Neoden 4 pick and place

[Brumby]

"Bad SW/FW means Bad elsewhere"
I don't subscribe to this view.

Neither do I.
A lot of Chinese hardware is "acceptable" or above,  but a poor UI is indicative of poor  quality, poor design and lack of testing  throughout the software.

+1

With hardware, capabilities are pretty well defined by mechanical elements.  Maximum speed, precision of movement and so on.  Getting something built to physical standard means there are things it is able to do.

Getting it to do them is another challenge.

The software (and I include firmware here) is a completely different kettle of fish to mechanical construction and it is the one component that, especially in today's machinery, is key to getting the most out of any unit.  It has the ability to have been written poorly or well.  It has the ability to support many features or few.  It can be fast or slow.  It can be awkward or easy to use.  It can be robust or fragile... and the list goes on.

One question is whether the manufacturer is looking to market a mechanism or an optimally functioning piece of equipment.

If a mechanism, then they've already done that (it seems) - and the supplied software is simply to demonstrate it isn't a glorified paperweight.  The vision may be (just hypothesising) that the software nerds out in the world will look at this as a juicy project - and delve into a (possibly open source) project to make it sing and dance, which will add enormous appeal for purchasing of the mechanism.

If they want to provide a turn-key package solution, then they really need to get a serious programming effort happening.

[mrpackethead]

"Bad SW/FW means Bad elsewhere"
I don't subscribe to this view.

Neither do I.
A lot of Chinese hardware is "acceptable" or above,  but a poor UI is indicative of poor  quality, poor design and lack of testing  throughout the software.

+1

With hardware, capabilities are pretty well defined by mechanical elements.  Maximum speed, precision of movement and so on.  Getting something built to physical standard means there are things it is able to do.

Getting it to do them is another challenge.

The software (and I include firmware here) is a completely different kettle of fish to mechanical construction and it is the one component that, especially in today's machinery, is key to getting the most out of any unit.  It has the ability to have been written poorly or well.  It has the ability to support many features or few.  It can be fast or slow.  It can be awkward or easy to use.  It can be robust or fragile... and the list goes on.

One question is whether the manufacturer is looking to market a mechanism or an optimally functioning piece of equipment.

If a mechanism, then they've already done that (it seems) - and the supplied software is simply to demonstrate it isn't a glorified paperweight.  The vision may be (just hypothesising) that the software nerds out in the world will look at this as a juicy project - and delve into a (possibly open source) project to make it sing and dance, which will add enormous appeal for purchasing of the mechanism.

If they want to provide a turn-key package solution, then they really need to get a serious programming effort happening.

I think this applies to all the Manufacturers of these kind of systems..     Seems an awesome Opensource project?

[Brumby]

I think this applies to all the Manufacturers of these kind of systems..

Yep - I was pretty much stating the obvious.

Seems an awesome Opensource project?

If Neoden don't step up to the plate, that's what I'm thinking.

[mrpackethead]

I think this applies to all the Manufacturers of these kind of systems..

Yep - I was pretty much stating the obvious.

Seems an awesome Opensource project?

If Neoden don't step up to the plate, that's what I'm thinking.

openpnp.org ?

[Brumby]

Sorry - I'm not familiar with specific groups or organisations, just the concept of open source.


Perhaps others might want to comment on the best way forward.

[mrpackethead]

Sorry - I'm not familiar with specific groups or organisations, just the concept of open source.


Perhaps others might want to comment on the best way forward.

I was just pointing you to what looks like an open source project that is already making good progress on this exact problem.

[Brumby]

Oh.  Sorry.

Thank you.

[Kjelt]

It would be a great improvement if Chinese hardware manufacturers would team up with a western software house ,

I agree but it's not going to happen - cultural thing about not losing face.

I wonder if it is cultural?
Perhaps (speculation) that Chinese people are happy to pay for tangible stuff like hardware but are reluctant to pay for (good) software?
The latter being easy to copy (without costs) and seeing the amount of illegal software in China I wonder if this is culturally defined?
To make things clear I am not judging here, just wondering if this is the current mindset.
Any native Chinese resident here that can/wants to comment ?

[coppice]

It would be a great improvement if Chinese hardware manufacturers would team up with a western software house ,

I agree but it's not going to happen - cultural thing about not losing face.

