Looking for Feedback from LightPlacer Owners

[RWB]

Hey everybody!

I'm interested in getting a LitePlacer PNP machine since the price is right and it looks like it has the ability to be very accurate with a few tweaks.

I'm really wanting to get some feedback from you guys who have built Light Placers and have used them to build populate your own personal boards. How has the machine worked out for you over time?

How hands off is the machine once you get it setup and running how you want?

Does it need constant tweaking to keep it running accurately?

What size parts would you say it is best with?

Tons of questions of course but really just want some honest user feedback.

Juha has provide excellent Email support so far and his build instructions look great.

Any info is appreciated

[mrandt]

Hey everybody!

I'm interested in getting a LitePlacer PNP machine since the price is right and it looks like it has the ability to be very accurate with a few tweaks.

Welcome to the community

The price is excellent and - as I put in in another post already - I believe Juha currently offers the most complete DIY P&P machine kit.

If you look at other projects you will find FirePick is still not ready, Protovoltaics Kickstarter was not funded, Varioplace not ready yet either... and most others are in a higher price range.

Unless you want to build your own machine from scratch (e.g. using Jason's OpenPNP hardware, Volker Besmens or Brian Dorey's machine for guidance and inspiration) I think you currently cannot get better value for the money.

Not everything on LitePlacer is perfect yet but it is a good starting point. Software is continiously being improved and the community is very active; developing and sharing hardware tweaks and add-ons.

You have to feel comfortable with building and tweaking the machine though. It is not an "off-the-shelf-ready-to-run" product. It works well once you have assembled and tweaked it and found your "operational scheme" but it takes some time and effort to get there.

If you want to buy a machine which just works after a few hours, I suggest you better spend 3-4 times the money and get a commercial desktop P&P.

If you like to make things, it is a nice and rewarding experience to build your own pick and place machine.

I'm really wanting to get some feedback from you guys who have built Light Placers and have used them to build populate your own personal boards. How has the machine worked out for you over time?

I used to assemble boards by hand - which is tedious and tiring work. I could not afford the commercial options so I started building a FirePick at first. I had serious problems with accuracy and also found that the machine would not serve my purposes so I discarded it and went for LitePlacer instead.

I built mine as an "alpha" version following Juha's CAD plans even before the kit became available - so of course it took me much longer than any usual customer. I think with Juha's kit, you can build the machine in a few afternoons - depending on your skills, experience and tools.

The software has a learning curve but if you follow the instructions you should be able to place your first board quickly.

I have had my LitePlacer for about 8 or 9 months now and am reasonably happy with it. I am mostly building prototypes and low volume series, most boards have less than 100 components and I do not use smaller than 0603. Accuracy and repeatability are good. I will post a few pictures of assembled boards when I have time.

I know other users who have successfully populated large numbers of boards; Adam (wayoutwest) has recently reported that he built hundreds of boards with many components and let his LitePlacer run overnight.

How hands off is the machine once you get it setup and running how you want? Does it need constant tweaking to keep it running accurately?

Setup takes a while, especially if you have many different components. If you use some sort of clever tray system or even build automatic feeders, this time can be reduced.

Once a job is configured and works, machine runs nicely on its own. Only user intervention is if you need to change the nozzle (large differences in part size) or refill some of the parts.

Some hardware add-ons greatly improve the error-rate; I consider a stronger pump and vacuum buffer reservoir (tank) in combination with a pressure sensor to detect failed pickups the most relevant changes.

I am also working on automatic nozzle change like Karl (knas) presented on this forum and automatic feeders - but this is all optional. For a start, component strips taped to the table work fine.

I hardly ever do tweak anything for accuracy. There are some calibration routines in software, mostly using the cameras, which I run before each job. But the mechanics are very stable - nothing to tweak there other than adjusting eccentric spacers on V-wheels and maybe tensioning the belts every now and then.

What size parts would you say it is best with?

0805 and 0603 passives work like a charm. 0402 is probably possible, but I currently have no need to go that small. For larger parts, placement ability mostly depends on the vacuum source and nozzle tips you use. Blunt needles combined with small rubber suction cups can lift larger IC without a problem.

With a custom nozzle holder and mating nozzles or grippers, special parts should not be a problem either.

