Neoden 4 pick and place

[thommo]

Oh - I forgot to mention another thought.

Consideration could be given to requesting Neoden hand over their 'vision library' strictly for use in 3rd Party SW running on a Neoden machine. Either as Dynamic or Static - whatever they felt most secure with.

It takes the pressure off them so they can concentrate on the hardware [and still offer it at the same price point], and rest assured that it has the best available, English Interface operating it.

This could move very fast once the ground swell began!

[vonnieda]

Hey Jason,

Would a reverse engineered prototcol help or do you need actual access to a machine? I'd like to help out of possible with this.

Hi elmood,

Protocol would be enough to get started. I would not be surprised to find out they are just using a modified / cloned Smoothie, to be honest. I'm seeing a lot of that recently.

To get a solid port going I'll eventually either need access to a machine or access to someone who has one and is interested in helping at a pretty in depth level. Ideally, the way this would go is someone who owns the machine is interested enough in running OpenPnP on it that they will do the development work and I can assist with the complex bits.

I don't own a N4, so my interest in this purely that I'd like OpenPnP to be available for more machines and after having seen the N4 in Shenzhen last year I was quite enamored with the hardware.

The first step, though, is to get some basic information about the hardware and protocol. From there I can kind of give an idea of whether this is something that would be a quick port or if it's going to require in depth work.

Jason

[vonnieda]

I wonder if it would be worth asking Neoden for info on protocols - probably not but can't hurt to ask...

BTW some pics of the internal hardware would be interesting to see...

Someone from the OpenPnP IRC channel asked them for information on their feeder protocol. If I remember correctly they said they would be okay with this but nothing has actually appeared.

[rx8pilot]

If I went down the development road.....I would never consider relying on the Neoden hardware. Not because it is useless, but because it comes from a source that could change at any moment or go away entirely. For the effort needed on the software side, it would be tragic if the hardware did not cooperate. Drivers, controllers, etc would have know and published interfaces and protocols.

With the goal of being low-cost, the bare hardware could be done better and lower cost than what Neoden is charging IMHO. Maybe not by much, but enough to justify a hardware design that is solid and reliably available.

[elmood]

thommo: I'm guessing that the chances of Neoden sending you any code or protocol docs or anything are probably zero. It's just not how these Chinese companies operate. The fact that they went to some kind of lengths to ensure that the English language pack can only be used with specific authorization from them leads me to believe that they take the software aspect of their design fairly seriously.

Jason: I would have more details on the driving electronics except that I can't find a simple way to open the N4 without some major disassembly. There don't appear to be any access covers other than by removing the build table and perhaps a lot of the mechanical parts on top as well. If anyone had theirs open I'd be curious to know how... for fear of messing up any calibration before getting more familiar with the machine I didn't, in fact, do what Dave Jones says to do.

[TheSteve]

Elmood - simply lay the machine in its back and remove the access port on the bottom to see everything.

btw, board 2 of 2 went pretty good. If not for the darned SOIC-8 feeding problems it may have done the 1550+ parts with no stopages. We have it running pretty slow - it took 1.5 hours or so to do the board. There was no optimization of the feeders at all as this was a board we hadn't planned to build.

[rwb]

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

[thommo]

You're probably right on that one elmood.

My thoughts were that if Neoden could see advantage in being a Hardware Only Operation, and market the machine in China with their own SW application, and elsewhere with a 3rd party English plug-in option, they would have nothing [or next to nothing] to lose.

On the other hand, we just need to identify a capable image-recognition source that will do what is required - even if it needs to be purchased as a license.

thommo: I'm guessing that the chances of Neoden sending you any code or protocol docs or anything are probably zero.

[TheSteve]

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

My biggest concerns with the machine at this point are not the software, it is that the feeder hardware simply isn't good enough. They seem to work well enough with 0805 sized parts but the 0603s we have done have not gone as well. The feeders just don't seem smooth enough and cause parts to jump out of the pocket, flip over etc. We didn't spend tons of time on the SOIC feeder but we did try to optimize it and at that point it still flipped the chip etc 10% of the time which meant the machine needed a constant presence/intervention for the entire build.

