New 3D-printed cut-strip feeder for OpenPNP - the BlindsFeeder

1,667 views
Skip to first unread message

ma...@makr.zone

unread,
Jan 18, 2020, 10:18:23 PM1/18/20
to OpenPnP
Hi

I've created the new "BlindsFeeder". Check it out here:
https://github.com/openpnp/openpnp/pull/936

BlindsFeeder-Open-Close.gif



And here:
https://youtu.be/dGde59Iv6eY

Cheers,
_Mark



ma...@makr.zone

unread,
Jan 18, 2020, 10:21:29 PM1/18/20
to ope...@googlegroups.com

Ok, the gif as anattachment?

_Mark

BlindsFeeder-Open-Close.gif

SMdude

unread,
Jan 18, 2020, 10:47:48 PM1/18/20
to OpenPnP
Nice job Mark! I like it!
Simple, minimal plastic and you can unload one project and load another without worrying about all your passives jumping out of their pockets.

Doppelgrau

unread,
Jan 19, 2020, 10:21:39 AM1/19/20
to OpenPnP
Nice work. I like it and I think I'll try to use it too.
A short question: when using the push cover, is it pulled for closing or pushed? (If pulled the cover can be pushed over areas not reachable).

Shai

unread,
Jan 19, 2020, 12:20:48 PM1/19/20
to OpenPnP
This is very cool! Do you close the blind with the part on the nozzle after you pick it? I couldn't tell from the video. If yes, wouldn't the part fall off? 

ma...@makr.zone

unread,
Jan 19, 2020, 2:11:07 PM1/19/20
to ope...@googlegroups.com
Am 19.01.2020 um 15:15 schrieb Doppelgrau:
Nice work. I like it and I think I'll try to use it too.

Cool!

A short question: when using the push cover, is it pulled for closing or pushed? (If pulled the cover can be pushed over areas not reachable).

The push-cover cannot be closed by the machine, it opened step by step as the parts are used up.

_Mark


Airhead Bit

unread,
Jan 19, 2020, 4:16:41 PM1/19/20
to OpenPnP
Funny, I was going to use the feeders Erich Styger made for his OpenPnP machine, yours have more flexibility. I don't have OpenSCAD installed but was wondering how long these are, I have those books of 'sample' parts with 100mm strips.
Great job on this project by the way.
Kevin A

Airhead Bit

unread,
Jan 19, 2020, 4:38:22 PM1/19/20
to OpenPnP
That took about a minute, 84mm and by changing tape_length=84 to 124 done! I guess that by playing around I can get 0805 in there...
Thanks!

On Sunday, January 19, 2020 at 2:16:41 PM UTC-7, Airhead Bit wrote:
Funny, I was going to use the feeders Erich Styger made for his OpenPnP machine, yours have more flexibility. I don't have OpenSCAD installed but was wondering how long these are, I have those books of 'sample' parts with 100mm strips.
Great job on this project by the way.
Kevin A
---------snip

ma...@makr.zone

unread,
Jan 19, 2020, 5:44:52 PM1/19/20
to ope...@googlegroups.com

Hi

No the feeders are only opened automatically, they aren't closed automatically. You need to do a "Close all Covers" for that.

The assumptions is, that more than one part is needed, more than one job is run, etc. You only want to close up shop, once you're really done.

I was contemplating adding a handler to the "Machine Disabled" event, but I think that might be too dangerous, as sometimes you want to disable the machine after a non-recoverable mishap (stalled stepper etc.).

_Mark

bert shivaan

unread,
Jan 19, 2020, 5:50:33 PM1/19/20
to OpenPnP
airhead, was that a minute after installing openscad? Does that imply it is dead simple to install and get things running?

--
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/9be36449-3b7a-48cb-6b27-df7523982b7b%40makr.zone.

ma...@makr.zone

unread,
Jan 19, 2020, 5:54:25 PM1/19/20
to ope...@googlegroups.com

Glad, you figured it out so quickly.

The length and also the number of tapes are only limited by the build volume of your printer.

You can also change how the trays and the covers are arranged on the print bed, so if you're limited by the arrayed tapes, but there is still space in tape length, you can switch covers_to_the_right to true.

If this is till not enough, you can generate the model twice, once for the trays, once for the covers, with the tray and cover switches and print separately.

Having said that, there might be limits due to accumulated errors in the tapes and/or in the print.

_Mark

--
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.

SMdude

unread,
Jan 19, 2020, 6:15:40 PM1/19/20
to OpenPnP
I found openscad really easy to modify an existing model.
Install it and take a look. There will be a list of parameters you can modify for length width etc.
If you want to delete features out to f the design you can just comment them out in the code.


On Monday, 20 January 2020 09:50:33 UTC+11, cncmachineguy wrote:
airhead, was that a minute after installing openscad? Does that imply it is dead simple to install and get things running?

On Sun, Jan 19, 2020 at 5:44 PM ma...@makr.zone <ma...@makr.zone> wrote:

Hi

No the feeders are only opened automatically, they aren't closed automatically. You need to do a "Close all Covers" for that.

The assumptions is, that more than one part is needed, more than one job is run, etc. You only want to close up shop, once you're really done.

I was contemplating adding a handler to the "Machine Disabled" event, but I think that might be too dangerous, as sometimes you want to disable the machine after a non-recoverable mishap (stalled stepper etc.).

_Mark


Am 19.01.2020 um 18:20 schrieb Shai:
This is very cool! Do you close the blind with the part on the nozzle after you pick it? I couldn't tell from the video. If yes, wouldn't the part fall off? 

--
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 ope...@googlegroups.com.

ma...@makr.zone

unread,
Jan 19, 2020, 6:38:01 PM1/19/20
to ope...@googlegroups.com

OpenSCAD is very compact and easy to install.

Many features in the BlindsFeeder have bools to switch them on/off. So be sure to check, before commenting things out ;-)

Also you can let all your TapeDefinitions in there (and with time you will have many).

Just set the corresponding arrayed_tape_lanes entry to 0 to not generate the unwanted ones.

e.g.  if you don't want the tape12mm, just set the 1 on top of it to 0.

_Mark

Airhead Bit

unread,
Jan 19, 2020, 7:12:48 PM1/19/20
to OpenPnP
No, the same minute, Install OpenSCAD then open the file, look at the numbers and try 124!  I found the latest generation of i9/5 processors from Intel along with 2 Samsung 960 M.2 NVme SSD drives gets me 65 second compiles on the Merlin code. I can't say that OpenSCAD is simple, _Mark just set it up so it looks simple.  

On Sunday, January 19, 2020 at 3:50:33 PM UTC-7, cncmachineguy wrote:
airhead, was that a minute after installing openscad? Does that imply it is dead simple to install and get things running?
---snip

Airhead Bit

unread,
Jan 19, 2020, 7:19:16 PM1/19/20
to OpenPnP
Where did you get the number from for the different size (0402/0603) components? I need some 0805 and capacitors with varying heights. It was simple to figure out the way you had set it up, Thanks!
Kevin A.

