Rapid Feeder - A new type of feeder

6136 views
Skip to first unread message

Shai

unread,
Jan 17, 2020, 12:24:38 PM1/17/20
to OpenPnP
In continuation of my last post here on I2C feeder design, which allows the flexibility to move feeders along an axis without the need for mechanically indexing them, here is a picture of my first working prototype without a sprocket. I hooked it up to a breadboard setup to test. My next plan is to get CNC'd PCBs. These PCBs are expensive at low volume and my hope is that others here may also be interested in this to help out in developing it. In doing so, I am wondering how many would like to join in on the first round of prototypes? I have most of the pieces already, just need to buy more motors to match the volume of PCBs and 3D print the cover of each feeder from PLA (which is a non critical part for dimensional accuracy). The sprocket will also be made out of 1mm thick PCB. This would include the motherboard to plug the feeders into as well. All the feeders will be SMT assembled, but ship mechanically un-assembled (to save space in shipping box). The feeder is designed to work with all tape types shown on this guide: http://www.vishay.com/docs/60034/tapereel.pdf This includes proper min bend radius of wider tapes. This first batch will only be for 8mm wide tapes, but the feeder is designed to be expandable to wider tapes by swapping the standoffs and dowels to longer ones. The entire setup runs on Arduino code and uses two optical sensors. What still needs to be figured out is the mounting of the feeders to the motherboard. Currently there are two holes on top and under the 5 pin spring connector as you can see, but these don't work well. The idea was to have it clip onto two rods, but PCBs don't flex well enough to make it work. Any suggestions appreciated.

How many people here would be interested in getting these feeders in classic green color and what is a price point that works for you so I can calculate the volume required to meet that price point? Please also let me know quantity you would be interested in. My hope is to get enough people interested so we can lower the price point with a group buy. Note that these have only been mechanically tested, but the code is ready and compiles.


2019-12-23.jpeg


IMG_7308.JPG



screenshot_166.png



Mike Harrison

unread,
Jan 17, 2020, 12:30:17 PM1/17/20
to ope...@googlegroups.com
On Fri, 17 Jan 2020 09:24:38 -0800 (PST), you wrote:

>In continuation of my last post here on I2C feeder design

No! No! No! - I2C is absolutely the wrong for this type of application for so many reasons.

RS232 or RS485. CAN if you must, but that's more expensive.

Even 5V TTL UART would be better than I2C



Marek T.

unread,
Jan 17, 2020, 12:37:32 PM1/17/20
to OpenPnP
Hi Shai,

Is there not any shuter?
Do you have some photos with the feeder fully assembled? With the reel and tape, showing how this is advanced and how the covering tape is taken off?

Shai

unread,
Jan 17, 2020, 12:44:56 PM1/17/20
to OpenPnP
@Mike, I don't see why this wouldn't work with I2C? It's capable of hosting 128 devices on one motherboard.

@Marek, see attached picture of how cover tape gets pulled. When it is loose, the optical sensor is clear and knows to cut off power to motor to stop pulling tape. I think by shutter you mean something cover the components after the tape is pulled? There is none. I designed it so the part is picked right where the tape is pulled off. The motors are not super fast and *shouldn't* make the parts bounce. 

screenshot_168.png


Shai

unread,
Jan 17, 2020, 12:51:45 PM1/17/20
to OpenPnP
Here is another view from the top. The opening dimensions change with the width of the tape. Wider the tape = bigger opening. So the plan is to have slight variations between different feeder widths. 

IMG_7956.JPG


Jon Raymond

unread,
Jan 17, 2020, 12:54:58 PM1/17/20
to OpenPnP
It's been explained to you before why I2C shouldn't be use. It's a noise issue. Do you have a working feeder(s) to showcase on a machine?

Jason von Nieda

unread,
Jan 17, 2020, 1:10:26 PM1/17/20
to ope...@googlegroups.com
Hi Shai,

Nice work - looks like it's really come a long way!

For mounting, how about something like this?

IMG_3400.JPG

Forgive the rough drawing. Basic idea is that the mounting point is a piece of t-slot extrusion (perhaps 20x20 or 40x40). The feeder has a fixed hook on the top that catches in one of the t-slots, and the feeder has a spring loaded lever on the bottom that catches in the bottom t-slot. Your slider PCB could mount to the "front" of the extrusion using a third t-slot.

