Pandaplacer A1 machine with feeders.

1,344 views
Skip to first unread message

Jarosław Karwik

unread,
Jul 9, 2024, 9:43:02 AM7/9/24
to OpenPnP
As to keep things separate I have created new thread for Pandaplacer ( and only Pandaplacer - this discussion started here https://groups.google.com/g/openpnp/c/JucIVeN_f0Q/m/k3kyE0ZgAQAJ?utm_medium=email&utm_source=footer  and we need to keep things separate)

I have bought the kit ( should arrive in 20-3 weeks) 
In the meantime I will be printing parts. I will show the kit once arrives ( I have not seen unboxing video for it)

zzimdog

unread,
Sep 3, 2024, 4:05:30 PM9/3/24
to OpenPnP
Hi Jarosla,
I just received my panda placer today and started the assembly. They have bee very responsive which is good because I already had questions during the first stage and second. The documentation has not caught up to the current changes and a little rough. 

Mike

PP.ca

unread,
Sep 3, 2024, 7:18:42 PM9/3/24
to OpenPnP
Make sure you try 0402s …

Jarosław Karwik

unread,
Sep 4, 2024, 2:46:45 AM9/4/24
to OpenPnP
Well, I was occupied by preparing my CHMT-36VA for sale so I could not fully assembly mine, but few pictures and comments.

The package itself:
Zrzut ekranu z 2024-09-04 08-44-35.png

Zrzut ekranu z 2024-09-04 08-44-52.png

Jarosław Karwik

unread,
Sep 4, 2024, 3:03:08 AM9/4/24
to OpenPnP
After unpacking
Zrzut ekranu z 2024-09-04 08-45-07.png

All nicely packed , all the components labelled.
However the assembly manual is not up to date:
- Several parts were updated and do not match the assembly pictures
- The frame is prepared for both AS1 and AS2 feeders, but the manual shows only AS1 assembly without mentioning it.
- Some parts of the manual show details of the assembly with bigger overall picture where the place actually is located.

I have the frame partially assembled and I have few comments:
- I would gladly pay a bit more for few 40x20 instead of 20x20. This would help stability as single 20x20 profile can be rotated when attached only using one center screw. Especially these under linear rails.
- One big questionable design choice - Z end position sensor. It activates when one head is in its down position -  this is simply wrong and may potentially destroy something. Should be mid position sensor, but not a notch - 0/1 transition detector when the heads reach equilibrium. Quite easy to make with transoptor.
- Belt tensioners - you can of course do it with springs on belts, but I do not like them, especially that there is not much space left for them. I have bought these ( ~5...8$ each, you need 3) and they can be easily added to the machine:
Zrzut ekranu z 2024-09-04 08-58-52.png

Marshall S. (Alakuu)

unread,
Sep 4, 2024, 5:33:16 PM9/4/24
to OpenPnP
I keep thinking I should add belt tensioners to my machine too. Wasn't sure if I should try and match the diameter of the idler or not though. They're a bit bigger than the standard ones I've found on amazon.

Jarosław Karwik

unread,
Sep 5, 2024, 5:49:07 AM9/5/24
to OpenPnP
Small difference will not matter - for idler. Would be different case for main driving wheel.

Here mine tensioner and original idler - it is more narrow with a bit different diameter.
I will just reuse original idler ( it means some drilling and screw replacement as inner diameter is different)

Zrzut ekranu z 2024-09-05 11-45-24.png

Dominik Kraus

unread,
Sep 7, 2024, 12:12:01 PM9/7/24
to OpenPnP
I've been using Pandaplacer close to a year now and I can say only positive things about it. It has been pretty accurate and reliable from the start. Leo is always very helpful if any questions arise!

I'm currently in the process of doing some quite large modifications to cope with some bigger requirements regarding speed. I've upgraded to LDO motors and thus swapped the motor drivers, added a Basler industrial camera and I'm right now adding two more nozzles for a total of four. I'm hoping to get this done in time till the "Yamaha" feeder arrive

PXL_20240907_151717746.MP.jpg

Jarosław Karwik

unread,
Sep 7, 2024, 5:52:41 PM9/7/24
to ope...@googlegroups.com

With these many changes you might have easier way taking old Yamaha pick and place machine and stripping it just to the frame - you would get relatively compact machine with unparallel stiffness and big enough for all the feeders you could imagine. I still have such frame - with modern servos ( I have replaced old DC engines) and large ballscrewes. 