[thommo]

hmmm
All I can say is that you should hold in there on that one a bit longer - for reasons I can't disclose, I'm bound not to say any more.

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

My biggest concerns with the machine at this point are not the software, it is that the feeder hardware simply isn't good enough. They seem to work well enough with 0805 sized parts but the 0603s we have done have not gone as well. The feeders just don't seem smooth enough and cause parts to jump out of the pocket, flip over etc. We didn't spend tons of time on the SOIC feeder but we did try to optimize it and at that point it still flipped the chip etc 10% of the time which meant the machine needed a constant presence/intervention for the entire build.

[thommo]

Any chance you can post some video of the FEEDER ISSUE elmood?
Odd this doesn't appear to occur in their videos, but I guess it's a bit difficult to tell for certain.

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

My biggest concerns with the machine at this point are not the software, it is that the feeder hardware simply isn't good enough. They seem to work well enough with 0805 sized parts but the 0603s we have done have not gone as well. The feeders just don't seem smooth enough and cause parts to jump out of the pocket, flip over etc. We didn't spend tons of time on the SOIC feeder but we did try to optimize it and at that point it still flipped the chip etc 10% of the time which meant the machine needed a constant presence/intervention for the entire build.

[elmood]

thommo: It's TheSteve you want on that one... so far I haven't had any weird feeder problems. It does definitely take some tweaking of the settings and for ages I couldn't figure out how to set the length of the feed advance... it's called "feed rate" which to me doesn't sound like "distance" but is.

[vonnieda]

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

My biggest concerns with the machine at this point are not the software, it is that the feeder hardware simply isn't good enough. They seem to work well enough with 0805 sized parts but the 0603s we have done have not gone as well. The feeders just don't seem smooth enough and cause parts to jump out of the pocket, flip over etc. We didn't spend tons of time on the SOIC feeder but we did try to optimize it and at that point it still flipped the chip etc 10% of the time which meant the machine needed a constant presence/intervention for the entire build.

I bet it's the peel box causing the issue. The feeder should be able to feed smoothly enough but the peel box is designed in such a way that it's going to be hard for it to peel smoothly.

I bought one of the feeders and peel boxes and tore it apart. The peel box uses what I guess I would either call a solenoid or linear motor to advance a one way meshing gear. As best as I can tell the peel box just slams the coil to one side to advance the gear a bit and then is returned by a spring. I assume it does this as many times as needed to peel the tape the amount required.

I took some pics of the thing: http://imgur.com/a/osEGC

Jason

[thommo]

Nice series of pics there Jason!

Tell me, what is it you do, for which they pay you?
How long have you had the OpenPnP project going for?

Finally, do you have anyone in mind that this N4 group could task [paid] with the job of creating a solution for their commercial requirements, and who you've happily worked alongside previously?

I think there's a very real chance of getting a 3rd party N4 SW solution up and running.

TheSteve  - I also wanted to thank you for sharing the pictures and your experience with the machine today 

Sounds like things are getting better once all the quirks get acknowledged and worked out.

Plus it looks like were getting closer to some possible software improvements even without Neoden helping out which is a step in the right direction.

...
I took some pics of the thing: http://imgur.com/a/osEGC

Jason

[vonnieda]

Nice series of pics there Jason!

Tell me, what is it you do, for which they pay you?
How long have you had the OpenPnP project going for?

Finally, do you have anyone in mind that this N4 group could task [paid] with the job of creating a solution for their commercial requirements, and who you've happily worked alongside previously?

I think there's a very real chance of getting a 3rd party N4 SW solution up and running.

Hi thommo,

I'm a software developer by trade. I mostly write Java and JavaScript for my day job, lots of web based work.

I've been working on OpenPnP for about 5 years. I started it because I needed a cheap pick and place machine to build some boards I was selling. It turned out to be a bigger project than I thought it would (like they all do) and since then it's become a passion of mine. It's a side project that is more like a second job

Unfortunately I don't have anyone I can recommend to take on the commercial aspect. I'd suggest you join the OpenPnP mailing list and ask there. There is a large audience and there may well be someone who is interested.

