Desktop Pick-and-place: overview

[Spikee]

NeoDen 5
http://pickandplace-neoden.en.made-in-china.com/product/ReAxBJYkVNGL/China-Neoden-5-Visual-Pick-and-Place-Machine.html

They are integrating their two part feeder now in one portable feeder. That can be plugged in like the CL feeders.
Expected release Q1 /Q2 2017.

Changes vs neoden 4:

Code: [Select]

A.All-in-one feeder,same as the one that you have seen in our factory last time,the feeder will be used electronic all-in-one type,NeoDen will offer 8mm/12mm/16mm firstly,then will consider 24mm and 32mm for customizing in the near future to meet most of the requirements of customers if possible;
B.The bracket of feeder will be used removable type,convenient for users to install and change the tape reel;
C.The up-looking camera will be used 4 cameras,to save the time while photographing,ensure higher production capacity finally;
D.The bottom cover will be increased by 12cm in height,convenient for agents and end-users to do the service and change spares once problems happened;
E.The weight of working table will be increased,to ensure the stability while machine working;
F.The plastic cover will be improved to be stronger and unbreakable;
G.The machine will stop working once plastic cover opened;
H.Tricolour light will be equipped on the machine,green light indicates machine working normally,yellow light indicates machine may have some errors,red light indicates machine have some problems.
Price is unknown at this moment. But based on that above link 7500 USD for machine with no conveyor, no stand and no included feeders would be a good guess.
Changes seem minor. Unless you plan on changing feeders very often I don't see much of and advantage. The new feeders
might also be a bit more expensive than the yamaha CL (copies) (pure speculation).

--
The cons of these customized feeders is that if a product goes uit of production or if you want to upgrade your machine
you will be stuck with them. With yamaha feeders you can migrate to many diffrent PNP brands or even to a yamaha machine.

[Prohiottoka]

I've been asked to write a quick follow up about the CHMT-530P4.

We are definitely happy with it. There are problems of course, but we can live with it.
We are using it for a while now, but we did not tested the limits of the machine. The finest components what we place with it are 0603 and sot363. You need to check the board before reflowing, the machine does make mistakes. (with using the camera function)

You don't want to order the feeders from them if you plan on placing fine components (or at least we did not receive the good ones) They are working, but not calibrated at all.  But again for that price, it can still worth it.

I believe they are in the beginning of making a fine product. With minor improvements the machine can be much better. And we know they are working on improvements. So maybe what you can buy now is even better than what we bought.

We plan to buy another machine in the beginning of 2017. I think we will buy another one from them.

[cmantunes]

There are problems of course, but we can live with it.

Would you mind sharing the problems, even if minor, you are experiencing?

Thanks!

Carlos

[Prohiottoka]

We are having a conversation with the supplier, and they promised us to send sw updates based on our suggestions.
So until we can test the new sw version, I don't want to complain about it in a public forum We have placed more than 1 000 000 parts, we are still satisfied with the price/performance ratio.

I think the biggest hw faults are:
- there is no lubrication for the feeders.
- there are only 2 vacuum sensors for the 4 heads.

There are problems of course, but we can live with it.

Would you mind sharing the problems, even if minor, you are experiencing?

Thanks!

Carlos

[mikeselectricstuff]

- there is no lubrication for the feeders.

You need to be careful with lubricant near a feeder, as paper tapes shed dust, which can mix with oil and form a sticky goo

[cmantunes]

I think the biggest hw faults are:
- there is no lubrication for the feeders.
- there are only 2 vacuum sensors for the 4 heads.

Thank you!

One quick question: is the dpv file a text file, something one could create with an Eagle script, for example?

[arcircuit]

Wanting to know if anyone knows of someone who would come out to our shop to help us learn how to use this the first time. We have a very old pnp machine and it will definitely be a learning curve for us.

[Prohiottoka]

Yes it is text. And quite understandable, so I think you can do it.

[markk]

Hi everybody

big thanks to the forum! I've spent hours if not days here and learned a lot.

Background

I'm an enthusiast hobbyist doing quite complex circuits. Up until now I hand placed everything using the pictured cartesian guide. I'm awful at it! It took me 10 hours to place these two boards (mostly 0603, 0.5mm pitch, 400 parts per board, 100 different ones, yes, it needs simplifying). Paste melted to puddles!  Had to do lot of rework. Circuit mostly works though 


Now I'm in the market for a Pick and Place machine. I'd like to do prototypes (two, three boards) and very small runs (max twenty boards per go, I guess).

Main dilemma

I'm a bit confused by the (lack of) distinction between prototyping and production.