On Sunday, January 19, 2020 at 3:54:25 PM UTC-7, ma...@makr.zone wrote:

Glad, you figured it out so quickly.

--snip

Alexander Goldstone

unread,
Jan 19, 2020, 7:20:10 PM1/19/20
to OpenPnP

In addition to the impressive work on the BlindsFeeder it looks like you have an interesting 'slates' concept for your base-plate. Can you please share some more information as to how you put this together?

bert shivaan

unread,
Jan 19, 2020, 8:27:12 PM1/19/20
to OpenPnP
Ok, I can be the dense guy in the room often, but need a bit of help here. Scad is waiting for me to open a file. The links provided seem to take me to a text file. Do I just save them as a .scad and open them? Is there specific names to call them?

--
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/e3bb6777-84d7-4b09-9e03-a4a881cba95f%40googlegroups.com.

bert shivaan

unread,
Jan 19, 2020, 9:00:10 PM1/19/20
to OpenPnP
never mind, got it
I have opened it now :)

ma...@makr.zone

unread,
Jan 20, 2020, 3:41:46 AM1/20/20
to ope...@googlegroups.com

Hi Kevin A. and everybody

Before I forget and as it was never actually mentioned in the video, be sure to print in green! For the "green screen" computer vision effect that is used extensively!

> Where did you get the number from for the different size (0402/0603) components?

All the main numbers are usually multiples of 4mm, so you can't go wrong even just guessing this from the physical tape.

Look at datasheets for parts that you don't have yet. There are generic datasheets like this:
https://www.analog.com/media/en/technical-documentation/white-papers/tape-and-reel-packaging.pdf

The tape_thickness, pocket_portrusion you must get from the datasheet or measure from the physical tape. Probably need a caliper. But don't worry too much, the printer can only resolve this in layers of e.g. 0.2mm and that's good enough.

pocket_width is the width the physical pocket across the tape. You can make it a bit larger as long as there is space. The model will make the blinds opening slightly larger.

tape_play, cover_play are empirical. Start from my examples. Be sure to set your filament diameter precisely in the slicer, then print small sample feeders. TIP: you can print a hole series of almost identical TapeDefinitions where you only vary the tape_play, cover_play values.

Play can only be tested with the tape loaded, because the cover is designed with a combination of sideways and downwards clamping forces in mind. Using double-sided tape on the full underside, tack the feeder on a flat hard surface (important, so it can't bend against the covers' spreading forces), load the tapes, peel off the cover foil, and then test how easy it is to move the cover. The cover must move quite easily but there should be no skidding after a fast push.

WARNING: if the cover really sticks, don't try on the machine. No point in ruining your nozzle tips. Forget lubricants, they might have side effects for soldering and it really should work without.

Note that PETG takes a day or so to really settle. The cover will slide much more easily after that. Also be aware that the longer the feeder is, the more friction the cover will get, so don't jump up from printing a tiny sample to filling up your print volume in one step. Also print multiple lanes to get a representative side-by-side tension.

These tape_play, cover_play are forgiving for 0603 and up. I hope my examples will mostly just work. For your 0805 "bricks", I expect you to get it right on the very first try.

Actually, everything is very easy for 0603 and up. You can just snap in the cover on top of the parts without danger of parts jumping out. Just ease in the cover edge with the blinds first, align the "closed" position and then press the whole cover in. Don't be afraid, you can twist and pinch a PETG cover a lot before it gives.

Having said that, things can be quite frustrating to get right for 0402 (so don't start there). The 0402 is so hard mostly because the cover must not only just "be there", but also press down just right on top of the tape to keep these incredibly tiny parts in their very shallow pockets. If this is not just right, the parts will be "rolled" into the air gap between cover and tape. It is quite a challenge to get the balance between this pressing-down and too much friction right. Expect to print a few :-)

My hope is that once people get it right for their printer model, they can reproduce 0402 feeders reliably, anytime. It works for me and my PRUSA i3 MK3. And I hope that the community will share back these tape_play, cover_play values and other tips so others with the same printer model can just get it right on the first try.

Everybody: If you have a printer other than PRUSA i3 MK3, please report back your findings, thanks!

If parts keep jumping out, I found that an anti-static coating on the feeder helps. This anti-static coating is also a must for ESD sensitive parts, I guess. 

_Mark
--
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.

ma...@makr.zone

unread,
Jan 20, 2020, 5:05:41 AM1/20/20
to ope...@googlegroups.com

Hi Alexander

While I strongly recommend the "slates" idea per se, I'm not happy with my implementation. The extrusions that I used as guides are too expensive, too hard to mount right and too much in the way of things. The two-layered slate bases are unnecessarily complicated to make. The idea to store the slates in the machine base is not a good one. When I designed it, I overlooked things like where spent tapes are supposed to go on automated feeders ;-).

https://makr.zone/design-of-the-machine-base/133/

I'll probably change this for a much lighter open frame design and store the slates away from the machine.

But the slates per se are a good concept!

They have to be heavy for 0402 work, to keep the vibrations down. The 1mm magnetic stainless steel top is probably overkill, though it adds the weight and it is nice to be able to go rough when scraping off double sided tape etc.. ESD wise it is also a good choice, I believe. I mounted the stainless steel on High Pressure Laminate (RESOPAL), so it ok to use liquids for washing. After some fails with glue, I just used double sided tape to mount the stainless steel on the HPL.

Other than that there is not much magic to it. Just find a way to attach these to the machine table easily and more or less repeatedly. Not much precision needed with the BlindsFeeder (+/-2mm). Make sure the surface of the slate (or better yet the surfaces of the tapes in their feeders) are more or less level with the PCB in Z for best computer vision precision.

For the future I'd like to create a solution (in OpenPNP or as a script) to group the feeders per slate and load/unload them as needed. Could even be QR code driven.

_Mark

bert shivaan

unread,
Jan 20, 2020, 6:23:25 AM1/20/20
to OpenPnP
Well after all these years I now want a 3d FDM printer. I have a SLA but of course it is not big enough for this. So couple quick questions:
1- a quick search on Amozon turns up at least 20 different green PETG filaments. None are called "signal green" that I found. How much will the shade matter? (I like the neon green :)
2- Would it be hard to assign numbers to the "trays" and have them "load" the parts when identified? This could be done simply with a fixed width binary scheme like holes that can be read, or a stick on label that we can read and tell openPNP the tray number. OpenPNP could also have specific info about the tray ie number of lanes, sizes in each lane , etc.
3- Any thoughts on Creality Ender 3 Pro? I can buy it locally for $209USD.

ma...@makr.zone

unread,
Jan 20, 2020, 9:39:28 AM1/20/20
to ope...@googlegroups.com
Am 20.01.2020 um 12:23 schrieb bert shivaan:
Well after all these years I now want a 3d FDM printer.

