As you have probably read in our AXIOM Beta Manufacturing Report, we are still building all the early adopter AXIOM Beta developer kits completely manually. Obviously we need to increase our production speed to grow the project and to be able to deliver a reasonable amount of cameras any time soon. So in the last four months, we have put a considerable amount of time and resources into automating our production processes. The most time consuming aspect is picking and placing components onto the PCBs, so following our project philosophy, doing this the open source way is the number one priority. In this article we want to report about the automated pick and place options, our experiences and also the ups and downs along the journey.
We consider being able to build and enhance affordable yet powerful electronic production machines a huge thing for hardware startups as well as open source projects. This technology in our opionion is an extremely enabling asset, so we see it as our duty to give back to these communities and push this technology forward ourself - not just for us, but for anyone else following with their own projects. Some people might see this holistic approach as a diversion from our core goals of developing and building a camera, but we think that only by being in total control of all aspects of camera production, can we stay truly independent as a community in stormy seas while reducing our dependencies on others.
Liteplacer
The Liteplacer is delivered as a self-assembly kit. It consists of so called 'Makerslides' and laser cut steel plates, electronics and a lot of screws and small parts. It's made and sold by Juha Kuusama from Finland. The mechanical assembly guide is excellent and depicts every step visually. Although the Liteplacer is not technically open hardware, it's very close as it's released under the creative commons non-commercial license - http://www.liteplacer.com/downloads/ states: "The licenses restrict commercial use. By “commercial use”, I only mean selling pick and place machines or software as a business. Selling boards made by the LitePlacer is perfectly ok".
Once the mechanical machine frame has been assembled, you are a bit on your own when it comes to sourcing the remaining electronic, cables and connectors and putting everything together. The Liteplacer comes with its own .Net application which is released as Windows binary only (Source Code). It is great to get the motors and sensors up and running and test all aspects of operation but for production we selected OpenPnp - more about that below. We have been upgrading and improving upon the basic machine by created a custom plate holding the TinyG motorcontroller and also ordered better top and bottom vision cameras and LED rings - the default Liteplacer cameras are 640 x 480 px and we need higher resolution for small parts. E.g. the 0201 (0.6 mm × 0.3 mm) components in the placement tests are sometimes incorrectly rotated 90 degrees because the component was just a few pixels in size and the camera couldn’t reliably see the orientation. The Liteplacer doesn't have the placing speed or accuracy of an industrial machine - but at a price tag roughly 97% cheaper, the bang for the buck is quite unbeatable. We are currently trying to find out why we have a changing camera/nozzle offset of up to 0.5mm and why it depends on where the head moves - we suspect a slight unevenness in the makerslides or V-wheels but it's proving hard to confirm - if you want to help out here please get in touch with us.
Complete Liteplacer on custom table with enclosures and custom plate for the machine control electronics
OpenPnP
When it comes to placing components, OpenPnP (open source software under GPL V3) offers some very valuable additional features when compared to the Liteplacer software - for example it runs on Linux and also offers bottom vision, part offset and orientation recognition. OpenPnP is under very active community development and we've already done our part by pushing small improvements upstream. We have successfully managed to place a number of different sized components with reasonably good accuracy - we even tried 0201 (0.6 mm × 0.3 mm) just for fun and some of them ended up in the right place :) . In general, we still see room for improvements - like higher resolution cameras (these have been ordered and are on the way), the calibration procedures or with the way we install components in tapes on the table, or custom designed semi-manual feeders. We are rather confident we will be able to produce the first AXIOM Beta PCBs this way soon.
In parallel to the Liteplacer efforts, we are also working with a much more solid/robust DIY pick and place machine built by Amescon (who provide office facilities where we are currently building the AXIOM Betas). So far this project has not got off the ground as we haven't found a software developer who could make the existing software work with our PCBs (it was designed for another use-case). As we recently got to love OpenPnP, we decided to give it a try with this machine. We can already move and home the head and receive images from both top and bottom vision cameras. This machine has slots for 24 pneumatic component feeders. Controlling these feeders and the vacuum valve are still work in progress. All in all we hope to have a robot army at our disposal soon, an army that can build cameras day and night - that would leave us more time to develop new camera hardware, feature and software and actually make movies with our cameras - we can still dream :).
Having wrestled with similar issues in a small tech startup may I offer this suggestion. Keep your eyes on the core mission of what you want to do. There are thousands of well qualified production houses that can efficiently provide a single proto up to the highest volume of production. Don't try to solve a problem that has already been well answered. Resources of time, focus and market opportunity are incredibly valuable. You will end up fighting with challenges unrelated to your core product and it will cost you, potentially dearly.
Hey guys,
I have to agree with Paul. I work for a small (<15 people) start up and we do PCB prototyping but production is of course done outside. Why try to assembly boards in house when really skillful people can do it for you for way less money?
I have the impression that this kind of activities would really slow down your current camera building effort.
That's what we thought as well, and so we got in contact with a bunch of local and some not so local assembly facilities, and after sorting out those which didn't meet our criteria we had each of them produce five of our sensor front-end PCBs. The three remaining facilities, two local and one overseas took their time and complained about various things, but finally delivered those five assembled boards. Unfortunately, only a fraction of the returned boards actually worked, and we had some trouble fixing the boards later on. We also spent a huge amount of money on building prototypes during the EU project, which is not something we can and want do on the limited budget we are working on, so we decided that we simply have to do it ourselves if we do not want to spend a lot of money and still get it done properly.
