be interested in implementing the software side, if it is even required.  Like I said earlier, a completely automatic feeder could likely be dropped on any machine, without any configuration, other than where to pick the part up.  For example, on an ECM machine it could be treated as either a tape, or linear feeder, with no interface the the machine itself.

If you get the feeder design working, you probably won't have to worry about writing the software.  If you keep the protocol simple enough, it should be a slam dunk to get that driver written... its a couple of hundred lines at most.  However I would definitely keep in mind how it would interface into OpenPnP because that's your interface, the closer you stick to it when implementing your designs, the easier it will be to integrate.   If I could try to condense down all the software progress that's happened in the last 20 years (end of the DOS days), it's pretty much gone towards modularity through abstraction layers and interfaces.  Other than that, it's just nerdy details.  If you understand the defined interfaces and develop your product around that, you'll be in great shape. 

Oh, the other thing I wanted to bring up, have you considered using a laser mounted to the end effector, and some sort of phototransistor mounted in each feeder, below the sprockets?  The idea being that if you were using a geared motor to drive the sprocket, and it was geared down enough, you could basically turn it on and off, while the sprocket tape would interrupt the phototransistor, it would basically be on while blocked, and would stop when it started conducting again.  The benefit here is that it's pretty cheap, and the feeders stay dumb, and it's actually pretty reliable.  Richard Spelling uses these on his custom feeders, and it seemed pretty clever.

[Michael Anton]

On Sunday, November 2, 2014 7:59:49 PM UTC-7, Neil Jansen wrote:

IPC docs seem to indicate that a minimum tape bend radius is 25mm, and that is for 8mm tape, so a 40mm sprocket, is still outside of this, but I'm sure it would still work.  One wouldn't have to wrap the tape around the sprocket either, but it probably helps to have more teeth engaged if possible.

One thing I've been slacking on is coming up with a master list of all the IPC specs and other documents that we should be looking at when coming up with a machine that's supposed to be taken seriously.  I've had access to these in the past at other jobs, but I working in the engineering dept. these days and aren't allowed anywhere near the SMT machines at my current job.  Could you post a list of any important documents that we should be looking at?  I don't mind buying them if they're not free.

I don't know that I'm the best person to ask, since I didn't even know that such specs existed until recently when I went searching for taping standards, thinking that a feeder design should probably conform to something that I didn't make up.  The one I've been using (that I found online for free, but probably shouldn't be) is EIA-481-C.  It's probably a bit out of date, but I figured it was close enough.  Sorry, I thought it was an IPC spec, as I have a bunch of those as well, but they usually pertain to footprint design.

I've been waiting on building a real machine for some of the development to settle down, but I figured that working on some fancier feeders would be a safe thing to start with, since it doesn't really depend on anyone else.

That's not a bad way to go really.  Not all of the jobs we have open are so modularized as an SMT feeder, and you've already got another working machine.  Sounds great to me.  I'm very grateful for any feeder work you're doing that could have future use-cases for FPD.

 

I had hoped to keep 8mm feeders to a 10mm width, but that is not looking trivial, so I'm planning to compromise at 11.5mm, which allows for 5 feeders in a 60mm block.  We can never have too many feeders after all, so the more compact they can be made the better.  The 8mm feeders will be the most difficult of course, and larger feeders should be almost trivial, as the motors can be larger etc.

If you can keep them to even 15mm that's quite a feat.  My current full-auto feeder is about 60mm wide for 1 piece of tape... but it's cheap and works well.  I know blade feeders are popular (if not ubiquitous) on real machines, however there's quite a bit of engineering in them to make them that thin.  What I did on my design was to sacrifice the density for something that could be with parts that were easily attainable.  I think making something like what you're doing is a complete job in itself and can't really be an afterthought, so it's great to have someone like yourself working towards that goal.

The key to getting these small, is to either find really tiny motors, like I've been attempting, or to gang them together with a slip clutch as someone else proposed.  I've found a bunch of really inexpensive small motors, so now it's just a question of whether or not they have enough torque (and I can't find any info on what the requirements might be), or can be geared easily.  I've ordered a bunch of inexpensive gears as well, so that I can play with this option if required.

I initially figured that it probably didn't really matter who made it, so I found the cheapest stuff I could, which likely came from Reprap Walmart.  For some things it works well, like the dump cup for the ECM machine that I printed the other day.