I wonder if it is cultural?
Perhaps (speculation) that Chinese people are happy to pay for tangible stuff like hardware but are reluctant to pay for (good) software?
The latter being easy to copy (without costs) and seeing the amount of illegal software in China I wonder if this is culturally defined?
To make things clear I am not judging here, just wondering if this is the current mindset.
Any native Chinese resident here that can/wants to comment ?

This is a factor. Most of the people freely copying licenced software in China are outraged if someone tries to walk off with the media it is on. The media is someone's property, and is to be respected.

[mrpackethead]

It would be a great improvement if Chinese hardware manufacturers would team up with a western software house.

I think this is what the SmallSMT people have done. The Software dev is done in Germany and and the hardware is built in China.     The Software doe'snt have the typical Chinese-isms about it from what I can see.

[uncle_bob]

It would be a great improvement if Chinese hardware manufacturers would team up with a western software house ,

I agree but it's not going to happen - cultural thing about not losing face.

I wonder if it is cultural?
Perhaps (speculation) that Chinese people are happy to pay for tangible stuff like hardware but are reluctant to pay for (good) software?
The latter being easy to copy (without costs) and seeing the amount of illegal software in China I wonder if this is culturally defined?
To make things clear I am not judging here, just wondering if this is the current mindset.
Any native Chinese resident here that can/wants to comment ?

Hi

Not native Chinese, but I've spent a lot of time in the middle of this sort of thing with them.

Often they are "hardware guys". Physical devices make sense to them and they turn out a fine gizmo. That's where they stop. They may not even sell the device themselves. They are specialists at a very narrow part of the process, but quite good because of it.  It is not uncommon for a large customer to buy the hardware and write software themselves. The big guy *was* building pick and place gear from scratch, so buying it outside is a big step for them.

Yes, it's a different approach.

Bob

[MTNELECTRONICS]

So far I have placed about 10000 chips with my machine, including some panelized boards with very close component spacing at 360 degree rotations.  The machine has done very well, but as others have noted the software leaves something to be desired.  The general interface is OK once you get used to it, but there needs to be some improvements on relative component location.  I have been doing the math by hand to make it work, but the software could easily have this function built in. 

It seems to be the typical case of a lot of these types of Chinese products---they go about 90%-95% of the way to producing an excellent product, but never reach 100%.  Like Mike has noted several times with the video; it has good production values with good shots, good graphics, etc., but they couldn't bother to have a native proofread and edit the script---so what could have been a good video is actually laughable.  This machine is similar---the hardware seems to be decent, the vision system for all of its simplicity has seemed to work well, yet the software is lacking some very obviously needed functionality, UI, and bug fixes---all things that would readily apparent and obvious to anyone who actually sat down and used the machine for a few hours. 

I am fairly happy with the machine, but hope that there are some software updates.  That said, I would never advise someone to purchase a product like this based on any hope or promise of future updates.  Only buy this machine if you are OK with working around its current limitations. 

[rwb]

@MTNELECTRONICS   Post some finished board pictures if you don't mind.

How often does the whole job run without any intervention needed by you?

Also after you learn how to work around the software does it run pretty smoothly?

[MTNELECTRONICS]

@MTNELECTRONICS   Post some finished board pictures if you don't mind.

How often does the whole job run without any intervention needed by you?

Also after you learn how to work around the software does it run pretty smoothly?

I'll take some pictures after my next run.  Those ones are already all reflowed, and once reflowed you can't really tell how accurate it was because everything sucks into its proper place.  I will say that for everything 0805 and above the placement has been visually very accurate and very repeatable across several boards using all four heads, and my farthest panel components are positioned about 250mm away from the camera. 

The vision system was really useful on these boards because there are 32x SOT89 chips arranged in two rows fairly tight 360 degree circles---they are so tight that they nearly touch when installed.  The components are actually quite loose in the tapes, so no matter how good your machine is there is a massive amount of variation in the pick rotational angle.  I was using three heads for them and the placement was good enough for none of them to ever hit each other and for all to reflow without any issues.   I had tried to do these on my Liteplacer PnP and gave up.  Even at slow movement speeds there was just too much variation in the pickup for it to work without rotational correction after the pick. 