:-)

I have a SLA but of course it is not big enough for this. So couple quick questions:

1- a quick search on Amozon turns up at least 20 different green PETG filaments. None are called "signal green" that I found. How much will the shade matter? (I like the neon green :)

I think any vivid green will do. Neon colors sometimes appear as "yellowish" in cameras/under LED light, so I'd be cautious.

I am astonished how resilient the pipeline is. I developed the pipeline in my office under fluorescent and day light, months ago, I never had to change anything for moving under the (bad) LED light of the machine. I even coated the feeders with an anti-static that adds blackish-blueish smudges in nooks and cranies, and still no change required.  So I guess its not so important.

Caveat: I needed to leave the camera's auto-color balance on. So I need some more than minimal settling time because of that. The manual color balance of the USB driver does not help, it only shifts the hue, which is actually counter-productive.

I plan to add true RGB color-balance to OpenPNP one day.

2- Would it be hard to assign numbers to the "trays" and have them "load" the parts when identified? This could be done simply with a fixed width binary scheme like holes that can be read, or a stick on label that we can read and tell openPNP the tray number. OpenPNP could also have specific info about the tray ie number of lanes, sizes in each lane , etc.

I think I don't understand. Please elaborate on the actual use case.  Where information initially comes from and why it would later be missing in OpenPNP.

What I was thinking about is unloading and then reloading  whole slates with feeders. But I would then probably create one big QR code on a known scanning position on the slate, that contains all the feeders (including the parts) as a compressed XML fragment (the same XML that OpenPNP saves to the machine.xml).

Another idea is showing the parts loaded in the camera view, similar to how the pocket numbers are already displayed in the BlindsFeeder.

Ideally this would be a build-in feature of the camera view, available for all the feeders (and perhaps other objects on the table).

3- Any thoughts on Creality Ender 3 Pro? I can buy it locally for $209USD.

Sorry, I only ever had this one printer, the PRUSA i3 MK3, and while I can highly recommend it (it just works!), this is without having had any comparison, really.

_Mark


bert shivaan

unread,
Jan 20, 2020, 10:15:35 AM1/20/20
to OpenPnP
Let me try to explain.
Assume we have 4 trays (4 lanes as shown in your example)
so then tray 1 would have some capacitor,resistor,regulator, and uP
tray 2 has different stuff
tray 3 has same as tray 2
tray 4 has only 1 thing in lane 3
When we load the parts into openPNP, we would tell openPNP that cap0402-.1uF is in tray 1 lane 2

But we have to be able to define tray 1,2,3,4
So we could just assign the number when we set it up as in your video. This case the number is only read by us humans
or
we coild "print" the number in the side by the Fidicual using binary like 10 bit. open square =1, no square = 0
this could be read by the camera and openPNP would know.

In either case I am hoping we can save the data with what parts are in each tray so if we remove it from the machine, when we put it back it will remember.
Keep in mind I see this also as a way to store some hundreds of strips, maybe not on the machine. So a database/XML file or something will be needed to know what parts are in what tray

ma...@makr.zone

unread,
Jan 20, 2020, 10:46:18 AM1/20/20
to ope...@googlegroups.com
> But we have to be able to define tray 1,2,3,4
> So we could just assign the number when we set it up as in your video. This case the number is only read by us humans

There is the param version_tag in the OpenSCAD BlindsFeeder model. I used it as a version indicator and that made sense when things were moving fast, but now we could use it as just a generic tag for the feeder array or "tray" as you put it.

This tag is 3D-printed there:

It is only (barely) human readable, but you could then name your feeders that way (press F2 on the list entry):

I though quite a lot about encoding data on the feeder itself (i.e. all the TapeDefinitions) and dismissed it. It is not practical (too much information, too low resolution, too slow to print). You could only encode the data that doesn't ever change and that does not include the loaded parts.

Instead I would opt for a cheap label printer to print a QR code with the full data including the loaded parts and stick it to the slate that is carrying all the feeders. This way you could also load a whole slew of feeders in one go. Whenever the data changes (differnet parts loaded) the sticker could be reprinted.  The QR code could be printed very small/dense, as the camera is magnifying a lot.

_Mark

--

bert shivaan

unread,
Jan 20, 2020, 10:58:16 AM1/20/20
to OpenPnP
I didn't want to encode any information in the tray about parts. Just give the tray (array) a unique number that could be read by openPNP (maybe)
Mostly the work will be done inside of openPNP. So really what you are thinking for the slate, I want to do with the array. I don't have a slate system such as you, so it seems more likely to me I would be moving the arrays by themselves. But that brings me to how to attach them.
I was hoping to store the information you suggest on the densely printed label inside openPNP by array number, or in your case it would be slate/array/part.

I am not sure I am explaining this well at all :(

--
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.

E O

unread,
Jan 20, 2020, 11:04:07 AM1/20/20
to OpenPnP
QR code stickers might be easier than coming up with a printable tag that's able to be read reliably. Plus you'd need to regenerate the STL/gcode for each one you print.
To unsubscribe from this group and stop receiving emails from it, send an email to ope...@googlegroups.com.

bert shivaan

unread,
Jan 20, 2020, 11:15:30 AM1/20/20
to OpenPnP
My "tag" thought was just 10 bit binary, the 10 bits always in the same spot so the camera/openPNP would know when there was no hole it is a 1.
Maybe the "holes" can be made as punch outs so all the arrays can be printed the same and the user just punch out the correct bits? 10 bits would give us 1024 unique arrays. Or go 12 bits for 4K?

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/8eac2660-68f5-48cf-b722-a314f3970946%40googlegroups.com.

ma...@makr.zone

unread,
Jan 20, 2020, 11:19:55 AM1/20/20
to ope...@googlegroups.com

> Just give the tray (array) a unique number that could be read by openPNP (maybe)

I'll keep that in mind. Holes are very slow to 3D print, but maybe some edge indents could code it. A bit like a Vine Weevil on a leave :-)

_Mark

Jim

unread,
Jan 23, 2020, 10:28:55 AM1/23/20
to OpenPnP
Hi Mark,

I played with your brand new OpenPnP 2.0 feature and want to test it just for fun to see how it works.

Before printing the stl-file i would like to know what was the maximum length where cover could be opened and closed successfully before friction gets too high?

Would a tape_length of 204mm (= 50 parts/4mm pitch) still work or is too long?

Jim

Shai

unread,
Jan 23, 2020, 10:57:48 AM1/23/20
to OpenPnP
I just realized that this would probably work great with an aluminum extrusion for base and plastic extrusion for a cover.

ma...@makr.zone

unread,
Jan 23, 2020, 11:17:01 AM1/23/20
to ope...@googlegroups.com

Hi Jim

the longest I printed was ~160mm, if I remember correctly, but there was no indication that this is the limit.