I bought a roll before I knew better but haven't printed with it yet.  I'll probably keep that advice in mind and only use it for really simple stuff without breakable features.

I'm not really up on modern PC software development anymore.  My development experience goes back to the days of DOS.  I'm hoping that if I can come up with the mechanics and electronics, that someone else will be interested in implementing the software side, if it is even required.  Like I said earlier, a completely automatic feeder could likely be dropped on any machine, without any configuration, other than where to pick the part up.  For example, on an ECM machine it could be treated as either a tape, or linear feeder, with no interface the the machine itself.

If you get the feeder design working, you probably won't have to worry about writing the software.  If you keep the protocol simple enough, it should be a slam dunk to get that driver written... its a couple of hundred lines at most.  However I would definitely keep in mind how it would interface into OpenPnP because that's your interface, the closer you stick to it when implementing your designs, the easier it will be to integrate.   If I could try to condense down all the software progress that's happened in the last 20 years (end of the DOS days), it's pretty much gone towards modularity through abstraction layers and interfaces.  Other than that, it's just nerdy details.  If you understand the defined interfaces and develop your product around that, you'll be in great shape. 

The only product I've used in recent years to do any PC development with is called Flowstone, by DSP Robotics.  It is like a really inexpensive version of LabView.  It works pretty well for doing PC control applications, and is very rapid to develop in.  I've been amazed at what I could throw together in a couple of weeks of time, and the learning curve wasn't too bad either.  I plan to use it in the future for building bed of nails testing jigs for production boards.  I think it would work really well for this sort of thing.  I think it has the ability to do all sorts of vision and audio processing as well, but I've never played with it.

Oh, the other thing I wanted to bring up, have you considered using a laser mounted to the end effector, and some sort of phototransistor mounted in each feeder, below the sprockets?  The idea being that if you were using a geared motor to drive the sprocket, and it was geared down enough, you could basically turn it on and off, while the sprocket tape would interrupt the phototransistor, it would basically be on while blocked, and would stop when it started conducting again.  The benefit here is that it's pretty cheap, and the feeders stay dumb, and it's actually pretty reliable.  Richard Spelling uses these on his custom feeders, and it seemed pretty clever.

I think I saw that one somewhere, but I thought he used the laser to trigger the advance cycle.  I didn't know he was shining it through the tape holes.  I would think that would be very unreliable if you had parts on clear plastic tape, which is pretty common.

[Paul Jones]

I generally use a 0.4mm nozzle on my ultimaker with PLA and it works great. I’ve got a draw full of helical gears I printed when I was trying to design a simple rotating head for a previous PnP machine, and they mesh surprisingly well! I even managed to get a small worm drive working, so printing gears is definitely doable.

I recently acquired a 0.2mm nozzle and it seems to work well so far. I used it to print the tiny 5mm lock pieces that fit into the aluminium extrusion (the stuff I bought was quite narrow so I had to redesign that part).

 

Cheers,

Paul.

--
You received this message because you are subscribed to the Google Groups "FirePick" group.
To unsubscribe from this group and stop receiving emails from it, send an email to firepick+u...@googlegroups.com.
To post to this group, send email to fire...@googlegroups.com.
Visit this group at http://groups.google.com/group/firepick.
For more options, visit https://groups.google.com/d/optout.

[Michael Stegen]

Hi Michael, 

Nice work! good to hear that another Michael is working on feeders too. ;-)

Please show us some pics from your work so far, i'm curious to see how you snap fit the sprocket wheel together (i assume the sprocket, and "tape support" wheel?)

I'll post some pictures from my progress so far.  I have manged to bring the thickness of each feeder down to 10mm.

I now snap fit them directly onto the frame, and the tape path feeds the spend tape just like Neils design under the frame.

Neil, have you tried other then paper tapes on your feeders? As i noticed that on the drag type feeder, i dont think you can use anything else. There is simply not enough room.

For plastic tape you'll need a cutout in the tape path for the parts, with support rails on both sides. 3mm on the side where the sprocket holes are, and  0,6mm on the other side.

-Mike

--

[Paul Jones]

PS I’ve put up some stl files from my old design on github, but I’m not sure they would be of any use. I was going to use a stepper motor to advance the tape originally, and have the motors offset to pack the feeders nice and tight.

My second idea (which annoyingly I have lost the design for) was using a miniature dc gearmotor and a worm drive.