After working out all of the initial bugs (feeder settings, pick height, placement height, etc.) it ran almost continuously without any issues.  The machine has a very hard time with 0805 capacitors but I highly suspect it is because I do not have the proper nozzles (they shipped me two sets of the size bigger than what 0805 really needs instead of one set of each size) so I've had to baby it along with 0805 caps and deal with a lot of misfeeds and slow movement until I they send me the proper nozzles.  I'll update this when I receive them. 

I haven't tried any 0603 or 0402 yet because as I mentioned above, I don't have the proper nozzles at the moment.  With the ones I have an 0603 would just get sucked inside!  I am SOL for a few weeks as they are away on holiday.  Very disappointing, but it is what it is at this point. 

The software hasn't crashed on me or shown any glaring bugs, but it really does leave a lot to be desired.  There are lots of functions that are not clear or require work arounds.  The workflow is making some sense to me now, but it is convoluted.  So far, once I get it going it works, but I question whether or not anyone who designed the software has actually used it on the actual machine, and if they have then their level of expectation for a $10K piece of hardware is very low.  Stupid stuff like having to always resize the columns just so you can see the information in the X and Y columns all adds up to a bit of frustration after a while (if you watch the "training videos" you'll see what I'm talking about). 

I also wish that there was some more user configurable options such as a user-variable speed for pick and especially place; it is extremely fast and fairly smooth, but with some components a slower speed would be nice.  Options for error handling during placement would also be nice.  There needs to be an easier way offset fiducial to components without having to resort to a calculator or spreadsheet. 

In the end it seems that if the machine keeps on working the way it has been I will be satisfied with it for what I am currently using it for.  That said, it is definitely not a perfect machine and some of the lack of polish, especially with the user interface, is readily apparent (I knew this from watching the videos and reading the manual).  While I hope that they make some improvements to the software, I did not buy the machine expecting it, and I do not advise anyone to purchase a machine like this hoping for anything more than what it does already at the time of purchase.

I have been following the SmallSMT thread and like what I see there as well.  I think that if I had known more about that before I had committed to the NeoDen I would be seriously considering one of those machines as well.  They seem well built, have more feeder options, higher maximum component height, and an owner/engineer who actually listens to customer requests and complaints and is capable of providing solutions---which is perhaps the most valuable aspect of all. 

[rx8pilot]

The software on my machine is seriously convoluted as well and it costs far more than an N4. The machine is not new and the software has been around since about 2008 or so - still current and sold today. So convoluted in fact, that I believe I have never seen worse. It is very flexible and quite reliable, just excruciatingly non-intuitive.

With that said, I have learned how to make it work over time. It is just bizarre. It's not Chinese either, is all USA convolution. I only have $10k in it, but they are selling them for $50k with the same feeder package I have. Not sure how many they are selling, but that is the asking price.

[thommo]

MTNELECTRONICS - thanks for the extensive update. Really enjoying all the learning.

What, if any, of these issues you've raised, have you sent to the guys at Neoden, and what has been their response so far? [understanding that they're all enjoying the CNY holidays currently]

As I've said previously, we don't have a machine yet, but I recall there being information in a video regarding:


setting up panels
use of fiducials
identification of 1st part on each of 3 boards in the panel array


Without fully understanding what you're referring to when you say "an easier way offset fiducial to components without having to resort to a calculator or spreadsheet", I imagine that there is presently no link between the panel fids [or even the individual PCBs'] and the components to be placed.

How would it be if you nominated the 1st component in the PnP list [which the GUI asks you to identify], as being a fiducial - with the property 'do not place'. This sounds like it may avoid the current situation and issue you're experiencing. It's a 'work around type of solution' I know, but I think all systems have their own idiosyncrasies.

To the FORUM generally:
Has anyone from the current 'owners' group' begun compiling a list yet? If so perhaps you'd consider publishing it. Others could then 'add' to it even if only by copy/paste initially. This would be tremendously helpful - for the Neoden guys, all the current owners, and the new/potential ones alike.

1. identified bugs w/repeatable procedure steps
2. GUI improvement/modification
3. new feature - mandatory
4. new feature - wish list

[rwb]

@MTMElectronics Thank you very much for the review! Looking forward to the pictures and updated feedback after you get the new nozzles.