I would guess that 0603 and above is no problem whatsoever. If if turns out too tight, you can increase the cover_play with no ill effects.

For the 0402 I would be cautious, because I also started to see geometry errors adding up in 100mm+ that start to matter, because everything must be sooooo precise as the nozzle tip really barely fits inside the blinds opening (and will fail the pick if it does not hit the opening precisely). But then again with 0402 you get ~50 parts inside 100mm, not so bad for prototyping. 

Note this is all "straight out of the box" yet, i.e. I never doubted (or even measured) whether the 3D printer is true to scale and I normally left the BlindsFeeder's "Normalize" switch on in testing, which means OpenPNP will go to "theoretical" coordinates rather than "stretch" and "shear" the feeder between the fiducials. So there might be more potential even for 0402.

_Mark

--
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.

ma...@makr.zone

unread,
Jan 23, 2020, 11:32:56 AM1/23/20
to ope...@googlegroups.com

Just be aware that the feeder is really custom-made for one type of tape. I guess it will work for all tapes of one manufacturer for the same package spec. Even between different manufacturers for common parts like chip resistors.

But there are differences between tapes of different part heights, because punched paper tapes and embossed plastic tapes do adapt to part height. This might be the case for the larger uF multi-layer capacitors, even for the same (small) footprint.

And the BlindsFeeder is not tolerant to varying tape thickness, at least not for 0402 and only limited for 0603. Never tried larger stuff yet.

Embossed plastic tapes might be more tolerant, as only the pocket depth changes with the part height, and perhaps a pick on a free-hanging pocket will work (if the feeder Z is set precisely).

_Mark

Jim

unread,
Jan 23, 2020, 11:50:03 AM1/23/20
to OpenPnP
Thank you Mark,

With my $165 printer i can print up 210x210mm, so 204+2mm cover gap would have place on the build plate.

I already dismounted some of my milled manual strip feeders to have some space for testing.

stripfeeder.jpg


Hopefully i'll get tomorrow my first green pla and then the fun can begin ;-)

Jim

ma...@makr.zone

unread,
Jan 24, 2020, 2:20:58 AM1/24/20
to ope...@googlegroups.com

> Hopefully i'll get tomorrow my first green pla and then the fun can begin ;-)

Be aware that PLA is more brittle than PETG. I tried once with PLA and one cover immediately cracked at one place (never did that with PETG, for many dozens of prints). However the cracked little PLA guy still worked!

So report back your experience with PLA, thanks.

_Mark

Am 23.01.2020 um 17:50 schrieb Jim:
Thank you Mark,

With my $165 printer i can print up 210x210mm, so 204+2mm cover gap would have place on the build plate.

I already dismounted some of my milled manual strip feeders to have some space for testing.

Jim

unread,
Jan 24, 2020, 4:51:09 AM1/24/20
to OpenPnP
On Friday, January 24, 2020 at 8:20:58 AM UTC+1, ma...@makr.zone wrote:

> Hopefully i'll get tomorrow my first green pla and then the fun can begin ;-)

Be aware that PLA is more brittle than PETG. I tried once with PLA and one cover immediately cracked at one place (never did that with PETG, for many dozens of prints). However the cracked little PLA guy still worked!

So report back your experience with PLA, thanks.

_Mark

 
Hello Mark,

Oh No - i confused that two types.
Used always the cheapest PLA i could get because it is so easy to print and it worked for me in most cases (mechanical robustness).
PETG, ABS, Wood etc. i never tried before - i fear that my old v5 head will not handle it very well.

That could now be a reason for a new printer purchase...

Jim

Marek T.

unread,
Jan 24, 2020, 6:16:01 AM1/24/20
to OpenPnP
PETG no problems if you can PLA.
ABS is another story but it goes about the closed chamber with stable temp conditions nott the nozzle. AFAIK.

ma...@makr.zone

unread,
Jan 25, 2020, 4:21:52 PM1/25/20
to OpenPnP
Hi

Added the Wiki page for the BlindsFeeder:


Linked from here:


Critical comments and English language help welcome.

_Mark



Jed Smith

unread,
Apr 13, 2020, 7:18:29 AM4/13/20
to OpenPnP
I've been experimenting with using a BlindsFeeder to augment my slowly growing reel feeder bank and seem to be having a problem.

Modifying the OpenSCAD parameters and printing it out went smoothly. I'm getting better at quickly running through the OpenPNP setup (intermittent user to keyboard interface glitches). And the tests where I run one part at a time work well.

But when I start running small test jobs I've been encountering issues. Namely it uses the first nozzle to slide open the cover, then picks up a part with that same tip. Then it goes to the next row, uses the first nozzle which dislodges the part to open the cover, while using the second nozzle to pick up the second part. I am using Peter's dual head, so I have two nozzles, and have auto tip changing setup.

I've currently configured all of my tips to be able to slide open the covers, but I don't know how to enable both nozzles to be able to slide covers back. Some of the rows don't have the blinds holes and use the solid covers. Am I missing something in how I have it configured or was the code not setup for dual nozzle heads?

ma...@makr.zone

unread,
Apr 13, 2020, 8:59:04 AM4/13/20
to ope...@googlegroups.com

Hi Jed

Ouch that's a classic bug!

My machine has only one nozzle, I tried hard to make it work with multi-nozzle machines but its tricky to do it all in pure theory. Sorry.

I must check the nozzles whether they have a part on, and I must also favor the nozzle that is compatible with the parts in the feeder. Unfortunately in feeder.feed() I don't get the nozzle that will eventually pick the part, so this will not be 100% conclusive, if multiple nozzle tips are compatible and without part. If this happens it will do some extra moves. 

Will try to create a bugfix PR fast.

Thanks for reporting this and I appreciate your ongoing help in testing it with multi-nozzles. :-)

_Mark

--
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.

ma...@makr.zone

unread,
Apr 13, 2020, 10:53:49 AM4/13/20
to ope...@googlegroups.com

Made a PR to fix this:

https://github.com/openpnp/openpnp/pull/979

Once its made available, I would be very grateful for you to test it, Jed. Thanks!

_Mark

Jed Smith

unread,
Apr 13, 2020, 11:52:32 AM4/13/20
to OpenPnP
Looking forward to testing this and will let you know. Thanks!

I'm impressed you got it as close as you did without being able to test the code on dual nozzles. Wish I normally got that close on the first try :)

ma...@makr.zone

unread,
Apr 18, 2020, 2:38:54 PM4/18/20
to OpenPnP
Hi Jed

Jason just merged my fix into OpenPNP 2.0, so I think this is ready for testing again. It should now always favor the nozzle that is also used for picking, so no extra moves are needed (if the nozzle tip is enabled for push&drag).

Thanks for testing this, Jed.