Loading the feeder on would mean angling it a bit to get the top hook caught and then lower it until the spring loaded lever rolls over the bottom slot and catches. Pull on the lever to remove.

I feel like maybe you looked at something like this previously, but I didn't see the email. Forgive me if you've already considered and discarded this idea.

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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/77b7477f-6c35-43c9-836b-a6a13274430c%40googlegroups.com.

Shai

unread,
Jan 17, 2020, 1:10:45 PM1/17/20
to OpenPnP
On Friday, January 17, 2020 at 12:54:58 PM UTC-5, Jon Raymond wrote:
It's been explained to you before why I2C shouldn't be use. It's a noise issue. Do you have a working feeder(s) to showcase on a machine?

One has to consider other constraints here. There are a very limited number of spring contacts on the market and the pin count maxes out at 5 pins I believe from the research I did. In addition to that, the spring connector would have to have a high cycle count. The spring contacts I am using have 5,000 cycle counts, ensuring the feeder lasts for a long time. I2C can be improved using a filter of sorts. The feeders are 10mm width, so you can fit quite a bit, over a short distance. As mentioned, I don't have working feeder yet with the PCB because they have to be CNC'd, which is expensive and why I was hoping to do a group buy for others who may be interested to experiment.
 

Shai

unread,
Jan 17, 2020, 1:15:15 PM1/17/20
to OpenPnP
Thanks Jason! It's been months in the works testing multiple prototypes before committing to CNC'd PCBs. I have thought extensively about mounting to extrusions, but there's a few issues. 

1. Requires to grow the PCB footprint significantly, which adds cost.
2. Are you suggesting a lever for each feeder? This brings back mechanical indexing (assuming the lever is on the machine side).
To unsubscribe from this group and stop receiving emails from it, send an email to ope...@googlegroups.com.

Shai

unread,
Jan 17, 2020, 1:17:06 PM1/17/20
to OpenPnP
I was hoping I can find some sort of metal or plastic clip off the shelf that can be sandwiched between the PCBs. Of course you can 3D print it... but that's a lot of printing :) It would also have to grow in width with wider tapes.

Jason von Nieda

unread,
Jan 17, 2020, 1:18:21 PM1/17/20
to ope...@googlegroups.com
No, the lever is part of the feeder. No indexing required. I would imagine the top is a 3D printed piece that goes between the two PCBs and the bottom is 3D printed or maybe CNCd if more strength is needed.

If it's not clear still, let me know and I'll try to do a better drawing :)

Jason


To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/84821fcc-7243-4606-b70e-ed2612d9398f%40googlegroups.com.

Shai

unread,
Jan 17, 2020, 1:23:59 PM1/17/20
to OpenPnP
Hmm CNC = much higher cost. Here is what I was thinking of 3D printing for a "clip" design that gets sandwiched between both PCBs.

screenshot_169.png


Jason von Nieda

unread,
Jan 17, 2020, 1:34:19 PM1/17/20
to ope...@googlegroups.com
I think you are going to find that if you use metal rods they are going to flex significantly when you are adding/removing/moving feeders, throwing off all your feed locations every time you touch one.

Here's a better drawing:

IMG_4363.JPG

I think the lever / catch can easily be 3D printed. It will be at least 6mm thick or so, and should be plenty strong.

Jason



On Fri, Jan 17, 2020 at 12:24 PM Shai <shai...@gmail.com> wrote:
Hmm CNC = much higher cost. Here is what I was thinking of 3D printing for a "clip" design that gets sandwiched between both PCBs.

screenshot_169.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.

Shai

unread,
Jan 17, 2020, 1:42:23 PM1/17/20
to OpenPnP
I agree that it can flex. That was my concern too. However, this also depends on the linear rod material and diameter. I have some samples here that are about 250mm long and don't flex. It may vibrate a bit when adding/removing feeders, but should return to position once untouched. The issue I see right away other than cost with your design is it would require increasing the feeder size significantly as there's no more room left.

Jason von Nieda

unread,
Jan 17, 2020, 1:45:49 PM1/17/20
to ope...@googlegroups.com
Fair enough. I think you are going to find that when you hang a bunch of leveraged weight off those rods they are going to flex quite a bit, unless you are talking like 20mm+ diameter, but I guess the only way to know for sure is to try :)

Jason


On Fri, Jan 17, 2020 at 12:42 PM Shai <shai...@gmail.com> wrote:
I agree that it can flex. That was my concern too. However, this also depends on the linear rod material and diameter. I have some samples here that are about 250mm long and don't flex. It may vibrate a bit when adding/removing feeders, but should return to position once untouched. The issue I see right away other than cost with your design is it would require increasing the feeder size significantly as there's no more room left.