Don't get me wrong, if we find the right assembly place which is flexible enough for our special requests and still affordable, we gladly drop our prototype and production plans. So please by all means, get us in contact with those really skillful people ....
You generally want to test waters with simpler parts, power board or something. The thing with camera sensors is they can't handle heat, if you didn't specify to manufacturer that sensor must be soldered manually they probably stuck it in the oven and cooked it.
Also these manufacturer do tons of designs from tons of companies and have huge amount of experience about it, if they complain its generally for a good reason and it pays to listen and fix the design. Its in their interests for the boards to turn out OK so you would come back for more, so if they complain they fear things are not going to work as intended and you won't be happy with the end result. The companies that don't complain are the type that don't give a fig about the end result, they will take your money and your design and it will turn out however it will turn out, usually for the worse.
I don't usually reply due to obvious reasons, like non commercial thinking like this. However, I was looking at a good link to advise people disappointed by the craft camera failure, to go over here and see a more original concept.
Grant Petty, of Black Magic Design, years ago turned up on sunrise TV with the wave solder machine that did all the placement. Now, as others are saying here there must be heaps of contract manufacturers with these machines that can do it for you. Ultimately, as previously stated In yesteryear I think, contract manufacturing is the way to go with limited volume cheaply. This means production prototyping. The sort of numbers floated around here are way high, the supply as well as the manufacture chain needs to by tightened up and get parts prices down. This is all business stuff. If you want to run it as a business, talk to Rai Orz who did the Drachen camera in Germany (look up the Drachen Feader film). He has a good head on him.
Disregard and delete the previous comment, there seems to he an appropriate comment, just not from Sebastian. It is still a matter of production prototyping for the machine it will be made on. If you expect to fix it latter, you are asking the wrong question. The original manufacture should be so high quality the defect numbers are so low you can afford to throw them away. So, if they are complaining about things it is worth listening to and adjusting your design to fit the process, I can imagine heaps of instances where design placement would not suit a machines abilities.
8 Comments
Having wrestled with similar
Having wrestled with similar issues in a small tech startup may I offer this suggestion. Keep your eyes on the core mission of what you want to do. There are thousands of well qualified production houses that can efficiently provide a single proto up to the highest volume of production. Don't try to solve a problem that has already been well answered. Resources of time, focus and market opportunity are incredibly valuable. You will end up fighting with challenges unrelated to your core product and it will cost you, potentially dearly.
Hey guys,
Hey guys,
I have to agree with Paul. I work for a small (<15 people) start up and we do PCB prototyping but production is of course done outside. Why try to assembly boards in house when really skillful people can do it for you for way less money?
I have the impression that this kind of activities would really slow down your current camera building effort.
That's what we thought as
That's what we thought as well, and so we got in contact with a bunch of local and some not so local assembly facilities, and after sorting out those which didn't meet our criteria we had each of them produce five of our sensor front-end PCBs. The three remaining facilities, two local and one overseas took their time and complained about various things, but finally delivered those five assembled boards. Unfortunately, only a fraction of the returned boards actually worked, and we had some trouble fixing the boards later on. We also spent a huge amount of money on building prototypes during the EU project, which is not something we can and want do on the limited budget we are working on, so we decided that we simply have to do it ourselves if we do not want to spend a lot of money and still get it done properly.
Don't get me wrong, if we find the right assembly place which is flexible enough for our special requests and still affordable, we gladly drop our prototype and production plans. So please by all means, get us in contact with those really skillful people ....
Thanks and all the best,
Herbert
You generally want to test
You generally want to test waters with simpler parts, power board or something. The thing with camera sensors is they can't handle heat, if you didn't specify to manufacturer that sensor must be soldered manually they probably stuck it in the oven and cooked it.
Also these manufacturer do tons of designs from tons of companies and have huge amount of experience about it, if they complain its generally for a good reason and it pays to listen and fix the design. Its in their interests for the boards to turn out OK so you would come back for more, so if they complain they fear things are not going to work as intended and you won't be happy with the end result. The companies that don't complain are the type that don't give a fig about the end result, they will take your money and your design and it will turn out however it will turn out, usually for the worse.
I don't usually reply due to
I don't usually reply due to obvious reasons, like non commercial thinking like this. However, I was looking at a good link to advise people disappointed by the craft camera failure, to go over here and see a more original concept.
Grant Petty, of Black Magic Design, years ago turned up on sunrise TV with the wave solder machine that did all the placement. Now, as others are saying here there must be heaps of contract manufacturers with these machines that can do it for you. Ultimately, as previously stated In yesteryear I think, contract manufacturing is the way to go with limited volume cheaply. This means production prototyping. The sort of numbers floated around here are way high, the supply as well as the manufacture chain needs to by tightened up and get parts prices down. This is all business stuff. If you want to run it as a business, talk to Rai Orz who did the Drachen camera in Germany (look up the Drachen Feader film). He has a good head on him.
Hope it ties well.
Disregard and delete the
Disregard and delete the previous comment, there seems to he an appropriate comment, just not from Sebastian. It is still a matter of production prototyping for the machine it will be made on. If you expect to fix it latter, you are asking the wrong question. The original manufacture should be so high quality the defect numbers are so low you can afford to throw them away. So, if they are complaining about things it is worth listening to and adjusting your design to fit the process, I can imagine heaps of instances where design placement would not suit a machines abilities.
Hello, is there any info
Hello, is there any info available on this custom PnP machine you mention? I'm interesting in building one as well and may be able to contribute.
BG
Please contact us via the
Please contact us via the contact form, we are happy to discuss share our build. The OpenPnP mailing list is also a good community for such projects.
Add new comment