Alignment correction for ICs (QFN, TQFP, BGA) is still work in progress - but with a high-resolution upwards camera I am confident that 0.4mm pitch QFN or TQFP should not be a problem and will work fully automatic in the near future. Even BGAs might be possible (have to prove that still...) [/quote]

Tons of questions of course but really just want some honest user feedback.

Juha has provide excellent Email support so far and his build instructions look great.

Any info is appreciated

Yes, Juha is very helpful and supportive. Build instructions are also good and this forum is another excellent source of information.

I hope, my experiences also help you make a decision.

Have fun!

[WayOutWest]

Definitely agree with everything Malte said. In addition,

I built mine as an "alpha" version following Juha's CAD plans even before the kit became available - so of course it took me much longer than any usual customer. I think with Juha's kit, you can build the machine in a few afternoons - depending on your skills, experience and tools.

I built mine from Juha's kit in 1.5 (long) days. The first day was a long 13-hour sprint to get all the mechanical parts done. After that there was an afternoon of wiring.

Personally my biggest attraction to the liteplacer is how hackable it is. There is nothing else in the PnP space that comes even remotely close to the liteplacer's combination of capaibility+hackability. Nothing. The only more-capable machines are proprietary commercial systems that leave you at the mercy of the manufacturer if you want to add an unusual feature. None of the other open-hardware-ish platforms work even half as well "out of the box".

I will say, however, that FirePick has better vision software. You didn't hear it from me, but if you swipe their vision library and use it with a liteplacer you'll have something pretty killer ....

I know other users who have successfully populated large numbers of boards; Adam (wayoutwest) has recently reported that he built hundreds of boards with many components and let his LitePlacer run overnight.

With two major caveats: (1) all my software is either custom or taken from the FirePick's vision library and (2) I run my machine very very slowly, doing a vision operation before every single pickup and every single placement (yes even for passives), which is considered an unacceptable performance killer by most people.

How hands off is the machine once you get it setup and running how you want? Does it need constant tweaking to keep it running accurately?

Some hardware add-ons greatly improve the error-rate; I consider a stronger pump and vacuum buffer reservoir (tank) in combination with a pressure sensor to detect failed pickups the most relevant changes.

Second that. Don't try to use a pressure sensor without a vacuum buffer, the readings will be useless.

Even BGAs might be possible (have to prove that still...)

I place several 0.8mm-pitch BGAs on every board. The machine is -- mechanically -- absolutely capable of doing this. But again, I do a very slow+careful upcam alignment before each placement.

BGAs are actually easier than people think; they self-align very strongly due to the surface tension of so many contacts and you can use tacky flux instead of solderpaste so misplacement-induced bridging is not as likely (smearing flux is not bad like smearing solderpaste!). I think they're easier to place reliably than 0402s, but maybe that's because I've been building boards with them for a while now. Just remember that rotation is critical: 0.5 degrees error maximum, preferably 0.25 degrees.

[RWB]

Thank you very much mrandt & WayOutWest for taking the time to give your review of the Lightplacer PNP and your experiences with it.

I agree for the money it looks like the best value on the market.

I have 2 machines that use the same X Y Z table setup with stepper motors ect and I know first hand how accurate they are when running correctly. I have a laser cutting / etching machine & CNC mill and they are both very useful and used on a regular basis. So I know the Light Placer has the potential to place BGA's accurately eventually.

@WayOutWest from what I can tell and as you have already mentioned it seems that to get this accuracy were all desire the machine just needs better software & vision to get your parts aligned perfectly after they are picked up. It sounds like you have this up and running now @WayOutWest and that's very exciting to hear. I wouldn't mind the Light Placer running slow if it can get all my fine pitch parts placed properly while I'm doing other things.

Is your custom software something that would be not to difficult for me to use if you shared it? Do you plan on wrapping up the code and releasing it for others to use in the future?

Right now the only other machine on the market that looks very good is the NeoDen 4 PNP with Vision. It supposedly place 5000 components per hour which is pretty good but as of right now the max part height is 5mm which makes some of my CAP's not place-able by the machine.

There is a guy on the EEV forum who just purchased a NeoDen 4 PNP and is doing a review on it, and many others are in line to buy one also. I'm following this thread to see how well the machine actually works vs how its supposed to work.