--
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.

Dave McGuire

unread,
Jan 17, 2020, 2:02:51 PM1/17/20
to OpenPnP
On 1/17/20 12:54 PM, Jon Raymond wrote:
> It's been explained to you before why I2C shouldn't be use. It's a noise
> issue. Do you have a working feeder(s) to showcase on a machine?

RS485 would be ideal for this.

I'm not at all throwing stones at what looks like a really nice feeder
design, but I'd be awfully nervous about signal integrity there. I2C
was not designed for this. Yes, it may be able to be made to work, but
there are protocols that were specifically designed for applications
like this one, and they're easy and cheap to implement.

-Dave

--
Dave McGuire
McGuire Scientific Services, LLC
New Kensington, PA

Shai

unread,
Jan 17, 2020, 2:14:41 PM1/17/20
to OpenPnP
I'm all up for implementing RS485. I'm not familiar with it, so if someone here is interested in offering a helping hand, I'll be happy to implement it. Seems like MAX485 chips are cheap on aliexpress, so perhaps its fine to implement it unless there's additional ICs needed? Note I'm currently using Attiny84 on feeders and Atmega328P on motherboard. So if there's not enough pins, it would require to upgrade to a bigger IC. Again, I'm not familiar with RS485, so if someone wants to help modify it to that, I'm happy to do so.

Marek T.

unread,
Jan 17, 2020, 2:19:17 PM1/17/20
to OpenPnP
Shai shatter is not matter of rapidity of the feederb only. In case where the ventouri is used, when there is no suction the nozzle is blowing. And then small parts are blowed away from the tape pockets. The shutter must be opened straightly before the tip touches to the part, moreless in the half of the distance. So in this case, my case, the feeder without the shutter is useless for parts <= 0805. Pity because it looks really cool.

Dave McGuire

unread,
Jan 17, 2020, 2:20:38 PM1/17/20
to OpenPnP
I'd be happy to assist you with the RS485 aspects of the design.

Shai

unread,
Jan 17, 2020, 2:23:37 PM1/17/20
to OpenPnP
Understood. I designed these feeders with my setup in mind (PNP I am still working on) which does not blow air when vacuum is off. I gathered that most OpenPNP machines that are built do not have air blowing, so I did not consider a shutter. A shutter would also not be easy to implement in this design.

Dave McGuire

unread,
Jan 17, 2020, 2:35:58 PM1/17/20
to ope...@googlegroups.com
Actually, and I don't know why I hadn't thought of this before, CANbus may be an even better option for THIS, especially if you're already doing I2C-ish things like multiple slave communications. I'm in the middle of a CANbus design now, and could also assist with that, if needed.

Marek T.

unread,
Jan 17, 2020, 2:40:04 PM1/17/20
to OpenPnP
You're right, probably most of people use vacuum pumps not compressors+ventouri. But if you'd implement it the product could be fully universal and really good!
I can turn off the blowing, it's adjusted with the screw, and then it works like machine with vacuum pump. But from my experience, with the blowing the small parts are placed better. Also when I turn off the the blowing, when do the discard the small parts like 0603 gets off the tip in some 3sec or more, 0402/0201 don't get off at all but remain on the tip - don't know why but it is.

I need some 30 good feeders. Now I have some feeders but they are not precise enough for 0402. I think about pneumatic yamaha but it's complex plate and not cheap, electrical yamaha - simple plate but expensive feeders...

Do you live in around of EU or US?

Shai

unread,
Jan 17, 2020, 2:42:30 PM1/17/20
to OpenPnP
Located in US and understood about the blowing reasoning. I think that is for another future feeder design ;)

James Carter

unread,
Jan 17, 2020, 4:23:04 PM1/17/20
to OpenPnP
People are being quite harsh here. You've come up with a good design and if it works for you that's great.

Push forward with it and see if the design works for you and if you have any issues with I2C you can always modify the design.

Good on you for getting a decent design that works in a small footprint.

James Carter

unread,
Jan 17, 2020, 4:26:17 PM1/17/20
to OpenPnP
P.S. I have a CNC I can lend some time on to help you prove this design out. Smallest endmill I generally use is 2mm.