Ref:

_Mark



Kevin Kirmse

unread,
Apr 18, 2020, 8:00:15 PM4/18/20
to OpenPnP
I am working with the BlindsFeeder and I having a problem with setup.

Under the location group in the setup if I use the "Capture the location that the camera is centered on" button I get a Java Exception.
The "Capture the location the tool is centered on" button works fine.


OpenPNP_Java_Exception.PNG



ma...@makr.zone

unread,
Apr 19, 2020, 4:56:43 AM4/19/20
to ope...@googlegroups.com

Hi Kevin

> I am working with the BlindsFeeder and I having a problem with setup.

Cool! The good news is that problem you encountered is not related to the BlindsFeeder. :-)

It seems you have set a Z probe Actuator on your machine head. It reads a Value from the Z probe and tries to convert it to a number, which fails.

If you really have a Z probe, you need to properly set it up (i.e. the read regex). If you don't, please remove the probe actuator from the head.

The NullPointerException should probably be replaced by a proper Error Message. I'll have a look.

_Mark

Pekka Roivainen

unread,
Apr 19, 2020, 6:08:37 AM4/19/20
to OpenPnP
Hi!

I have a problem facing issues with getting the scad files from OpenPNP. I am running on Linux, Debian Buster, and when I press the "Extract 3D-Printing Files", I get a promt for the save directory, but after choosing the directory, I get "The EDIT action is not supported on the current platform!" error popup. The BlindsFeeder-Library.scad file is generated on the chosen directory, but not the other one. A quick googling shows that those edit, open etc actions in Java appear to work well only in Windows.. Is there a way to get the files from some other place? Git Repo maybe?

Thanks,
Pekka


On Sunday, January 19, 2020 at 5:18:23 AM UTC+2, ma...@makr.zone wrote:
Hi

I've created the new "BlindsFeeder". Check it out here:
https://github.com/openpnp/openpnp/pull/936

BlindsFeeder-Open-Close.gif



And here:
https://youtu.be/dGde59Iv6eY

Cheers,
_Mark



Pekka Roivainen

unread,
Apr 19, 2020, 6:13:31 AM4/19/20
to OpenPnP
Hah. Just after posting this message, decided to do one more google search and found links to the files on the original Pull Request https://github.com/openpnp/openpnp/pull/936.

Maybe worth adding to the OpenPNP wiki as well if the link in OpenPNP is difficult to fix on Linux.

BR,
Pekka

ma...@makr.zone

unread,
Apr 19, 2020, 6:35:10 AM4/19/20
to OpenPnP
Hi Kevin

Added a PR.


With proper Error Message:


_Mark

ma...@makr.zone

unread,
Apr 19, 2020, 7:07:18 AM4/19/20
to ope...@googlegroups.com

Hi Pekka

Make sure to get the newest "develop" version which is (always) here:

https://raw.githubusercontent.com/openpnp/openpnp/develop/src/main/resources/org/openpnp/machine/reference/feeder/BlindsFeeder-Library.scad
https://raw.githubusercontent.com/openpnp/openpnp/develop/src/main/resources/org/openpnp/machine/reference/feeder/BlindsFeeder-3DPrinting.scad

The one linked in the PR might have been be outdated (not sure).

I have updated the PR and the Wiki with these links, thanks for the hint.

I will change the way the files are extracted, to make sure they were both extracted, even if the "EDIT"command then fails.

_Mark

--
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.

ma...@makr.zone

unread,
Apr 19, 2020, 7:44:39 AM4/19/20
to ope...@googlegroups.com

Hi Pekka

it will now always extract the Files and then show this:

 

when the Desktop command failed.

Pull request:

--> https://github.com/openpnp/openpnp/pull/990

_Mark

Pekka Roivainen

unread,
Apr 19, 2020, 4:46:57 PM4/19/20
to OpenPnP
Thank you Mark for your quick response!

First feeders printed and after few trials, they seem to work nicely, covers slide with just nice friction on them etc. No OpenPNP tried yet.. I made the wall structures just a tiny bit stronger than on your default config. The biggest problem for me was the cutout for the thorns in the bottom of the cover. The area between the cutout and the edge of the cover is so narrow that it becomes a single line of extrusion on the first layer. And I have always had issues getting those to stick to my print bed. Everyone keeps telling me that PETG sticks like crazy, but for me, in the case of these very narrow features, it has never been the case. But, it is a problem of my printer, not the model for sure and for now I solved it by removing the cutout. Paper tapes are thick enough that the thorns are not protruding them anyway..

Thanks again, this will probably be my cut tape solution in the future.

BR,
Pekka
To unsubscribe from this group and stop receiving emails from it, send an email to ope...@googlegroups.com.
--
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 ope...@googlegroups.com.

ma...@makr.zone

unread,
Apr 19, 2020, 5:24:37 PM4/19/20
to ope...@googlegroups.com

Cool Pekka, nice to hear.

Have you tried 3D-Printing-Spray? It was the key to success for me. Zero problems since!

I tried all the glue sticks and whatnot and this was just a mess.

_Mark

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/2a73cb30-8666-43ed-9d9e-a5afd4e4b242%40googlegroups.com.

Marek T.

unread,
Apr 19, 2020, 6:58:27 PM4/19/20
to OpenPnP
Dimafix, C-stick, what are you using Mark? And still heating the bed or do you print on cold?

ma...@makr.zone

unread,
Apr 20, 2020, 2:01:42 AM4/20/20
to ope...@googlegroups.com

I was using the stick provided by Prusa. And once with that blue tape. No good. I'm always heating the bed as preset in the default profiles, sometimes adjusted to what the filament manufacturer says. And I'm cleaning with 99% Isopropanol a lot. Sometimes the spray gets accumulated and then I just wash the removable plate with warm water with some fluid soap, then dry and then Isopropanol again.

If you can remove the plate, spray it away from the machine. I didn't at first and the spray then caked the rails etc.

_Mark

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/0154167e-6f55-4332-b531-231a0fe1c9d7%40googlegroups.com.

Jonathan Oxer

unread,
Apr 20, 2020, 3:48:26 AM4/20/20
to OpenPnP
I found that the magic trick with PETG is to run the first layer very slowly. I turn my printer down to about 30-40% for the first layer, and then turn it back up again. Doing that has made PETG 100% reliable for me, it sticks every time even with fine details.

Jon

Marek T.

unread,
Apr 20, 2020, 4:07:18 AM4/20/20
to OpenPnP
I often suffer with problem that some single corners of printed parts don't keep the bed glass good enough. They lost contact sometimes when the nozzle is already at z 3-4mm. But it's still first layer holding problem.
Few days ago I have find custom edition of slick3r with ironing and option of ears instead of brim - and it works much better for me.

Michael Anton

unread,
Apr 20, 2020, 4:45:04 AM4/20/20
to OpenPnP
That's actually commonly done with most filaments on the first layer, and is why this option even exists in slicers.  It's so common in fact that I'd probably not have thought about suggesting it if someone had problems.