Everybody is talking about prototyping, but it seems to me all the machines are more or less straight production machines (or try to be). All the talk is about reel feeders and while there is the odd tray or cut strip holder available, nobody in the commercial market seriously seems to address the fact that one has to place say 50 different parts from small quantity feeders. 

Like I said I'm not a professional but I can't imagine real pros will buy a full reel of parts when they prototype some new design, do they? I typically buy some 10 parts perhaps (costly ICs) that's a very short cut strip. Even if the parts are very cheap, does it make sense to buy a full reel for an odd value precision programming resistor?

Even if the pros don't bother about the cost or the storage space (or the environment), is it worth the hassle to fiddle with the reels? (remember: you also have to unload another reel you will surely miss later!)

Even if your machine has cassettes, is it worth having a gazillion spare ones for every odd part? Even if the cassettes only cost $50, this cost veeery quickly adds up!

So I'm a bit astonished that none of the commercial machines seem to address this in a practical way.

What am I missing? 

OpenPNP vision assisted feeders

Only OpenPNP really seems to go practical. That’s the way I envision feeders for prototyping:


Or even lose parts:


The problem: nobody offers a quality assembled OpenPNP compatible machine. The liteplacer kit is the only reasonable option for those that really don't want to get sidetracked constructing their own machine (you still have to assemble the kit, buy power supply and cabling, build a working table with up-camera inset, etc.). It sure is a great machine to get started but having zero reel feeders is just as bad as having only reel feeders. Also I wonder if the rather low speed will be a limiting factor.

OpenPNP on commercial machines

I see some commercial machines have been converted, mostly due to their built-in software being 

The TVM920 port is noteworthy (watch from 0:55 for speed):
https://youtu.be/Es060QcynB0 
https://groups.google.com/forum/#!topic/openpnp/UA8h0TvXpKg

The machine also has a large working area for custom feeders, trays etc. But it is too expensive for me. Especially if you include shippings costs and a reasonable stash of feeder cassettes to go with it. The machine is also very large and incredibly heavy and it requires rear access space.

If anybody knows another port, please point it out to me, thanks! 

Reference:
http://openpnp.org/hardware/


Porting another machine

Looking at the other machines, there seem to be some "cost stepping effects". You get the ~$3000-$4000 class and then nothing and then the $7000-$10,000 class.

The lower class seems to make you unhappy in the long run, it seems, judging from this forum. Even former enthusiast are ground down. Too much fiddling with the cheap feeders? Too little space for cut strips?

The higher class simply seems over-engineered for ridiculous speeds. If the user reports in this forum are in any way representative, this top speed will hardly matter as the machine will be idling most of the time while the user-slash-baby-sitter is wrestling with some of the many, many feeder, cover tape, nozzle and whatnot problems.

If I understand things correctly, the speed or rather acceleration, jerk, snap, crackle and pop all work against precision and in order to still comply, they have to throw tougher mechanics at the machine, increasing cost, increasing weight, more cost, more weight...

The market seems to design for "sports car" emotions, not for function   

Neither class is right for OpenPNP. Or am I missing something? 

Any insight here would be very welcome!


Missing combination

There should be a ready to buy machine that combines a wide but light mechanical construction with a reasonable count of cassette feeders (i.e. Yamaha CL). Perhaps as an addon module. The machine should be "desk against wall" compatible, reasonable depth, cassettes loaded from the front. A machine two persons can carry through normal doorways.

Why not put some barcode style visual rulers on the PCB holder, all around. Make up for any "wackiness" due to mechanical slightness with full scale top camera assisted calibration every once in a while. Make those rulers of PCB material and they'd even be temperature compensated.


My own evaluation

It is now clear that there is currently no ideal machine. Should I give it up? Or choose from a less than ideal machine?

Personally I currently favor one of the SMALLSMT machines. Straight from the above spec, the VP-2500LED seems to be the "best of the bad" matches. The 4 head machine has 16 CL feeder slots and sports a large moving area it seems. Sadly cassettes load from the rear. Also I don't need 4 heads (if there is a nozzle changer). Again the machine seems over-engineered for my needs. The cost is way above of what I originally envisioned.

The VP-2500DP-CL16 has only two heads and adds two banks of push feeders.  Perhaps I could order it with only one push feeder. Or remove one later and keep it for future/changed needs. This would add some space for cut strips.

The VP-2000S would be the financially sensible compromise. But it seems just too small. No cassette feeders.

Michael from SMALLSMT has promised to release a HAL DLL that would allow porting OpenPNP. I'm currently negotiating to get the API before I buy the machine, no luck so far.