I also think the Small SMT machine hits a sweet spot for many different things including the better software.

And the M10V PNP also looks great but more expensive and slower.

[rwb]

Bump.

Any of you NeoDen 4 users have any more feedback about the machine and how its working out for you?

[elmood]

About a week later than we expected due to some things, but we finally tried to make a board today! The delay was mostly caused by the Puhui T-962C oven we bought along with the Neoden 4 being utter crap. I don't want to derail this thread, but basically don't buy one. Neodentech should offer something a bit less horrible. Things that get hot shouldn't be made with adhesive tape that burns while the product is operating normally.

The first run went somewhat okay, but not great. We had a few placement problems because the PCB was actually not located far enough into the machine. It's important to check that the entire PCB is within the movement area for all nozzles or else there will be an error during the job. Simple, but wasn't obvious until about half way through the job.

Another problem (since we restarted after a few errors) seems to be that the software lets you restart part way through the job (cool feature!) but it seemed like it would start at the component before the one I selected in the list. I didn't recheck this yet but it seemed weird when we were using it.

The main issue is that parts on the right half of the board were placed incorrectly. I've double checked all my math and the coordinates SHOULD be right. The Y positions were bang on, but the X is off by about half an 0805 resistor width.

We also placed a TQFP part and it was in a completely wrong spot. I'm not sure if it didn't image correctly but it was about 0.25" or more off. It was coming from a tray, so I'm not sure I'm using the trays correctly. The user manual has a lot of nice screen shots but the text is more or less unusable.

Also, when stopping a job the machine appears to drop the part randomly, instead of putting into the trash box. It also kicks the board out, but not enough to actually reset it. (you have to put the board outside the machine so it can draw it in on the rails for the edge detection to work properly)

Anyway, I think we're getting there... for a first try I'm not completely discouraged, but there is definitely some more learning to do.

BTW, if anyone has tips on using a manual solder stencil printer I would be glad to learn more.

[uncle_bob]

Hi

Usually a part like your TQFP that comes up 1/4" off from all the others is a part coordinate issue. More or less, one side of the world talks about the center of the part and another side of the world talks about the part corner right by pin one. The net result is that the space mission misses Mars. I'd check your input data files and see what they say. The offset is about right if it is a 12 to 14 mm square part.

I don't have a Neoden so I can't really dig any deeper into the issue. No idea about how they handle the coordinates.

Bob

[rwb]

Thanks for the Update Elmwood 

Keep us updated as you progress forward.

[TheSteve]

I assume you're using the camera with the larger parts - make sure they are being detected/centered properly. You may need to fine tune the camera brightness like we did with the SSOT28 parts we use. And as I've mentioned before be sure to do a nozzle calibration.

[elmood]

TheSteve: Is there a way you've found to prevent multiple parts from being picked up at the same time? I found that the TQFP worked perfectly when it was the only part on the head. Previously my job had two different nozzles in the chiplist one after the other. The TQFP got picked up first, followed by a SOIC on a different nozzle. When it went to image the TQFP the other chip messed up the detection and then it placed it in the wrong location.

[TheSteve]

TheSteve: Is there a way you've found to prevent multiple parts from being picked up at the same time? I found that the TQFP worked perfectly when it was the only part on the head. Previously my job had two different nozzles in the chiplist one after the other. The TQFP got picked up first, followed by a SOIC on a different nozzle. When it went to image the TQFP the other chip messed up the detection and then it placed it in the wrong location.

Yes, you select if you want the parts picked/placed "jointly" or not. Pretty sure you can also select the camera to operate per nozzle or "jointly" as well. We do all of our camera parts one at a time.

[elmood]

I selected both individual and large component in the case of my TQFP part and it seems if adjacent parts in the pick list use different nozzles it will pick up multiple parts at once and then go and image them one at a time. But because there are multiple parts on the head at once the vision code cannot deal with the extra part showing up in the image.

Also, even after running the calibration of nozzles my resistors and small parts still appear to be placed incorrectly. I'm not sure where the error is coming from. I carefully set the first chip location and believe that the mark points are correct.

So those are my two currently issues. I'm not sure it's worth running another board with paste until I can figure them out.