Mine is like this but with about twice as many gears – I think it was 60RPM @ 5v. Cost was not much, from eBay.

http://www.diyrc.com/E-Retract-Motor2.jpg

 

 

Paul.

[Paul Jones]

Geez, 3^rd time lucky!

https://github.com/PeeJay/Misc-PCB-Pick-and-Place

 

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Paul Jones
Sent: Monday, 3 November 2014 10:07 PM
To: fire...@googlegroups.com
Subject: RE: [FirePick] Mike's Feeder Development

 

PS I’ve put up some stl files from my old design on github, but I’m not sure they would be of any use. I was going to use a stepper motor to advance the tape originally, and have the motors offset to pack the feeders nice and tight.

My second idea (which annoyingly I have lost the design for) was using a miniature dc gearmotor and a worm drive.

Mine is like this but with about twice as many gears – I think it was 60RPM @ 5v. Cost was not much, from eBay.

http://www.diyrc.com/E-Retract-Motor2.jpg

 

 

Paul.

 

From: fire...@googlegroups.com [mailto:fire...@googlegroups.com] On Behalf Of Paul Jones
Sent: Monday, 3 November 2014 9:13 PM
To: fire...@googlegroups.com
Subject: RE: [FirePick] Mike's Feeder Development

 

I generally use a 0.4mm nozzle on my ultimaker with PLA and it works great. I’ve got a draw full of helical gears I printed when I was trying to design a simple rotating head for a previous PnP machine, and they mesh surprisingly well! I even managed to get a small worm drive working, so printing gears is definitely doable.

I recently acquired a 0.2mm nozzle and it seems to work well so far. I used it to print the tiny 5mm lock pieces that fit into the aluminium extrusion (the stuff I bought was quite narrow so I had to redesign that part).

 

Cheers,

Paul.

--

[Neil Jansen]

On Mon, Nov 3, 2014 at 5:41 AM, Michael Stegen <mikes...@gmail.com> wrote:

Neil, have you tried other then paper tapes on your feeders? As i noticed that on the drag type feeder, i dont think you can use anything else. There is simply not enough room.

For plastic tape you'll need a cutout in the tape path for the parts, with support rails on both sides. 3mm on the side where the sprocket holes are, and  0,6mm on the other side.

 

Have you seen the latest version of my drag strip feeder that has the cutouts for one part and two sprocket holes?  Admittedly I haven't tried it with other types of tape, but I guess I'll try that when I get home today.  I don't foresee any issues at this point, the new design is a lot better than what I had.  

[Michael Stegen]

Hi Neil, 

Yes i have seen the latest version, looks good. It's just that some tapes will not fit.

For example, i've got a reel of 10uf 16V caps, on (transparent) tape. That tape is 1,9mm thick.

[Neil Jansen]

Oh, I understand now.  That's will be addressed by making different parametric versions of the design.  However, I wanted to vet the 8mm paper tape version first before making all the other versions.  Once that's working great, I'll add the pocket for deeper parts like caps, and then make all the other versions for wider tape (12mm, 16mm, 32mm, 48mm, etc...).  I did originally keep the paper tape and the cup versions separate, because of the position of the leaf spring underneath the tape. Some commercial smt feeders have removeable pieces to switch between cup and paper tape, but we'll just cross that bridge when we get there.  Printing out another feeder is cheap and easy, so why make small parts that can be lost.  

The current design, once parametrically modified, should be able to hold very deep parts like electrolytic caps and such although I've obviously not gotten that far yet, so we'll see.  I should hopefully be getting to the drag-pin feeding code in OpenPnP very soon to get that working well with the firepick mechanical pieces.  As I've said before, it's certainly the hardest and riskiest thing to do on the entire machine, so I'm not taking it lightly... I'm going to try a few clever things in order to increase the reliability of it.

[Bruce Bock]

This might give you some ideas.

I was a contributer for this.  Since these were listed as open source he shouldn't have an issue

with me sending them to you.

https://www.kickstarter.com/projects/1973759599/open-source-smd-parts-feeders

http://www.richardspelling.com/?p=683

[Douglas Pearless]

Hi Team,

As an idea, I have seen a video of a DIY Pick and Place machine where the pick head it self is used to advance the tape.