Here is the forum post for the Neoden 4 review if your interested also : http://www.eevblog.com/forum/reviews/ne ... -place/60/

Guys I would love to see some pictures of the boards that you are producing with the Light Placer so please do share if you can.

@WayOutWest if you have any youtube videos showing the speed of your current board runs that would be awesome.

What I like about the Light Placer is that it can place all my Caps where the NeoDen 4 can not. I think if WayOutWest's better vision software would allow me to place my 0.5mm chip + taller caps even at a slower speed I would just buy the machine. But since I do not have a immediate need to populate boards at the moment I'm waiting to see how the NeoDen 4 works out for others.

Again thanks for your feedback and looking forward to more hopefully

[mrandt]

Hi RWB,

the way I see it Neoden 4 plays in a completely different league...

They optimized for speed (multiple heads) and larger volume manufacturing (conveyor belt mechanism). If you buy some feeders with the machine, you easily pay > 10k$ for their machine.

Their hardware seems to be quite nice and clean - and probably more accurate and faster than a DIY machine. From what I heard, their weak point is software. Everything including the controller is closed source - so you have to trust some Chinese company to support it for many years... Not sure if I would build my business on that assumption.

LitePlacer is more of a hacker's / maker's tool - a kit which gives you a good starting point and can be modified and enhanced as needed. Everything about it is open source. CAD plans are as freely available as the software source code. If you don't like Juha's software (or any of the forks currently out there), use OpenPNP or write your own... Besides Adam's fork of the software, there is also Reza's fork which is currently being enhanced by Karl. I know of at least two other people coding their own fork of the software - and so far most people have shared their enhancements. Last not least, Juha is also enhancing his "trunk" version of LitePlacer windows program.

I think it is just a matter of time until we have an even more complete controller software which is also easier to use.

For me, the openness was the most important point - besides the budget

What do you want to use LitePlacer for? How many PCBs do you want to produce in which time? How many parts per PCB? How many different layouts? Special parts? How much time and / or money are you willing to spend? Do you see value in the learning experience of building your own PnP? I believe all this will influence your decision...

If I had little time but a lot of money and wanted to produce larger quantities of boards with same layout, I would buy a commercial machine like Neoden, Manncorp, Essemtec or the like. If I wanted to save money, learn about PnP bottom-up and would produce smaller quantities of many different PCB (think prototype), I would jump on LitePlacer train again

I saw your CNC mill in the other thread. Let me use that for a comparison:

You could have bought a professional CNC from a well reknown brand, let's say Tormach, and spent 25k$ or more. You get a nice piece of hardware delivered to your door with a fork-loader, ready to run with a complete toolchain of software. Probably more capable than the mill you have know; sturdy, made from steel and able to mill hard materials. Probably over-engineered for many DIY or semi-pro purposes, but definetely nice.

However, you went for a (supposedly) much cheaper machine that likely required some form of assembly, is made from aluminium and thus will not allow to mill hard materials (at least not at higher speeds) and has a learning curve when it comes to tuning and software.

Same decision to make for PnP here - just my two cents.

Regards
Malte

[WayOutWest]

Besides Adam's fork of the software,

It's not a fork of the LitePlacer code; it's all written from scratch in Java+C (except the FireSight+OpenCV libraries that I use). Just need to correct that so I don't get even more people asking me for patches.

I love the liteplacer, but I think the choice of a proprietary programming language for its default software is it's biggest, and possibly only, defect.

[mawa]

I love the liteplacer, but I think the choice of a proprietary programming language for its default software is it's biggest, and possibly only, defect.

I must answer back that C# is no more proprietary than Java, Python or C++.

Calling that a "defect" is that kind of "java blindness" I encounter since many years.

I run a CRM Software company since 1992 and use C# in my company for more than 9 years now.

IMO bashing C# because it's a Microsoft product is unfair and biased.

You find dozens of DIY projects where e.g. an arduino device has a C# PC UI front end. You can even get an arduino IDE for VS from Visual Micro for free.

You can get the IDE aka Visual Studio 10 and now replaced by Visual Studio Community for free and a whole bunch of ready to use UI controls and more than enough libraries including openCV for almost anything you can think of.

You can create easy to debug and easy to deploy multi threading applications which run out of the box on the majority of PCs.