--
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/f7903818-aafc-45f1-a536-e8946232413fn%40googlegroups.com.

mark maker

unread,
Sep 8, 2024, 12:53:07 PM9/8/24
to ope...@googlegroups.com

Or simply buy a second Pandaplacer. People often forget that upping the number of PnP machines is the easiest and surest way to speed things up. A guaranteed factor two is very hard to get with hardware that just costs twice as much.

With two machines, you can spread out the human work (loading/unloading PCBs, refilling feeders, etc.) while the other machine is still chugging away. Plus you get more reliability, if one machine is broken, you still get to put out PCBs on the other.

I learned that with 3D printers. Very convenient to have two 😎

https://youtu.be/cNCjjvCT4Fc?feature=shared&t=236

_Mark

zzimdog

unread,
Sep 8, 2024, 5:56:30 PM9/8/24
to OpenPnP
I must say that Andrew and Leo have provided outstanding support it is actually the most responsive and thorough I can recall. I finally have the machine assembled and though the assembly was a nightmare for me trying to figure out with the current status of the documentation. I was having to reference everything I could find online as the pics are so tight it is hard to tell orientation or what part of the machine I was supposed to be working on. Definitely was the most confusing thing to follow I can remember. Hopefully they will put up some pics that at least show the machine from different angles or maybe at least a pic or drawing of the completed stage. 

On the other hand compared to getting my lightplacer setup on openpnp this thing was a breeze. Most of it was preconfigured and was done with it in minutes. 

I haven't used the machine in production yet as I need to print out tons of feeders and I am use to just laying down cut tape with the liteplacer so it will be a big change for me. The design detail of the machine seems really good the way everything is purposely designed and not like most things I usually end up buying with all off the shelf stuff clogged together to make something work.  Especially the feeder design with everything fitting in perfectly and cable routing and access holes to reach screws etc. Look forward to getting it up and running and if it works as well as it seems so far I definitely will be buying at least one more.

They need to get the documentation up to the standards for the rest of the machine for guys like me that don't have a lot of experience modding and dealing with pick and place machines though. Some of us just need them up and running so we can put them to work.

If there are any other rookies out there getting one of these the following helped me along.

There are tons of extra parts not sure what some of them are for yet. 
There are end stops in one of the bags and didn't notice them until I was already done and trying to figure out what all the stuff left over was for. So might want to use them during that part of the assembly. After assembling with the new ones following the images for the original way of doing it they but right up to the linear rails. I ended up needing to move them back some though as the limit switch would not trip in that position. 

Some of the connectors are replaced with the one of the y connectors in the box with the bottle.


The rails that the feeder boards mount on underneath need to be screwed down to the hole that is closest to the outside of the frame. 

There are 4 pins on the feeder board and only 3 pins on the feeder connector the 4th pin on the right dies not plug into anything with these feeders. The orange wire should be on the right when looking at a feeder from the back.

These videos were helpful with figuring out how to assemble the feeder especially when I couldn't figure out which servo arm to use since none of the different styles fit. 

How to align the feeders and blocks With an alignment tool link that is helpful to print out. 

The 2 step calibration model linked on one of the videos to print bolts right on to the bottom camera plate holes. 

Jarosław Karwik

unread,
Sep 12, 2024, 4:36:20 AM9/12/24
to OpenPnP
Belt tensioner combined with small reinforcement
I decided to add extra T profile as I have weakened the profile connection reusing the screw which connects them together ( for the block reinforcing the tensioner).
Zrzut ekranu z 2024-09-12 10-32-01.png Zrzut ekranu z 2024-09-12 10-31-53.png

Jarosław Karwik

unread,
Sep 12, 2024, 4:46:06 AM9/12/24
to OpenPnP
Final look
Zrzut ekranu z 2024-09-12 10-45-36.png

Mike Menci

unread,
Sep 12, 2024, 5:48:14 AM9/12/24
to ope...@googlegroups.com
Now tensioner nut has no functionality - it is fixed. 
- isn’t it? 
Sent from my iPhone

On 12 Sep 2024, at 10:46, Jarosław Karwik <jarosla...@gmail.com> wrote:


Final look

Jarosław Karwik

unread,
Sep 12, 2024, 6:22:50 AM9/12/24
to OpenPnP
No, it is still right - I get 6...8mm adjustment


Zrzut ekranu z 2024-09-12 12-19-08.pngZrzut ekranu z 2024-09-12 12-18-55.png

Thad

unread,
Sep 12, 2024, 6:16:33 PM9/12/24
to OpenPnP
I have this built a few weeks ago, but have not been able to get the nozzle tip detection/calibration to complete and have not put much time into it. Anyone with a functioning one got any tips or a better pipeline setting?