In this instance:

1. The pick and place head moves over to the tape
2. Then vacuum pick shaft is inserted into the guide hole in the tape beside the next component, i.e. the one that will be placed next.
3. The the tape is pulled forward to position the next component using the head itself, and 
   
4. The pick head moves up from the guide hole, across to now pick up the component and places it on the PCB

This has the advantage that no extra parts are required, BUT it may be necessary to instead have a dedicated pin on the head as when picking larger components the nozzle will be bigger than the guide holes in the tape.

Thoughts?

Cheers

Douglas

[Michael Anton]

This is exactly what Neil is doing with his drag feeders.  I think he is using a dedicated pin, rather than the nozzle though.

This may not work for all tape widths however.  At some point, the force required to pull the tape may be more than the machine is capable.  The other downside, is that it is much slower than the automatic feeders that some of us are proposing.  This may not matter much for a hobbyist machine, but it becomes more important for limited production machines, like some of us wish to build.

It's always nice to have more options available, and each individual can decide what their requirements are.

Mike

[Neil Jansen]

Hi Douglas,

Mike is right, that has been the whole point of our machine since day 1.  It is partially my fault for not having that information in a central place for you to read, however I think if you spend some time reading some of the older posts on this group, then it will easily answer a lot of the questions and show a lot of the proposed solutions to the problems that we're facing.  We've got a lot of very smart and creative people here and we've come up with all sorts of great ideas over the last few months.

Mike's comments about the bigger feeders not working with drag-feed are correct as well, however there are a few ways to mitigate this.  Larger tape feeders are obviously easier to motorize.  They don't even have to be proper size NEMA 17's or 14's, they could be the smaller 28BYJ-48 geared type, or a micro metal gearmotor, hopefully geared down quite a bit with a bit of vision to stop it at the right spot.  

Or, if it's the type of tape with sprocket holes on both sides (common on 48mm, etc), then you could possibly even keep the drag feed, and use two sprockets tied together with a common axle.  Pulling one side just ahead of where the sprocket engages would pull the whole thing evenly.

--

[Michael Anton]

Ok, since you asked for some pictures, here is what I have so far.  I've only spend a weekend on the design so far, and much of that was making sure that everything was fully parametric, and could be modified for different numbers of teeth, and tape width by updating parameters in a spreadsheet.  I also drew an accurate model of the motor, as I always make assemblies in Cubify as I go along.

The two halves of the sprocket are snap fit together.  There is a bump around the hub on the toothed side, and a matching groove on the other side that keeps them locked together.  I could release an early version of this part if people want to work with it.  It is a nice model, in that you can change the number of teeth, thickness, and it automatically figures out the diameter and tooth profile to fit the tape properly.  Or, at least that is the theory, as I haven't actually tried any sizes other than this one.

The plan is to have a large diameter pivot on the housing that the motor mounts inside of, and that the sprocket rotates on, so that it has some support other than the wimpy motor shaft.  This pivot is thinned out where the motor is closest to the sprocket, so there really is room for it.  The motor fits nicely underneath the sprocket, and it would be possible to do this in a 10mm wide feeder, though the thickness where the motor mounts, would only be 1mm, but it is for a small area, so it would be ok.  This is the same thickness of the sprocket flanges, and teeth.

If I can figure out a way to drive the sprocket without having the motor mounted inside it, then I could make the sprocket smaller, and the torque would be better.  I could also add another level of gearing, but at that point it seems like I might as well pick another motor.

I could probably make the sprocket smaller than it currently is, as it accounts for up to 5mm deep pockets on the tape, which is probably overkill for 8mm wide tape.

Mike

[Douglas Pearless]

Well this looks really promising.

I assume you have already seen this: https://www.youmagine.com/designs/open-pick-n-place-tape-feeder

Cheers

Douglas

[Michael Anton]

Yes, I saw that one and Richard Spelling's design, both quite some time ago.  I'm keeping their approaches in mind.  Thanks for pointing them out though.

I think Neil already has the simplest solution, though I may come up with another way to handle the cover tape removal that would help improve it a little bit.  I'm looking for something in between the complicated commercial designs, and Neil's, with the goal of keeping the feeders as narrow as possible so that more fit on the machine.  It has been my experience that you can never have too many feeders.

I wonder how many already patented technologies we are likely to reinvent in the process?

Mike

[Paul Jones]

As far as I know there is no law against publishing a design based on a patented idea, and no law against building it individually for personal use. Even if a prototype was patent encumbered we’ll come up with something better soon enough anyway.

I had a look through patents a while ago and I didn’t really find anything useful. Most of them I thought were unnecessarily complicated, and not at all DIY friendly.