I think it is sad when somebody who is publishing his DIY project uses a language + IDE that you either

• cannot run because it and its development IDE only runs under Linux,
  or you need to buy some base product like Mach3 in Brian Doreys PnP solution
  or you cannot afford like Volker Besmens PP4 project written in Delphi in a professional IDE and uses proprietary Delphi libraries and therefore cannot be developed/adapted using Lazarus.

Therefore I am very grateful that Juha chose to write his code with VS and C#.

[Mark Harris]

As a note, from someone with a long long background in professional C# development.. most of C# is now open source, pretty much everything except the windows interop/winforms/wpf stuff is open source now. The .NET CLR has always been open source, there were many closed source libraries it used which were not and used windows interop calls, so would not run on other platforms. This is where the Mono project came in - since Microsoft released a huge amount of code as open source in the last few years, mono is playing catch up big time trying to get it all integrated rather than trying to reverse engineer (ie: decompile the IL and turn it into sensible code)..

Pretty much everything for the web platform (ASP.NET/ASP.NET MVC/Razor etc) is all open source....

As an added bonus, .NET/C# is one of the fastest languages to develop in, with all the handy stuff that saves so much time (LINQ, all the easy multi-threading stuff, vast amounts of libraries) and is the most used language in professional environments. Just search any job advertisment site for "C#" vs "PHP" or "C++" or "Java"... and you'll find that C# probably has more jobs listed than all the others combined. Companies use it for a reason - it's fast, it's reliable, it doesn't chug memory down (cough..java..cough) and it's very easy to debug, understand and use.

Being the most used professional language has big bonuses, when a professional developer contributes code its usually of exceptional standard compared to the average open source project. Personally, LitePlacer's code is a horrific mess (one enormous form, no architecture to speak of), but that will change and improve. Compare your average open source .net projects code base to your average open source php/phyton etc code which are primarily amateur/hobbyist languages, and almost always you'll find the .net project's code is cleaner, faster, easier to understand and better optimised. Obviously there are plenty of exceptions but the vast majority of .net open source projects i've used have first class code bases... the vast majority of traditionally linux based language projects are a nightmare of bad coding practices and security holes (ever looked at wordpress or drupal?)

It's like Arduino project code bases vs those developed for ARM 7/9/Cortex without the use of "maker-oriented" helper libraries.. the ARM projects tend to be built by professional embedded developers or hobbyists who have been there done that and made many of the mistakes you make when learning

Likewise look at Fritzing/EaglePCB projects on average vs those using Altium... those of us using altium are either pirates or professional engineers who have a clue what we are doing Sure, you do get professional engineers using Eagle, but the vast majority are are using Cadence/Altium, and the vast majority of hobbyists use Eagle/designspark/fritzing (for those with truly no clue). Half the designs I work on would take 10-30 times longer in Eagle vs Altium and be of much lower quality in eagle (hand routing in altium is just so far superior, especially for high speed/impedance matched signals.)

Personally, I'd prefer something a bit more proprietary/closed source than something a bit more rubbish/made by people with no clue any day of the week!

I love the liteplacer, but I think the choice of a proprietary programming language for its default software is it's biggest, and possibly only, defect.

I must answer back that C# is no more proprietary than Java, Python or C++.

Calling that a "defect" is that kind of "java blindness" I encounter since many years.

I run a CRM Software company since 1992 and use C# in my company for more than 9 years now.

IMO bashing C# because it's a Microsoft product is unfair and biased.

You find dozens of DIY projects where e.g. an arduino device has a C# PC UI front end. You can even get an arduino IDE for VS from Visual Micro for free.

You can get the IDE aka Visual Studio 10 and now replaced by Visual Studio Community for free and a whole bunch of ready to use UI controls and more than enough libraries including openCV for almost anything you can think of.

You can create easy to debug and easy to deploy multi threading applications which run out of the box on the majority of PCs.

I think it is sad when somebody who is publishing his DIY project uses a language + IDE that you either

• cannot run because it and its development IDE only runs under Linux,
  or you need to buy some base product like Mach3 in Brian Doreys PnP solution
  or you cannot afford like Volker Besmens PP4 project written in Delphi in a professional IDE and uses proprietary Delphi libraries and therefore cannot be developed/adapted using Lazarus.

Therefore I am very grateful that Juha chose to write his code with VS and C#.

[WayOutWest]

Well then.

[mrandt]

But I also like Java

*troll* *scnr*