Pekka Roivainen

unread,
Apr 20, 2020, 5:09:42 AM4/20/20
to OpenPnP
It happens that I received a sample pouch of a adhesion solution together with my last filament order (3DJake EasyAdheasy). I didn't try it so far, as there is not too much info about using them with PETG online. Everyone just says that PETG needs something to help removing it from the plate.. But I went ahead and applied that to my bed and Wow! Just Wow! :) Every single feature of the print sticked very well. I already print the PETG first layer ridiculously slowly (12mm/s) and that for sure is a key factor, but hasn't just been enough for me. I also get the PETG curling up after few mm of print height sometimes. Let's see if this magic stuff will help in that as well. That EasyAdheasy is available in a bottle with built-in sponge, so no spray. Which is good, as I don't have a removable surface.

So thanks for the suggestion Mark, it really made my day.

Pekka

ma...@makr.zone

unread,
Apr 20, 2020, 5:13:18 AM4/20/20
to ope...@googlegroups.com

Yeah, bottom layers come with veeery slow speeds configured in the PRUSA standard profiles.

For larger parts, like the PushPullFeeder, I was able to more than double that speed. But for the BlindsFeeder I would not dare do that.

The more I read about all these discussions, the more I appreciate how everything just works with PRUSA printers and the whole software and profile package.

The first printer I bought assembled, but I recently assembled a second Kit printer mostly done by my girls as a "lockdown project". Worked just as flawlessly from the first try. And it's all done with simple parts, no fancy stuff. Y rails are zip-tied, if you believe it. Miraculous in a way.

_Mark

--
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.

Kevin Kirmse

unread,
Apr 21, 2020, 3:00:24 PM4/21/20
to OpenPnP
Questions on the blindsfeeder system.

I have a board which uses more then one blindsfeeder prints. I have found that if I set up an additional feeder print the location will sometimes get copied to all blindsfeeder prints. How should I handle more then one print?

How does the blinds feeder handle a print that is not true square? My 3d printer appears to have to be skewed. I have used the fiducials to calibrate but the calibration changes the location of the top most fiducial. It is changed on the order of a mm. The parts at the top of the feeder do not line up when I use the "Move the camera to the selected feeder's current part location". This happens whether on not I have "Normalize" checked. The parts at the top of the feeder print appear to be positioned properly compared to the fiducial at the top of the print.

ma...@makr.zone

unread,
Apr 21, 2020, 3:58:08 PM4/21/20
to ope...@googlegroups.com

Hi Kevin

The BlindsFeeder system automatically creates arrays (groups) of feeders that are on the same 3D print.

First make sure you define the first feeder completely i.e. all three fiducials must be set and Apply must be pressed. Now the system knows the area the array covers.

Create a second feeder of the same array. Move the camera above one part in the tape and press Auto Setup. It will know that the new feeder lane belongs to the same array as the already defined first feeder, because of the same area it occupies inside the fiducials.

Watch this example:
https://youtu.be/dGde59Iv6eY?t=485

Create a third feeder of a different array. As long as the fiducials do not occupy the same area as the first feeder, this will create a new array. Again make sure to set all three fiducials and press Apply. Etc.

Sometimes not pressing Apply will create confusion, so I recommend pressing it a lot. Only when you apply a location, will the automatic work! Note that Jason is currently working on a change in OpenPNP that gets rid of Apply.


>How does the blinds feeder handle a print that is not true square? My 3d printer appears to have to be skewed.

> This happens whether on not I have "Normalize" checked.

You are right, switching off "Normalize" should deal with this and again don't forget to press Apply. But the skewed transformation will only come into effect when you press "Calibrate Fiducials" after switching off "Normalize" (or if calibration is later triggered by homing or by a job starting, etc.).

I have never really tested skewed coordinates, so it may be possible there is a bug. If this still does not work, please stay on it and report your findings, so I can fix it for you and everybody else!

I'm very interested in this use case, because it's cool to use brains to cope with simple and affordable tools, that's why I made it!

_Mark

Harjit Singh

unread,
Apr 21, 2020, 8:58:15 PM4/21/20
to ope...@googlegroups.com
Tongue in cheek/half serious: another application for the affine transform method?

--
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.

ma...@makr.zone

unread,
Apr 22, 2020, 5:47:54 AM4/22/20
to ope...@googlegroups.com

Hi

The BlindsFeeder does use AffineTransform, of course ;-)

https://github.com/openpnp/openpnp/blob/46edeafd3bb0a89733b7363a79cdd799433a2534/src/main/java/org/openpnp/machine/reference/feeder/BlindsFeeder.java#L1432

https://github.com/openpnp/openpnp/blob/46edeafd3bb0a89733b7363a79cdd799433a2534/src/main/java/org/openpnp/machine/reference/feeder/BlindsFeeder.java#L1486-L1490

And - also tongue in cheek - that's why I suggested to Tony, in his cool generalized location transformation feeder groups pull request, to allow for each feeder to have their own nested transform so they can also use full AffineTranform to allow for shear (and now we have a real use case too)

;-) 

It's a bit buried in this this comment.

_Mark

bert shivaan

unread,
Apr 22, 2020, 9:31:08 AM4/22/20
to OpenPnP
Hey Mark, I have been working on printing one of these for a few hours now. (cumulative time)
Looking at your pictures, it seems like mine does not melt together to become as smooth as yours. Heck you can even see the facets in the "spikes".
What head temperature do you use for PETG?
I know different brands/spools are different, just wondering where you are.
Also can you look to tell me your retention settings? I see it looks as if you have NO strings when you print. 

ma...@makr.zone

unread,
Apr 22, 2020, 9:59:25 AM4/22/20
to ope...@googlegroups.com

Hi Bert

I have different PETG and those vary a lot regarding temperature. The green stuff is from Extrudr and it uses significantly lower temperatures than all the other brands I have.

I start from the "Generic PETG" profile from PRUSA then look at the spools and set the recommended temperature. If it is a range, I go ~2/3 high. Then sometimes I go down when I have excessive stringing. Then sometimes back up, when the bottom layer does not fuse right anymore (then I just cope with stringing).

Extrudr's is really very nice all around, except for diameter tolerance, where it sucks. For those small feeders its not such a problem, just measure the diameter before each print and enter in the Slicer.

These are the two temperature extremes, one for Signal Green from Extrudr:

and the other for White from 3D Jake (for the PushPullFeeder body):


> Also can you look to tell me your retention settings?

Is this the same as "retraction"?

I haven't changed anything there for the BlindsFeeder. This is the default setting for the printer:

There is also no override on the Filament.

For the PushPullFeeder I have switched the hidden retractions off:

Hope this helps and good luck with the BlindsFeeders!