If you want to send me something to make I'll run it for you.

ja...@loveelectronics.co.uk

Michael Anton

unread,
Jan 17, 2020, 5:05:40 PM1/17/20
to OpenPnP
You probably have far more gearing on that design than you actually need, especially since you are using a worm gear, and a gear head motor. I think at the very least, you could drop out the middle gear.  If you find out later that you need more gearing, you can always move to one of the other N20 motors with a higher gear ratio.

Shai

unread,
Jan 17, 2020, 5:10:11 PM1/17/20
to OpenPnP
To reply to Michael Anton's email that got sent to the emailing list instead of here:

"You probably have far more gearing on that design than you actually need, especially since you are using a worm gear, and a gear head motor. I think at the very least, you could drop out the middle gear.  If you find out later that you need more gearing, you can always move to one of the other N20 motors with a higher gear ratio."

Adding that gear was necessary because I am using off the shelf gears. You can't 3D print these gears either for many reasons. And there's no off the shelf worm gear that can attach directly to the sprocket or even the middle gear. This was literally the only combination of gears/worms that are possible with off the shelf parts that are cheap.

I should probably re-iterate to everyone that the goal of this feeder is to make a feeder that is cheaper than any other out there. Of course reliability is important, otherwise it's useless. But, if the cost goes up too high, you may as well get a metal stamped feeder off Ebay or Aliexpress.

Marek T.

unread,
Jan 17, 2020, 5:18:45 PM1/17/20
to OpenPnP
Not harsh but honest.

The group is to exchange the thoughts, not to agree with everything.

For me the feeder looks very very cool. Except the shutter lack and i2c ;). But it's still far from good working... Some 1-2 years ago, here was a man named Darren with Assisi very promising project of the feeder. I remember "he had a dream" but it looks he woke up and forgot about it ;). Let's wait so...

Marek T.

unread,
Jan 17, 2020, 5:20:59 PM1/17/20
to OpenPnP
Assisi = also...
Damned keypad auto-corrector...

Mike M.

unread,
Jan 17, 2020, 6:10:42 PM1/17/20
to OpenPnP
Shai
You can make sch for bouth options like this:
and place under the chip two resistor pads - than you are ready for bouth options !?

Mike

Shai

unread,
Jan 17, 2020, 6:13:17 PM1/17/20
to OpenPnP
Mike,

The Attiny84 I am using on the feeder doesn't have RX/TX pins. I would need a different IC. Just how much better is RS485 than I2C? And is anyone willing to code the Arduino?

John Plocher

unread,
Jan 17, 2020, 7:32:22 PM1/17/20
to ope...@googlegroups.com
On Fri, Jan 17, 2020 at 3:13 PM Shai <shai...@gmail.com> wrote:
The Attiny84 I am using on the feeder doesn't have RX/TX pins. I would need a different IC. Just how much better is RS485 than I2C? And is anyone willing to code the Arduino?


RS422/485 is "just like" RS232, but with different electrical details that make it a good fit for this use case - the bus is differential, and so can be robust in the case of noise and differing ground floors on either end.
It exists in both a 2-wire and 4-wire versions; in your case, the 2-wire half duplex version should be sufficient...

The 'Tiny should be able to serial on the pins you would have used for I2C; if you are banging something up with the Arduino IDE, you can use the SoftwareSerial library:

#include "SoftwareSerial.h"
const int Rx = SDA; // change as needed
const int Tx = SCL; // to match your circuit

SoftwareSerial rs485Serial(Rx, Tx);

void setup(){
pinMode(Rx, INPUT);
pinMode(Tx, OUTPUT);
rs485Serial.begin(9600);
}

byte resp[...size...];
void loop() {
rs485Serial.write(... some command...);
resp = rs485Serial.read(); ...
}

RS485 is a multi-drop half duplex protocol that can easily support many feeder clients, the protocols used on top of it tend to be half duplex so that you don't have to worry about devices talking over each other at the same time.
This implies that each feeder/device/slave will need some sort of fixed ID, so you can say things like "feeder 42, do _this_" and have just the feeder with ID 42 respond - not really different from giving each feeder its own unique I2C address.  From a protocol design perspective, think in terms of initially iterating through an ID range to enumerate the installed feeders and then a loop of "read, modify, write" to get things done.