 

Paul.

--

[Karl]

Mike, that is quite startling. I never realized that sprocket teeth could be 3D printed. Great work!

[Michael Stegen]

Nice work Mike!

Karl, just FYI I'm also using a 3D printed sprocket and support wheel in my feeder. 

2014-11-07 15:35 GMT+01:00 Karl <ka...@firepick.org>:

Mike, that is quite startling. I never realized that sprocket teeth could be 3D printed. Great work!

--

[Michael Anton]

I wasn't sure it would work either, as I tried printing one in conductive ABS, and it was a dismal failure.  I'm glad that I tried one in PLA!!

Mike

[Karl]

Ahhhh! So PLA is structurally stronger in fine detail vs. conductive ABS?

It seems like ESD management of feeder now gets interesting. With 3D PLA sprockets, those PLA tips will be ESD generators potentially. I guess conductive ABS could be used for the wheel and feeder body. Maybe that's OK. I'm worried that the ESD spray coating won't last on sprocket teeth.

[Michael Anton]

I find that I can never get the same detail out of ABS as I can from PLA, conductive or not, as it just never seems to print as well.  ABS is more impact resistant, but it is softer, and more flexible as well.  I'm not sure that it is the best candidate for the sprocket, even if it printed correctly.  That said, the conductive ABS that I have is perhaps not the best, since it is from RepRap-Walmart, rather than PushPlastic.

It is my intention, once I have a working design, to look at making the sprocket out of waterjet or laser cut aluminum, at least for the toothed part.  This could be electrochemically deburred as well to remove all of the sharp edges left from cutting.

Perhaps another option would be to have a grounded carbon brush wipe the sprocket teeth as it turns to remove built up charge.  Similar to what is done in laser printers to stop charge from building up on the paper as it moves through the printer.

Mike

[Karl]

Thanks, Mike. That makes perfect sense.

[Neil Jansen]

I can be the guinea pig, I have the 3Dxtech conductive ABS from push plastic and a really small hotend (0.35mm) that I bought a while back and never used.  I'm not sure when I'll get to it but hopefully in the next few weeks, worst case.  I would agree with Mike that waterjet cut metal cogs would be a great way to go as well. 

--

[Douglas Pearless]

Hi Team,

A quick heads up, I am working on a very simple and low cost feeder design.

As my CAD skills are rubbish, I will try to describe how it works.

I am intending to use this motor assembly: http://www.aliexpress.com/item/-/1925287996.html for s system that sits ON TOP of the SMD tape and not below.

In essence the output shaft is connected to a 3D printed shaft that has a slot that the start of the cover tape is locked into using a simple bar along the length of the tape.

The cover tape is wound back and this pulls the SMD tape through, with the sensor detecting when the next hole lines up.

Note the sticky side of the tape is wound inwards so the roll of wound up tape rides smoothly over the inbound cover tape.

I have done a bit of research and while tape breakage was an issue several years ago, it does not appear to be such an issue now.

I am intending to use IrDA ( http://www.aliexpress.com/item/HSDL-3201-001-IRDA-MODULE-115-2KBPS-8-SMD-HSDL-3201-3201-HSDL/1902146114.html ) to communicate with the Firepick itself and intent to use a EFM32ZG108 micro and a cheap motor driver ( http://www.aliexpress.com/item/-/1981658877.html ) and have a simple button for the user to push to configure the micro via IrDA.

I think the feeder could be build for $10-15, maybe.

Thoughts?

Cheers

Douglas

[Douglas Pearless]

Fixed a typo:

In essence the output shaft is connected to a 3D printed shaft that has a slot that the start of the cover tape is locked into using a simple bar along the length of the shaft.

[Michael Stegen]

I have seen something similar used on this pnp machine:

https://www.vbesmens.de/en/pick-and-place/automatic-feeder.html

But for detecting holes in the (transparent) tape, he uses a small switch instead of a beam of light.

2014-11-24 1:28 GMT+01:00 Douglas Pearless <douglas....@gmail.com>:

Fixed a typo:

In essence the output shaft is connected to a 3D printed shaft that has a slot that the start of the cover tape is locked into using a simple bar along the length of the shaft.

--

[Douglas Pearless]

Yes, I have now checked that out, I am looking at a much smaller implementation that is much simpler, though I may choose a different motor and gearbox as I guess there is no long term guarantee of the unit I found.