_Mark

bert shivaan

unread,
Apr 22, 2020, 10:10:26 AM4/22/20
to OpenPnP
Thanks Mark, for me everything helps. Complete newbie here so trying to sort out setting issues vs user error - most are the latter for sure.

--
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.

Marek T.

unread,
Apr 22, 2020, 10:26:31 AM4/22/20
to OpenPnP
Official slick3r PE is not too stable in my oppinion, and people who manage it on git are often quite "hard-head" to understand quite obvious problems and fix them.
The most annoying are operations made while printing that are not visible in the simulation, like printing some crazy brim acros half of the bed and back to the object. It's hard to accept for me it's living with its own life

Unofficial Slick3r++ is much better in my oppinion. Many bugs fixed, many new options or parameters added including ironing or ear-brims. And Supermerill ready to hear what people say.
To unsubscribe from this group and stop receiving emails from it, send an email to ope...@googlegroups.com.

Kevin Kirmse

unread,
Apr 22, 2020, 10:44:16 AM4/22/20
to OpenPnP
I recreated my Blindsfeeders with a better understanding on how the system works. I was able to make use of several feeder prints.

I did some further work with the skewed print.
Normalization was off when I opened the feeder.
Bottom left is fiducial 1.
Bottom right is fiducial 2.
Top left is fiducial 3.
I ran fiducial calibration.
The top left fiducial moves right by about 1.5mm after calibration.
This causes the feeders at the top of the print to be out of alignment.
The feeders work properly if I manually set the fiducials.

ma...@makr.zone

unread,
Apr 22, 2020, 12:12:10 PM4/22/20
to ope...@googlegroups.com

Hi Kevin

thanks! This sounds like a bug. It seems the Blindsfeeder will normalize the fiducials even if you tell it not to (when Normalize is off). I will investigate.

_Mark

-

bert shivaan

unread,
Apr 22, 2020, 9:52:46 PM4/22/20
to OpenPnP
another printer question, what size nozzle?

--
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/068cb32d-2c1c-70ec-2f1b-e9633e40a354%40makr.zone.

ma...@makr.zone

unread,
Apr 23, 2020, 3:19:32 AM4/23/20
to ope...@googlegroups.com

Standard 0.4mm

Btw, you see this and more on the "Printer Settings" screenshot I posted earlier. Plus in the video here:

https://youtu.be/dGde59Iv6eY?t=93

_Mark

John deGlavina

unread,
Apr 23, 2020, 9:36:29 AM4/23/20
to OpenPnP
Hey Mark, 

Any good idea on what to do when the pocket_portrusion is high enough for the camera to be out of focus looking at the fiducial vs the part? Everything on my bed is 10-11mm high.


On Saturday, January 18, 2020 at 10:18:23 PM UTC-5, ma...@makr.zone wrote:
Hi

ma...@makr.zone

unread,
Apr 23, 2020, 3:59:59 PM4/23/20
to ope...@googlegroups.com

Dig a hole into the table?

Is it the parts that are so high, or is the table higher than the PCB?

If the latter, that's not so god IMHO. Maybe consider moving the PCB up too and refocus the cameras, yes that's both cameras, bottom vision focal plane should also be on the same height.

It helps a lot with precision at least if your machine is "mortal" i.e. not super perfect, mechanically.

_Mark

Matt Odin

unread,
Apr 29, 2020, 2:22:12 AM4/29/20
to OpenPnP
Hello,

Aren't you worried about bending your nozzle tip, especially if you then try to accurately place 0402 or 0201 components with the same nozzle? 
matt

ma...@makr.zone

unread,
Apr 29, 2020, 11:11:53 AM4/29/20
to ope...@googlegroups.com

Hi Matt

First it is important to note that the BlindsFeeder is optimized to use a nozzle tips other than the pick nozzle tip for opening the cover. So if you are concerned, just use one of the heavy duty nozzle tips or you could even 3D-print a specific "push" nozzle.  Each nozzle tip has a Checkbox where you can allow/disallow Push & Drag.

At the Job start, the BlindsFeeder will automatically open all the covers of all the feeders that are involved in the Job, so even if you use a special nozzle tip, only one extra nozzle tip change is needed. It will even use a Travelling Salesman optimization to choose the best path through all these feeders.

Having said that, no, I'm not concerned. The load on the nozzle tip is very small. Apparently, other people use their nozzle tips even for drag feeding, which is a much higher load and much easier to go wrong.

You can read this conversation and perhaps ask John/CircuitSetup

https://github.com/openpnp/openpnp/pull/812#issuecomment-565803849

_Mark

--
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.

eSlavko

unread,
May 11, 2020, 4:22:35 AM5/11/20
to OpenPnP
I just wonder how this work on small machines.
So I have for example 3 arrays and just two can be in machine in same time.
How to enter/solve/callibrate that as they share same area on PnP.
As example..

Can I put feeder 1 into place 1 on machine and callibrate all lanes. (this should work of course)
After callibrating can I replace feeder 1 with feeder 2 in same area on machine and calibrate lanes?
same to third feeder.
So if I can do that then I assume that when I start job with feeder 1 and 2 on machine and in mid of job I can replace feeder 1 with feeder 3 and just enable feeder3 and dissable 1?


ma...@makr.zone

unread,
May 11, 2020, 7:02:32 PM5/11/20
to ope...@googlegroups.com

Hi Slavko

no, that is currently not possible without some scripting around it. The BlindsFeeder works by the fiducials and will "fuse" feeders that occupy the same area. You'll have a mess.

_Mark

--
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.

Slavko Kocjancic

unread,
May 12, 2020, 2:36:50 AM5/12/20
to ope...@googlegroups.com
So the feeder is defined by area (fiducials) and not the name and area.

I was thinking to make place for 4 feeders on machine and make bunch of
them and then just change as is needed and enabled/dissabled current
one. Good to know that this is not possible in current state.

The other my idea was that I attach some qr code to each feeder with
namo or number. And to force some script to search the feeder locations
on machine and automatically enable/dissable feeder.

(I have same idea with strip feeders as I taped strips onto piece of
MDF. The idea was to tape fiducials and qr code onto that mdf and at
start just search wich mdf (sets of strip) are present on machine.)
>> <mailto:openpnp+u...@googlegroups.com>.
>> <https://groups.google.com/d/msgid/openpnp/1748069e-a7de-4e73-8369-e05bbf7c6724%40googlegroups.com?utm_medium=email&utm_source=footer>.
>
> --
> 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
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/openpnp/4ab676bb-4082-5dc1-b949-2cd38bfe7503%40makr.zone
> <https://groups.google.com/d/msgid/openpnp/4ab676bb-4082-5dc1-b949-2cd38bfe7503%40makr.zone?utm_medium=email&utm_source=footer>.

ma...@makr.zone

unread,
May 12, 2020, 5:27:00 AM5/12/20
to ope...@googlegroups.com

Slavko,

First: as a manual workaround, you can shift a whole array of BlindsFeeders by just changing one fiducial coordinate manually and press Apply. The BlindsFeeder will automatically translate all the other feeders on the same array (same fiducial area). All geometry is preserved. The larger your array the more time saved.