Jarosław Karwik

unread,
Sep 13, 2024, 4:03:52 AM9/13/24
to OpenPnP
See comments on the bottom - it is known that this is a challenge with Samsung:

Jarosław Karwik

unread,
Sep 19, 2024, 11:51:58 AM9/19/24
to OpenPnP
Now my mod with belt tensioners. Much easier to adjust now.
I have also added some reinforcements to the frame ( simple L connectors)

Zrzut ekranu z 2024-09-19 17-47-38.png
Zrzut ekranu z 2024-09-19 17-47-27.png
Zrzut ekranu z 2024-09-19 17-47-11.png

bing luo

unread,
Sep 19, 2024, 12:30:42 PM9/19/24
to OpenPnP
It's hard to imagine, all the structural components are 2mm thick iron plates. You can even easily bend it with your hands.

AAAAA.jpg


Compare what we have made,We have too many concerns about the intensity. 
微信图片_20240919113412.jpg微信图片_20240919133143.jpg
微信图片_20240919112318.jpg微信图片_20231211104856.jpg
Message has been deleted

Jarosław Karwik

unread,
Sep 20, 2024, 4:27:00 AM9/20/24
to OpenPnP
Hi,

Let me summarize few things about Pandaplacer mechanical design:
- It is very optimized. Up to the point in which you can bend these 2mm metal sheets - however during normal operation the forces on these components are relatively low ( compared to hand force),  so it will work ( even if it feels a bit creepy).  I would prefer these sheets to be more like 4mm - then it would be much stiffer and the cost increase should not be that big
- You could still use something like 2mm, but it would have to me "sandwich" design - meaning two plates instead of one - separated by spacers and all the wheels, belts. This would also give benefit of easy implementation of belt tensioners.
- If I was tempered to make such kits, I would design it with metal sheets, but made it with  CNC bending ( it is quite cheap even for small series).  This would make it both stiff and cost effective - for the price of more work on initial design.
- Pandaplacer uses 20x20 aluminium profiles - I would gladly pay for 20x40 or combination of 20x40 and 40x40 ( for main pillars)

- The cabling - using flat cables is quite smart idea. We will see how long it lasts, although printers have been using this succesfully for ages.  For machine that size I think I  would go for umbilical cable - even if it makes machine much higher ( which is not important)










czwartek, 19 września 2024 o 18:40:14 UTC+2 tianl...@gmail.com napisał(a):
I really appreciate Pandaplacer's low-cost design. I always thought that only CNC machining of aluminum alloy could achieve the strength and precision required for PNP. Now it seems that thin iron sheets can also be used, and we plan to develop a similar and inexpensive kit for everyone.

bing luo

unread,
Sep 20, 2024, 5:25:27 AM9/20/24
to OpenPnP

This is a 4mm thick 201 stainless steel plate that I cut with a laser. I use it to make cheap PNP, and it is impossible to bend it by hand. However, it is very heavy, so it cannot be used on the mounter.   The rigidity of aluminum alloy is similar to it, with only one-third of its weight. 

微信图片_20240920165253.jpg  


On the mouter,   I have to spend a lot of money on CNC  for this fixed C-axis motor accessory because I need to ensure its verticality.

微信图片_20240920171154.jpg

I got some inspiration from Betz's website.  But I spent more money than Betz because I used a piece of aluminum alloy for CNC  to ensure verticality, and Betz used CNC to machine two pieces of aluminum alloy and then combined them together with screws.

微信图片_20240920171803.png


Mike Menci

unread,
Sep 20, 2024, 8:17:37 AM9/20/24
to OpenPnP

Jarosław Karwik

unread,
Sep 20, 2024, 8:25:07 AM9/20/24
to OpenPnP
Yes - there are many. I have just select ones which fit almost directly.
However - beware !

They are made from soft aluminiom and most asymmetric ( single sided) cannot actually take the force directly.
That is why I have added reinforcements - either L block or screws to make it usable
Zrzut ekranu z 2024-09-20 14-22-06.png     Zrzut ekranu z 2024-09-20 14-21-45.png

I have selected this way as it was easy off-the-shelf way to add tensioner - starting from scratch there are easier and more compact ways to achieve it.

Bingluo

unread,
Sep 20, 2024, 8:35:57 AM9/20/24
to openpnp
This  is  a  charmhigh  PNP, and their handling method completely exceeded my expectations. Fold the belt first, then place a piece of iron sheet on it, and then tap the iron sheet to tension the belt through friction. Finally, use a mechanical screw to tighten it.