  -John
 

John Plocher

unread,
Jan 17, 2020, 10:00:07 PM1/17/20
to ope...@googlegroups.com
also

... depending on the bus driver chips you choose (1-unit load or 1/4 unit load), at a slow baud rate (19,200, not 256,000) and short bus (3', not 5,000'), you should be able to handle up to 32 - 64 nodes on a single bus.

  -John

Jan Juranek

unread,
Jan 18, 2020, 4:44:02 AM1/18/20
to OpenPnP
Hey Shai,

perfect design!! Will you make this design open source?  We still don't have our machine built, but we can help you with coding as we are doing this for living :)

Let me know.
Honza

Dne pátek 17. ledna 2020 18:24:38 UTC+1 Shai napsal(a):
In continuation of my last post here on I2C feeder design, which allows the flexibility to move feeders along an axis without the need for mechanically indexing them, here is a picture of my first working prototype without a sprocket. I hooked it up to a breadboard setup to test. My next plan is to get CNC'd PCBs. These PCBs are expensive at low volume and my hope is that others here may also be interested in this to help out in developing it. In doing so, I am wondering how many would like to join in on the first round of prototypes? I have most of the pieces already, just need to buy more motors to match the volume of PCBs and 3D print the cover of each feeder from PLA (which is a non critical part for dimensional accuracy). The sprocket will also be made out of 1mm thick PCB. This would include the motherboard to plug the feeders into as well. All the feeders will be SMT assembled, but ship mechanically un-assembled (to save space in shipping box). The feeder is designed to work with all tape types shown on this guide: http://www.vishay.com/docs/60034/tapereel.pdf This includes proper min bend radius of wider tapes. This first batch will only be for 8mm wide tapes, but the feeder is designed to be expandable to wider tapes by swapping the standoffs and dowels to longer ones. The entire setup runs on Arduino code and uses two optical sensors. What still needs to be figured out is the mounting of the feeders to the motherboard. Currently there are two holes on top and under the 5 pin spring connector as you can see, but these don't work well. The idea was to have it clip onto two rods, but PCBs don't flex well enough to make it work. Any suggestions appreciated.

How many people here would be interested in getting these feeders in classic green color and what is a price point that works for you so I can calculate the volume required to meet that price point? Please also let me know quantity you would be interested in. My hope is to get enough people interested so we can lower the price point with a group buy. Note that these have only been mechanically tested, but the code is ready and compiles.

Michael Anton

unread,
Jan 18, 2020, 6:25:43 PM1/18/20
to OpenPnP
I think that you can drive your larger gear directly with the worm, or at least that's what I was doing in the design I worked on, and it looks like it should work ok, even if it isn't ideal.  It's not like there is a huge load to get it to turn.  Like you, I was using easy to get off the shelf gears.  I've attached a 3D pdf of the version I was working on, which may give you some ideas.

At the very least, you should be able to drive the small gear with your worm, and skip the larger gear next to it, since it has the same modulus as the final gear.
Feeder Assembly.pdf

Shai

unread,
Jan 18, 2020, 6:48:51 PM1/18/20
to OpenPnP
Hi Michael,

I see how you did it. If I do it like that, I have to make the feeder wider. Currently my feeder is 10mm thick, which is very skinny and allows more feeders to be mounted.

Brynn Rogers

unread,
Jan 18, 2020, 9:28:09 PM1/18/20
to OpenPnP

The Tiny102 and tiny104 have a UART.      I looked and the ATtiny's have like 64 byte of ram and some smallamout of code space (1k to 8k).
I have gone all ARM,   for almost the same price you can get an arm M0 that has 1Kram minimum, and like 8k code minimum, Way nicer to use...
NXP  MKL02 or something   0.50 each for some variations...

And please do not use I2C.   

Brynn

Shai

unread,
Jan 19, 2020, 10:44:30 AM1/19/20
to OpenPnP
One option is to use a fuse holder such as this: https://www.digikey.com/product-detail/en/keystone-electronics/3622/36-3622-ND/2745822 so the feeder clips onto 5mm diameter linear rods. However, this means downsizing from the original 6mm rods I thought of using, which seem strong. Not sure how strong 5mm linear rods are, but of course you can probably add 3D printed supports along the path to strengthen it, although not ideal since you lose feeder space. But these are the only fuse holders that fit inside an 8mm tape feeder. The next size up is 6.35mm fuse holder and those are too wide to fit into an 8mm feeder.