Assuming your machine has less than 1m Y working area, I suggest subtracting -1000mm from the Y coordinate for project 1, -2000mm for project 2 etc. So you can quickly move in or out the feeder arrays from your work area and you still see what is going on. Just think hard before pressing Apply and/or frequently backup your machine.xml :-)

Don't forget to disable/enable the feeder accordingly.

Any don't forget to set the soft limits on your machine so if you mistakenly feed from this feeder, it will just warn you and not crash.

Someone was also mentioning Scripts to load and unload feeders in bulk from the machine. Please search in the group.

What you ultimately want, is one of the goals of this PR by Tony:

https://github.com/openpnp/openpnp/pull/943

The PR is rather "invasive" and therefore a discussion is going on, whether a more fundamental approach with a hierarchic Group -> Slot -> Feeder structure was in order:

https://github.com/openpnp/openpnp/pull/943#issuecomment-620976140

I also feel that way, it would be nice to see the feeders structured in the Machine Tree.

The point is: you need to wait and use scripts or the manual workaround instead.

_Mark

Slavko Kocjancic

unread,
May 12, 2020, 8:06:46 AM5/12/20
to ope...@googlegroups.com
On 12. 05. 20 11:26, ma...@makr.zone wrote:
> Slavko,
>
> First: as a manual workaround, you can shift a *whole array *of
> BlindsFeeders by just changing *one *fiducial coordinate manually and
> press Apply. The BlindsFeeder will automatically translate all the other
> feeders on the same array (same fiducial area). All geometry is
> preserved. The larger your array the more time saved.
>
> Assuming your machine has less than 1m Y working area, I suggest
> subtracting -1000mm from the Y coordinate for project 1, -2000mm for
> project 2 etc. So you can quickly move in or out the feeder arrays from
> your work area and you still see what is going on. Just think hard
> before pressing Apply and/or frequently backup your machine.xml :-)
>

If understand correctly if I misstype two feeders to same -1000 I wreck
setup for both?

> Any don't forget to set the soft limits on your machine so if you
> mistakenly feed from this feeder, it will just warn you and not crash.
>

I have soft limits ON all the time. I'd like to have soft limits for Z
too but can't find that. I wreck the nozzle allready. The nozzle was at
-22. Then was commanded to +1. But the nozzle stays at -22 as smoothie
has soft limits too (0 to -25) and command was just ignored.

bert shivaan

unread,
May 12, 2020, 8:14:40 AM5/12/20
to OpenPnP
It seems like smoothie should have responded to the invalid move command. Maybe something is not correct in regex?


--
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/642181dd-682d-a00a-3557-a3df5d7da2bc%40gmail.com.

Slavko Kocjancic

unread,
May 12, 2020, 8:24:22 AM5/12/20
to ope...@googlegroups.com
I think that openPnP does soft limits. I do that on smothie too as I
can't find where to set limit for Z in openpnp. If I command smothie to
go out of coordinates the response is "ok" but doesn't move at all.

if I set soft_endstop.halt to true then report error but to restore I
need to do M999 and homming again.
> <mailto:openpnp%2Bunsu...@googlegroups.com>.
> --
> 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
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/openpnp/CA%2BKNHNzFc94JXSX9-wgjfO%3D5UkvVZ2qqPbiJ27S%2BtiWFAvyqqA%40mail.gmail.com
> <https://groups.google.com/d/msgid/openpnp/CA%2BKNHNzFc94JXSX9-wgjfO%3D5UkvVZ2qqPbiJ27S%2BtiWFAvyqqA%40mail.gmail.com?utm_medium=email&utm_source=footer>.

ma...@makr.zone

unread,
May 12, 2020, 8:27:25 AM5/12/20
to ope...@googlegroups.com

> If understand correctly if I misstype two feeders to same -1000 I wreck setup for both?

If their fiducials coincide to with I believe +/- 4mm, yes. But nothing catastrophic happens, they are just fused together. You could untangle them through machine.xml.

But better backup and then move on/off feeders.

Note, it took me quite a while to realize: you can always save config, no need to quit OpenPnP.

> commanded to +1. But the nozzle stays at -22 as smoothie has soft limits too (0 to -25) and command was just ignored.

Good point. Ignoring commands in a controller and not halting the machine is not a good idea. Have you checked if there an option in Smoothie to halt the machine on soft limits violation?

_Mark

ma...@makr.zone

unread,
May 12, 2020, 8:30:26 AM5/12/20
to ope...@googlegroups.com

> if I set soft_endstop.halt to true then report error but to restore I need to do M999 and homming again.

Ah there it is. I absolutely believe one must set that to true.

The halt is a hassle, but otherwise you can get the larges crashes!!

I added the M999 to my ENABLE and HOME commands. Easy to recover.

_Mark

Slavko Kocjancic

unread,
May 12, 2020, 9:01:14 AM5/12/20
to ope...@googlegroups.com
On 12. 05. 20 14:30, ma...@makr.zone wrote:
> > if I set soft_endstop.halt to true then report error but to restore I
> need to do M999 and homming again.
>
> Ah there it is. I absolutely believe one must set that to true.
>
> The halt is a hassle, but otherwise you can get the larges crashes!!
>
> I added the M999 to my ENABLE and HOME commands. Easy to recover.

That's nice idea. Now it works with only one drawback.
If commanded outside the openpnp show messagebos that controller return
error. (nobody know what eror.) if I command to go outside of YX the
openpnp say that it's outside of soft limit. Why the Z is ommited jere
at all?

ma...@makr.zone

unread,
May 12, 2020, 9:06:39 AM5/12/20
to ope...@googlegroups.com

> Why the Z is ommited jere at all?

Soft limits are defined as camera coordinates. You can capture X/Y from the camera but not Z.

Maybe I should add the limits to my new axes instead? 

@Jason, what do you think?

_Mark

Jason von Nieda

unread,
May 12, 2020, 11:06:28 AM5/12/20
to ope...@googlegroups.com
Yep, I think that's where they belong.

Thanks,
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/2b97cef8-c550-08ee-fdf8-a8d8b348ab1d%40makr.zone.

ma...@makr.zone

unread,
May 21, 2020, 11:48:01 AM5/21/20
to ope...@googlegroups.com

Hi Jason, Slavko, everybody

Limits will be per axis and Z can no have them too. Only Rotation axes are excluded (obviously).

I used different "icons" to make it clear its neither the tool nor the camera captured/positioned but the raw axis coordinate.

...its not actually icons but HTML text :-)

_Mark

Reply all
Reply to author
Forward
0 new messages