---- Replied Message ----
From Jarosław Karwik<jarosla...@gmail.com>
Date 09/20/2024 20:25
To OpenPnP<ope...@googlegroups.com>
Cc
Subject Re: [OpenPnP] Re: Pandaplacer A1 machine with feeders.

Jarosław Karwik

unread,
Sep 26, 2024, 11:32:02 AM9/26/24
to OpenPnP
Now fully assembled:

Zrzut ekranu z 2024-09-26 17-19-21.png


This part needs some work - such loose cables is not a nice thing.

Zrzut ekranu z 2024-09-26 17-19-31.png

One thing to be fixed - not sure I have missed something. The limit switch just needs adjustment - I will put something more permanent here later:
Zrzut ekranu z 2024-09-26 17-29-38.png




Leftovers and spare parts. Why the metal table ????
And what to do with fan - have not seen it in documentation

Zrzut ekranu z 2024-09-26 17-19-41.png

Now time for software check and power on.

Mike Z

unread,
Sep 27, 2024, 3:50:41 PM9/27/24
to ope...@googlegroups.com

I had to move the new style endstops back a little bit for the limit switch to hit and also need that little bit of extra room to be able to hit all 26 feeders on the sides. I cannot tell from the pic but how did you mount your bottom camera? The metal plate is to mount the camera and discard cup. I didn't have 2 in my kit but did have some extra parts but not as many as I thought after the base Assembly.


Mike Z

unread,
Sep 27, 2024, 4:20:29 PM9/27/24
to ope...@googlegroups.com
The fan I believe goes in the optional head cover. I haven't printed that yet as I'm still working on all the feeders. I ran the backside of one of my boards last night as I have enough feeders built to handle those parts and the difference between the panda placer and my old lite placer was astonishing. It built the board way faster and I am just doing the manal nozzle changes and trying to figure out how to use the second nozzle correctly. I think once I have the automatic changing enabled it will just take care of it. I did have some issues with the vacuum part detection and ended up disabling it as I couldn't figure out what I am supposed to do to change the detection level. I did see the graph and my vacuum settings were way lower than what it is set for by default. I tried changing them but again I have no experience with it and wasn't successful. On my first run it did a good job of placing all the parts and picking them all and the only problem I had was leader tape coming loose on one of the feeders. I haven't used them before either I have been suffering with cut tape glued to the table for the last 8 years.

Jarosław Karwik

unread,
Nov 8, 2024, 11:05:27 AM11/8/24
to OpenPnP
Now finally assembled with feeders ( only two until I  validate their operation)
Will test it on my next week small assembly

Zrzut ekranu z 2024-11-08 17-02-19.png

Zrzut ekranu z 2024-11-08 17-02-25.png

Zrzut ekranu z 2024-11-08 17-02-31.png

I had to move the bottle inside the machine - otherwise it would interfere with component spools.

Zrzut ekranu z 2024-11-08 17-02-43.png

Jarda Dvořák

unread,
Nov 22, 2024, 12:39:20 PM11/22/24
to OpenPnP
Hi Jaroslaw!
How its going with machine?
I have recently ordered one so I am curious, exited and little bit worried - because I can see how long you have been assembling it.
Your posts will be very valuable for me. Thank you! :-)

Dne pátek 8. listopadu 2024 v 17:05:27 UTC+1 uživatel jarosla...@gmail.com napsal:

Marshall S. (Alakuu)

unread,
Nov 30, 2024, 10:04:36 PM11/30/24
to OpenPnP
Hey! Haven't checked in ages but wanted to pop back in. Over a year of pretty steady use and I haven't had any issues with the PandaPlacer. Well other than one issue. The tensioning springs on the nozzle heads kept snapping. I realized that if I just don't replace them and make sure at start up the nozzle was even there's absolutely no concern.

I have feeders with 20k+ feeds on them alone without needing to replace any parts! 

Side question tho, has anyone gotten async working? I'd love to not have the poor machine sitting waiting on 12mm of feeder time per component on some of my larger parts. 

cborn

unread,
Dec 3, 2024, 5:16:33 AM12/3/24
to OpenPnP
I've been been trying out some more feedertypes on my PandaPlacer.
The BlindsFeeder is really nice, both in the OpenSCAD software and the feeder software implementation.

I printed out a base to hold the feeder array, based on the PandaPlacer tray holder, which lets me take it out easily for loading, or swap with another feeder array.
Base was designed in Rhino, but would probably be better off done in OpenSCAD and linked to the size of the array.
 