The software that comes with the SMALLSMT machines seems very good compared to all the competitors, having many of the features that are missing for others. The machine control is optimized and very fast and I see it could work very well for production.

But - again - not so well for prototyping . I'd rather use OpenPNP's cool ad hoc vision capabilities (listed above) and live with the simpler, "stop-and-think-between-moves", machine control.

Have I missed an option?

Thanks for all responses!

-Markk

[mikeselectricstuff]

Automated P&P for 1-off prototypes is rarely useful unless there are a large number of relatively few parts.
The time taken to set everything up will usually be way longer than to manually place, once you have a good manual system.
It certainly isn't useful enough for it to be worth anyone selling a machine aimed at 1-offs.

Manually placing with a foot-operated vacuum pen can be pretty quick - couple of seconds per part when placing parts from the same strip on a board you're familiar enough with to know where they go without looking it up.

Think about what you need to to when manually placing a board, up to the point where you actually place a part - find the right part, take parts out of tapes, make sure it's the right way round etc.   
For an automated system you not only need to do that, but them tell the system where it goes, and put it somewhere it can pick it up.

For more complex PCBs, a small P&P can be useful for placing things that there are lots of - decoupling caps, pullup resistors, LED arrays etc. but for anything there's only a few of, doing it manually is going to be much quicker, with a little practice.

[JuKu]

Automated P&P for 1-off prototypes is rarely useful unless there are a large number of relatively few parts.

Actually, I found the exact opposite. If you have a just a few part types and your software can give you placement printout for each part (mine can), it would be fast. But if you have tens of different parts, most of the time is spent hunting the spot where to place it. Having a computer on the table helps somewhat, but it is still "type a part number, look the screen, find the same spot on board" -game.

... on a board you're familiar enough with to know where they go without looking it up.

With prototypes this is not the case, by definition.

The time taken to set everything up will usually be way longer than to manually place, ...

Depends. To assemble a prototype, you need to get the parts out from your storage to your work area. no way around this. the exception are parts you use all the time, like bypass caps - those are likely ready anyway. Then it is all about the software. If the software needs part size, part heights and so on, or you must use and setup feeders, it could indeed be tedious.

Think about what you need to to when manually placing a board, up to the point where you actually place a part - find the right part, take parts out of tapes, make sure it's the right way round etc.   
For an automated system you not only need to do that, but them tell the system where it goes, and put it somewhere it can pick it up.

Pessimist. :-) Find the tape, put tape somewhere so you can handle it (like with double sided tape to a table or manual tray, place the strip tape to a holder etc). Remove cover tape. These you need to do anyway, manual or automatic. For a machine that has prototype workflow, you then only need to tell which pre-defined spot the tape is (on LitePlacer, two mouse clicks). Knowing which way around it is, where it goes etc is something that the machine already knows from your data.

Besides, not all of us can handle 0603s or 0402s manually to begin with. (I can do 0603s but not 0402s anymore, and I need do everything SMD under a microscope anyway - another thing that makes manual work tedious for me).

(Disclaimer: I'm the guy who built LitePlacer, a pick and place machine specifically for prototypes. Why? Because https://www.liteplacer.com/about/the-story/.)

[nisma]

If doing one off pcb, it is crucial to not neet a lot rework for importing the data.
You need a layout tool that emits real protel centroids. Eagle, Altrium and possibly (not verified) Kicad is fine, Diptrace and EasyEDA as example are know
to not work.
Further it saves a lot of time to have a laser barcode reader with the capability to allows enter function keys.
Annother thing to consider is that you need the ability to split the design to as example 4 or 5 assembly steps.
This for allowing you to setup a limited number of parts, and during it's assembly, you continue to setup the next lot of parts.
Prerequirement is that you'r vision pipeline is setup correctly and bulledproof if using openPnP.
Cost saving could be having auger type dispenser for saving stensil costs.

[nisma]

Back to question, for prototyping, having a big area is really a bonus. Liteplacer works.
Liteplacer have a cost of 1777€+shipping , chinese kit costs 455€ shipped.
I know it is 1699 + cable chain (29) + terminal block (1.9) + ferrule set incl. crimper (26) + nuts(13) + 4x ferrite nut(2.3 each) that makes 1777.8€  Missing is the better uplooking HD camera , roughtly 1800€.
The big difference, with liteplacer you have aftersales support, with chinese kit not at all
including openpnp support list, that explitly don't give support for setting up motion or other important
prerequisites requiring to run PnP software.
If you know someone that is able to setup cnc machines, it is possible to go with the cheaper kit, otherwise it's hard.