Shai

unread,
Jan 19, 2020, 1:50:52 PM1/19/20
to OpenPnP
I've gone ahead and modified the feeder design to include these fuse clips. My main concern is the solder would break after multiple uses, so I cut 5mm slots into the PCB to guide the feeder perpendicular to the linear rails. There's a hole in the center of this fuse so that solder can protrude through for better bonding, perhaps that would help. There's also no cycle count on these fuses, but considering it's made out of a strong material, I think it will last long. See attached pictures. Thoughts? Very "tacky" but it should work and it's simplest/inexpensive design I can think of... still open to ideas :)


screenshot_171.png  screenshot_170.png



Shai

unread,
Jan 19, 2020, 2:17:20 PM1/19/20
to OpenPnP
Here's a better design actually to ensure the linear rods don't get pushed too far into the fuse holder and are guided more straight into the fuse holder. Another concern I realized would be to ensure the feeder isn't too heavy and the fuse holders can grab onto the linear rods well enough.

Only way to find out is to test :) If all else fails... I guess the best option is to just make a through hole and slide these feeders onto the rails and then mount it...

screenshot_172.png


Jason von Nieda

unread,
Jan 19, 2020, 2:23:42 PM1/19/20
to ope...@googlegroups.com
Hi Shai,

Are you intending to have the reels hang off the feeders, or will you have a separate reel holder? If the former, there is no way those fuse holders will bear the weight of most reels.

How about something like this instead:

IMG_7479.JPG

Sort of like a keyslot. You set the top entry point on the rod, so the weight is borne by by the stronger PCB or plastic, and then you use the fuse holder at the bottom simply to hold it in place.

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 view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/3b44de30-c7f8-4953-98e7-0c4ba1edaf39%40googlegroups.com.

Shai

unread,
Jan 19, 2020, 3:12:04 PM1/19/20
to OpenPnP
Hmm, good point. I completely missed the opportunity to do that! I currently don't plan to hang the reels on the feeders themselves, but perhaps in a future version. I will try to modify the design to do it how you suggested.

Shai

unread,
Jan 19, 2020, 3:58:00 PM1/19/20
to OpenPnP
Much better :)

screenshot_173.png


Marek T.

unread,
Jan 19, 2020, 4:09:48 PM1/19/20
to OpenPnP
Much better because keeps a bit better the feeder against loosing the x location.
However I'm not sure if this lower (on you picture) holder will hold it strongly enough.

Maple_Dude

unread,
Jan 19, 2020, 8:17:37 PM1/19/20
to OpenPnP
Looks interesting

Michael Anton

unread,
Jan 19, 2020, 9:09:25 PM1/19/20
to OpenPnP
Currently, mine is at 10mm as well, though in truth I was only trying to get 12mm, so I may have planned on having a thicker base to mount it on initially.  I was planning to use an aluminum PCB for the base of the feeder, but I don't require any routing which at least makes them fairly inexpensive.

Since I use a spring clamp to hold down the tape, I don't need to trap the tape on both sides, and the adjacent feeder forms the second side, so the feeder is narrower than it appears.  The large plastic part that supports all the assembly is only 8.4mm thick, which leaves 1.6mm of PCB to get 10mm in width.  

I planned to clip my design onto 20mm T-Slot extrusions, using a clip similar to what Jason proposed, but from the side rather than the bottom.  So, the front clip catches the front of the extrusion, and then the rear clip locks it into place.

Given that the tape can be up to 8.3mm wide when you include tolerances, that means that the skin on each side of your tape is under 1mm?

Shai

unread,
Jan 20, 2020, 10:59:18 AM1/20/20
to OpenPnP
Michael,

So I plan to cut a depth of 1.1mm. The manufacturer told me their tolerance is 0.15mm for each direction. Is it actually that? Who knows, this is a factory in china and I doubt their sales people even know their specs too well (from experience). So the total width for a tape, I planned to be at 8.2mm give or take. If you were to buy some old tape from aliexpress, yea it might be too thick or wide or low quality and may not work. I've tried a few and some are a bit harder to push on a 3D printed resin sample, but then again my 3D printed resin sample wasn't firmly flat like a PCB would be. I think the only way of knowing if this will work is trying prototypes, but prototyping a milled PCB is very expensive when you don't have a CNC to do it in house and need to send it out :)

Shai

unread,
Jan 20, 2020, 11:03:44 AM1/20/20