Jason

[thommo]

Thanks Jason,

I'll hold on and see where this excitement leads in the short term. Right now it feels good and it may get up. Thanks for joining the conversation.

Do you think you could possibly find more time if, say, the group sent you a Neoden 4 as payment - in advance?

Nice series of pics there Jason!

Tell me, what is it you do, for which they pay you?
How long have you had the OpenPnP project going for?

Finally, do you have anyone in mind that this N4 group could task [paid] with the job of creating a solution for their commercial requirements, and who you've happily worked alongside previously?

I think there's a very real chance of getting a 3rd party N4 SW solution up and running.

Hi thommo,

I'm a software developer by trade. I mostly write Java and JavaScript for my day job, lots of web based work.

I've been working on OpenPnP for about 5 years. I started it because I needed a cheap pick and place machine to build some boards I was selling. It turned out to be a bigger project than I thought it would (like they all do) and since then it's become a passion of mine. It's a side project that is more like a second job

Unfortunately I don't have anyone I can recommend to take on the commercial aspect. I'd suggest you join the OpenPnP mailing list and ask there. There is a large audience and there may well be someone who is interested.

Jason

[rwb]

vonnieda - Out of curiosity how did you end up over here in this thread? Did we summon you with our combined energy? 

[vonnieda]

Thanks Jason,

I'll hold on and see where this excitement leads in the short term. Right now it feels good and it may get up. Thanks for joining the conversation.

Do you think you could possibly find more time if, say, the group sent you a Neoden 4 as payment - in advance?

Thommo,

I'll have to get back to you on that. I'd like to collect some more information about the machine before I'd be able to commit to doing a full port for it. While I think it's unlikely that there's any security features, I'd hate to find out I can't reverse engineer the protocol for some reason.

It's an interesting thought though, so let's let this bake for a little while and see what comes of it.

Jason

[vonnieda]

vonnieda - Out of curiosity how did you end up over here in this thread? Did we summon you with our combined energy? 

rwb,

It's a fairly safe bet that if someone on the Internet is talking about pick and places machines I am probably lurking somewhere nearby

In this case, someone from this thread posted it on the IRC channel several weeks ago and I've been keeping up with the posts since then. Mostly I've just been enjoying the conversation, but when the topic turned specifically to writing pick and place software I figured I should speak up!

Jason

[thommo]

Jason - where are you based?

Thanks Jason,

I'll hold on and see where this excitement leads in the short term. Right now it feels good and it may get up. Thanks for joining the conversation.

Do you think you could possibly find more time if, say, the group sent you a Neoden 4 as payment - in advance?

Thommo,

I'll have to get back to you on that. I'd like to collect some more information about the machine before I'd be able to commit to doing a full port for it. While I think it's unlikely that there's any security features, I'd hate to find out I can't reverse engineer the protocol for some reason.

It's an interesting thought though, so let's let this bake for a little while and see what comes of it.

Jason

[vonnieda]

Jason - where are you based?

Seattle, WA, US

[Ichan]

@TheSteve, thanks a lot for sharing you experience - at least my confidence goes up a little that i will not loose the bet.

My biggest concerns with the machine at this point are not the software, it is that the feeder hardware simply isn't good enough.

Good news and bad news...

...so far I haven't had any weird feeder problems.

Turned back to be good news. So where is the feeder weirdness? I saw feed and peel "strength" setting, what is that - speed setting?

-ichan

[Ichan]

Hmm... seems we have to ask Neoden for their recommendation about these feeder setting related with components / reel type.

-ichan

[Smallsmt]

If you like to adopt our machine to an open source software we have all hardware descriptions and controller commands for our machine
And it's possible to control all speed and acceleration settings for any motor in our machine.
The vision is based on USB digitizer so it's easy to adopt to OPEN CV we use too.

Best regards
Michael

[mikeselectricstuff]

On the other hand, we just need to identify a capable image-recognition source that will do what is required - even if it needs to be purchased as a license.

What's wrong with OpenCV?