[rx8pilot]

Automated P&P for 1-off prototypes is rarely useful unless there are a large number of relatively few parts.
The time taken to set everything up will usually be way longer than to manually place, once you have a good manual system.
It certainly isn't useful enough for it to be worth anyone selling a machine aimed at 1-offs.

Manually placing with a foot-operated vacuum pen can be pretty quick - couple of seconds per part when placing parts from the same strip on a board you're familiar enough with to know where they go without looking it up.

As I am reading this....I am manually placing a 1-off PCB while I sit right next to my perfectly capable P&P system. Plain and simple - it is faster to manually assemble with my vacuum pen and cut tape holders. Way faster, in fact. Before I got my line up and running, I hand assembled everything so I got that process dialed in. I spent around $500 on assembly specific tools and got pretty quick at setting up and placing.

Like Mike said, that may not be true if you have a board with a whole lot of the same part.

[mikeselectricstuff]

... on a board you're familiar enough with to know where they go without looking it up.

With prototypes this is not the case, by definition.

It is if you only designed it a few days before.

[ar__systems]

... on a board you're familiar enough with to know where they go without looking it up.

With prototypes this is not the case, by definition.

It is if you only designed it a few days before.

No way you are going to remember locations of resistors on the board of 100 parts after even few days I usually don't get to assemble boards after just few days any way.

[rx8pilot]

I make a PDF with color coordinated parts. The color matches the label on the pickup tray. Once I got the template - it does not take long to make one for an all new design and it makes the process very easy. I made many 100's of double sided high-density PCB's this way before I got my P&P machine up. For a true one-off effort - I simply pull up the layout on the computer screen right in front of me and start going. Click the part to ID - place the part. Repeat until all parts are down.

[ar__systems]

Click the part to ID - place the part. Repeat until all parts are down.

Pretty much the way I work. Except that I have a BOM application that sends command to CAD s/w to highlight all parts of the given nominal.

PDF thing is good but there are not too many clearly distinct colors. Similar colors will create a room for confusion...

[markk]

Thanks for all the answers and insights. Explains a lot.

Manually placing with a foot-operated vacuum pen can be pretty quick - couple of seconds per part when placing parts from the same strip on a board you're familiar enough with to know where they go without looking it up.

Wow, it surely takes me much, much longer even if I know perfectly where the part has to go. And most of the time I don't - or I'm too unsure not to check. After one two hours of doing this, it becomes a pain. Muscles tense up, eyes hurt with straining, etc. Solder paste turns to puddles. 

... for anything there's only a few of, doing it manually is going to be much quicker, with a little practice.

Well, I fear it's more about talent and patience and perhaps old age   
(though I never was the "manual" nor the "precision" guy... guess, there's a reason I went into software in my professional life)

... If you have a just a few part types and your software can give you placement printout for each part (mine can), it would be fast. But if you have tens of different parts, most of the time is spent hunting the spot where to place it. Having a computer on the table helps somewhat, but it is still "type a part number, look the screen, find the same spot on board" -game.

I make a PDF with color coordinated parts...

Yes, I made my own little helper. An interactive HTML tool for EAGLE. You can click on the parts list to highlight the parts or click on the part on the PCB, to scroll to the parts list line and highlight it:


I just put it on github, see here:
http://makr.zone/pcb-manual-placement-helper-in-html/39/
https://github.com/markmaker/pcb-placer

Without this I would be completely lost. But sadly it does not place the parts by itself, does it? 

Find the tape, put tape somewhere so you can handle it (like with double sided tape to a table or manual tray, place the strip tape to a holder etc). Remove cover tape. These you need to do anyway, manual or automatic. For a machine that has prototype workflow, you then only need to tell which pre-defined spot the tape is (on LitePlacer, two mouse clicks). Knowing which way around it is, where it goes etc is something that the machine already knows from your data.

Besides, not all of us can handle 0603s or 0402s manually to begin with. (I can do 0603s but not 0402s anymore, and I need do everything SMD under a microscope anyway - another thing that makes manual work tedious for me).

(Disclaimer: I'm the guy who built LitePlacer, a pick and place machine specifically for prototypes. Why? Because https://www.liteplacer.com/about/the-story/.)

That's exactly the way I feel. Thanks!

I'm more and more inclined to buy a liteplacer. If there were some real work videos of it placing complex boards, with some DFN, QFN stuff perhaps - ideally also with OpenPNP/bottom vision - it would help a lot to come to a conclusion. But all I keep finding is "demo", "first place" or "work in progress this or that". Like nobody actually really finished building/improving their machines and finally made some boards! (same phenomenon with OpenPNP in general, actually).