[Michael Anton]

I'm not convinced that there is ever a long term guarantee on any item. :-)  One option to circumvent this is to buy a lot of them...

Take a look at the N20 series of motors that can be had all over the place.  They are more expensive, but appear to be quite well built, and use metal gears.  I picked up a 30rpm one up on ebay for about $4, just to see if they are useful for this purpose.  These have a lot of torque, but alas, the drive will need to be rotated 90 degrees, which is not completely trivial.

Mike

[Neil Jansen]

I'm not convinced that there is ever a long term guarantee on any item. :-)  One option to circumvent this is to buy a lot of them...

Take a look at the N20 series of motors that can be had all over the place.  They are more expensive, but appear to be quite well built, and use metal gears.  I picked up a 30rpm one up on ebay for about $4, just to see if they are useful for this purpose.  These have a lot of torque, but alas, the drive will need to be rotated 90 degrees, which is not completely trivial.

For motors, I'm going to try using these on my new feeder design:

 

I'm hoping it's got enough torque to do the job.  Would be sort of easy to gear it down if necessary, but we'll see.  The size is right, and I have a strong preference for stepper motors over analog motors for driving the tape (firing off a specified # of steps is easier than using a DC motor and feedback).  At $30 they're not cheap, but they're still a better deal than a full-size commercial feeder.  I've pretty much decided that I want to get full-auto feeders working before I try to get the drag-pin feeders to work.  These seem a bit more practical than the first full auto feeder design that I had, as they take up less space and are more modular.

 

[Douglas Pearless]

Looks like our two designs are going to diverge a wee bit here :-)

I am battling my CAD system (FreeCAD) as I learn how to use it properly (I don't use Windows any more, just OS X).

I have a design of a universal feeder that uses a scissor jack arrangement with a knob that controls the width of the side panels of the feeder tray, and a telescopic rod for the spooler and the cover tape peeler too.

It also has a similar depth arrangement for different thickness of tapes. 

There are probably going to be three sizes, the first for 8-24mm tapes, second for 32-72mm and the third for 88-120mm wide tapes.

For the vast majority of users, they only need the first feeder design and it is changeable by just rotating the tape width knob and tape depth knob.

It will probably take me a week to get the design down so I can share it.

Then again, it may turn out to be a design fail :-)

Cheers

Douglas

[Michael Anton]

I like those motors too, but it is too bad about the cost.  That is the type of motor that I've been looking for and couldn't find as an inexpensive one.  It would be possible to mount this in the center of the sprocket like I am doing with the small geared motor, and it would probably allow the sprocket to be smaller than mine.

I figure that using a stepper vs a DC motor is a bit of a tradeoff.  The stepper requires more electronics to drive it, but doesn't need as much feedback, though you will still need some way to index the stepper design, or the pockets may not always be at a consistent location after loading the tape.  The DC motors are simpler to drive but need more feedback.  The feedback doesn't have to add much to the cost though, so likely a DC motor design will be much less expensive.  I found some really small slot opto sensors that I'm hoping to use to pick up the sprocket teeth.  A couple of them placed slightly out of phase should give really good feedback.  The motor I have turns slow enough, that I suspect it won't take anything too fancy to get it to advance a precise amount.

Mike

[Michael Anton]

On Monday, November 24, 2014 8:00:54 PM UTC-7, Douglas Pearless wrote:

Looks like our two designs are going to diverge a wee bit here :-)

I am battling my CAD system (FreeCAD) as I learn how to use it properly (I don't use Windows any more, just OS X).

There are days when I have a real battle with Cubify, so I feel your pain.

 

I have a design of a universal feeder that uses a scissor jack arrangement with a knob that controls the width of the side panels of the feeder tray, and a telescopic rod for the spooler and the cover tape peeler too.

It also has a similar depth arrangement for different thickness of tapes. 

There are probably going to be three sizes, the first for 8-24mm tapes, second for 32-72mm and the third for 88-120mm wide tapes.

For the vast majority of users, they only need the first feeder design and it is changeable by just rotating the tape width knob and tape depth knob.

It will probably take me a week to get the design down so I can share it.

Then again, it may turn out to be a design fail :-)

I hope you are referencing the EIA-481 specs for component tape during this process.  There are a number of tolerances and dimensions that are important to keep in mind to make these things universal.