For the QR codes I initially used a Brother P-Touch labeller, but the reflections on the laminated tape made it to hard to get  consistent reading.
In the end I set up a Google-sheets spreadsheet I could just enter the part numers in, and have it print them out with the QR codes on plain paper, which works really well.

See below for my current test setup.

PP_BlindsFeeder.jpeg

I then spent a long time trying to get the AdancedLoosePartFeeder working, with little success, mainly because I'm still not familiar with the CV Pipeline, and because the WiKi instructions for the configuration are still not done so I'm not quite sure how to set it up.
I did run a training pipeline which captured an image of my part, and then had to manually copy the name into the Feed Pipeline, but couldn't get it to recognize it.
Any help on how you go about configuring this would be appreciated!
Loose parts ore on the table closest to the camera.

Next I tried the ReferenceLoosePartsFeeder, using a green-screen pipeline Mark had posted, with a bit of tweaking to get it to identify my parts.
I was pleasantly surprised when a "feed & pick" located a part and picked it up!
I then got the "No part found" dialog while the part was still stuck to the nozzle, probably vacuum settings or something I need to look into.

However the next bit was a surprise - I hit the "Recyde" button, and the machine went and opened the BlindsFeeder array for the 1206-10K part (I had the same part set in both) and put the resistor into an empty pocket in the BlindsFeeder.
I wasn't expecting that!
And it did the same for the next part, and the next, moving to another empty slot each time.

It appears we have the making of a system to pick up loose parts and sort them into pockets, something that could be really useful given the number of loose parts I have!

Was this intentional, or just a lucky happenstance?

Jarosław Karwik

unread,
Dec 3, 2024, 5:30:19 AM12/3/24
to OpenPnP
Once you have access to  cheap automatic feeders - why do you bother with eve blind feeder ?  Especially for resistors where even full reels are cheap

cborn

unread,
Dec 3, 2024, 6:15:52 AM12/3/24
to OpenPnP
Totally agree, but at this stage I'm playing around with the various feeders trying to learn how they work, and using a standard (and cheap) part makes it easier.
I do a placement run, and afterwards sweep all the parts, including discards, back into the pile of 10Ks, and start again.

I still have a lot to learn about fiducial vision (I think some sliders for tweaking pipeline settings, like in bottom vision, would be useful), and I haven't even started looking at bottom vision yet!

I'm also running from latest test source, building in both Eclipse and  Intellij IDEA for comparison, and I'm new to Java, so  lot of learning going on!

Jan

unread,
Dec 3, 2024, 6:34:12 AM12/3/24
to ope...@googlegroups.com
Hi Cborn!

On 03.12.2024 11:16, cborn wrote:
[...]
> However the next bit was a surprise - I hit the "Recyde" button, and the
> machine went and opened the BlindsFeeder array for the 1206-10K part (I
> had the same part set in both) and put the resistor into an empty pocket
> in the BlindsFeeder.
> I wasn't expecting that!
> And it did the same for the next part, and the next, moving to another
> empty slot each time.
>
> It appears we have the making of a system to pick up loose parts and
> sort them into pockets, something that could be really useful given the
> number of loose parts I have!
>
> Was this intentional, or just a lucky happenstance?
>

Even thought the ability to take a part back is a per feeder feature,
I'd consider your observation a bug as I'd expect the part to be put
back into the feeder it has been picked from. Anyhow, if the feeder it
was picked from does not support to take it back and the blinds feeder
does, it could be considered a feature and it's likely ok to keep it as
is as long as it did not caused any damage.

Jan

Matthew Johnson

unread,
Dec 30, 2024, 10:24:10 PM12/30/24
to OpenPnP
Would you mind sharing the files? I also use Rhino for all my 3d work.

cborn

unread,
Jan 2, 2025, 7:04:51 PM1/2/25
to OpenPnP
Hi Matthew,

Ive added them to my repo fork here for now, possibly not the best place but I'm not sure where that would be!

It includes the OpenSCAD I used. As I mentioned, I think it would be better to make a version that automatically sized to fit the OpenSCAD output, I may look at doing that if I get a chance to spend some time learning OpenSCAD.
Another option may be using GrassHopper.

Cheers,
David

Jarda Dvořák

unread,
Jan 3, 2025, 1:50:31 AM1/3/25
to OpenPnP
Hello everyone,
my Pandaplacer is not complete yet but I have made two models I would like to share with you.
It is a controller box and AS2 servo arm replacement.
Dne pátek 3. ledna 2025 v 1:04:51 UTC+1 uživatel cborn napsal:
Reply all
Reply to author
Forward
0 new messages