Mike

Cheers

Douglas

[Neil Jansen]

On Mon, Nov 24, 2014 at 11:35 PM, Michael Anton <3d.m...@gmail.com> wrote:

I like those motors too, but it is too bad about the cost.  That is the type of motor that I've been looking for and couldn't find as an inexpensive one.  It would be possible to mount this in the center of the sprocket like I am doing with the small geared motor, and it would probably allow the sprocket to be smaller than mine.

$30 isn't cheap, but if it works it will let me work on software, which is a good tradeoff.  

 

I figure that using a stepper vs a DC motor is a bit of a tradeoff.  The stepper requires more electronics to drive it, but doesn't need as much feedback, though you will still need some way to index the stepper design, or the pockets may not always be at a consistent location after loading the tape.  The DC motors are simpler to drive but need more feedback.  The feedback doesn't have to add much to the cost though, so likely a DC motor design will be much less expensive.  I found some really small slot opto sensors that I'm hoping to use to pick up the sprocket teeth.  A couple of them placed slightly out of phase should give really good feedback.  The motor I have turns slow enough, that I suspect it won't take anything too fancy to get it to advance a precise amount.

It's possible with the downward looking camera that we can provide feedback via computer vision, if a few conditions are met: camera latency, advance speed (mm/sec or sec/mm), and CV operation computation speed.   That certainly makes the driving circuitry a lot cheaper... Just a mosfet to each feeder's DC motor.  The software wouldn't be fun (or maybe would be fun depending on how you look at it), however, it's not written at this point, so to me, the lowest-common denominator is some sort of motorized feeder that can index on its own, which is the stepper motor route.

Also, come to think of it, I think on the RobotDigg site, that the NEMA 23 ultrathin motor is meant for the actual component tape movement, as it's torquier, and I bet the smaller NEMA 11 one I linked is for the cover tape winding... As it's very weak and would probably stall when tightened to a certain point, and can be adjusted by the driver.  That brings the cost of a single feeder to the $70-100 USD point, which is not cheap, but it would be an extremely convenient full auto feeder with cover winding, in a super tiny form factor.

[Michael Anton]

I'm hoping to come up with a full auto feeder that doesn't rely on vision at all for the feedback.  It would be completely stand alone, and hopefully requires no interface to the machine.  I don't think it will be as expensive as what you are proposing, but the design is not done yet, so things could change.  Perhaps once I have accomplished it the hard way, it can be simplified later with additional software, but for the moment, that is outside of my skill set.

I'm still hoping that the cover tape winding can be done passively based on the movement of the tape as I've seen in other feeders.  Mechanically, so far it doesn't look too bad, and I have most of it worked out, but I haven't actually tried it yet.  It does at least allow the feeder width to be kept narrow, and doesn't require another motor to drive it, which is pretty nice.

I'm hoping that you might leave some of the feeder development to the few of us that are working on it for the moment, and focus on the things that only you can do easily.  I would hate to see a bunch of duplicated effort slowing the process down.  I know that in the long run, you will want to add your handiwork into the designs, but perhaps that can come after we have something working.  The fully parametric design that I am working on would be a good start for others, even if they don't like how I've implemented everything.

Please don't take this as my suggesting what you should or should not do.  I'm just eager to actually see the whole thing working, and to build one myself, so the sooner the majority of the machine design settles down, the sooner I can begin...

Mike

[Neil Jansen]

I'm hoping that you might leave some of the feeder development to the few of us that are working on it for the moment, and focus on the things that only you can do easily.  I would hate to see a bunch of duplicated effort slowing the process down.  I know that in the long run, you will want to add your handiwork into the designs, but perhaps that can come after we have something working.  The fully parametric design that I am working on would be a good start for others, even if they don't like how I've implemented everything.

Please don't take this as my suggesting what you should or should not do.  I'm just eager to actually see the whole thing working, and to build one myself, so the sooner the majority of the machine design settles down, the sooner I can begin...

I will need "something" to work on vision and system integration, but none of the current SMT feeder designs suit my fancy.  What I'm coming up with is just a sprocket on a stepper motor, with some squirted plastic around it.  Nothing crazy.  

Personally I think having lots of different feeder designs is great for diversity.  I've mentioned before that I think it's the single biggest area for diversity in the entire machine.  There are so many different ways to make an smt feeder, and all have compromises (some more than others).  I'm not going to spend a lot of time on my design.  Just a few hours (days at most) to get something usable for software development, even if it's expensive and impractical in the long run.