Thanks again,
Mark

[Kjelt]

That looks great Mark, use it from Safari on an iPad so to put it near the pcb and it would be a supertool 

[mikeselectricstuff]

Yes, I made my own little helper. An interactive HTML tool for EAGLE. You can click on the parts list to highlight the parts or click on the part on the PCB, to scroll to the parts list line and highlight it:

Seriously ? Eagle can't highlight all parts of the same type by itself? From what I've seen it eems like you have to DIY pretty much anything useful in Eagle..

[rx8pilot]

That is Eagle. It sucks beyond comprehension.



Sent from my horrible mobile....

[jmelson]

Seriously ? Eagle can't highlight all parts of the same type by itself? From what I've seen it eems like you have to DIY pretty much anything useful in Eagle..

Hmm, Protel 99 can't do that in one click either.  You can do that pretty easily with a couple clicks to get into global editing, but it sure isn't one click.  To use it as an assembly guide, you'd really want it to be one click.

Jon

[Spikee]

Automated P&P for 1-off prototypes is rarely useful unless there are a large number of relatively few parts.
The time taken to set everything up will usually be way longer than to manually place, once you have a good manual system.
It certainly isn't useful enough for it to be worth anyone selling a machine aimed at 1-offs.

Manually placing with a foot-operated vacuum pen can be pretty quick - couple of seconds per part when placing parts from the same strip on a board you're familiar enough with to know where they go without looking it up.

Think about what you need to to when manually placing a board, up to the point where you actually place a part - find the right part, take parts out of tapes, make sure it's the right way round etc.   
For an automated system you not only need to do that, but them tell the system where it goes, and put it somewhere it can pick it up.

For more complex PCBs, a small P&P can be useful for placing things that there are lots of - decoupling caps, pullup resistors, LED arrays etc. but for anything there's only a few of, doing it manually is going to be much quicker, with a little practice.

I use my cheapish pick and place exclusively for prototypes. Generally these have between 50 and 150 passives. With a run of about 2-10 boards.
Since about 80-90% of the placement is passives anyway I just run it trough the PNP. And do the small IC and whatever by hand.
This combination works out great for me.

The situation before was that I / my client had to spend about 300 - 700 usd for small batch assembly in China. And we have pretty much always have issues with this.
Be it small confusions on their end, or messing up parts / plain replacing them with other shit and later on you are debugging why it is not working. Which is also quite expensive.
+ Th lead times are generally long as in 2-4 weeks.

Doing it youself might not be cheaper perse. But everything is in my hands. I can assembly 1 board and test it. And do a new iteration and new assembly 2-4 days later.
A complete run for Ordering parts,  boards, stencils and assembly -> basic test -> shipping to firmware guy / client takes about 1-1.5 weeks.

That is where the strength is with these kind of low end machines In my mind / experience. Even with a reasonable hourly rate as a western freelancer in this area is is about the same price / slightly more expensive to do it yourself than to get a company in Asia do it for you.

The risk is so much lower and the possible issues / communication time / problem fixing time can easily double the amount that you pay for assembly due to the designer having to put extra time and effort in it. I think many people underestimate this possible hidden cost.
 

[JPlocher]

I am a hardware/software engineer in California who develops custom electronics for hobby use.  I produce boards with 603 and 805 components, with no 402s or smaller, 20 to 100 components per board, with a board size sweet spot of 10cmx10cm in job runs of 5-30, several hundred per year.  I am looking at a PnP machine, a stencil printer and possibly an oven (my PID toaster oven works fine, but, well, its a toaster  )

The learning curve for PNP machines seems to be: read thru several days worth of pages here on EEVBlog, watch videos, master Alibaba and Google Translate, do it again because it didn't sink in the first time, bemoan the fact that there is no commercial offering that supports openPNP, then cogitate on it all...

Having done so, I'm coming to a few observations I'd like to validate:

• There is a ~$5k range of systems and a ~$10K range, the difference being primarily drag feeders -vs- Yamaha pneumatic
• Cost goes up with the number of feeders required (for both types...)
• Vision is great, but it competes with placement speed - and isn't really needed for larger passives (etc)
• The provided PnP software is functional, but not very usable - and each vendor has a different, large list of issues. 
• Hardware is generally good to great, but the software limits how well it can be used.
• It is a small world, with vendors and manufacturers all intermingling their product lines and trying to clone/copy the good stuff while improving on perceived shortfalls.

The price point for a usable system (PnP, feeders, Printer, Oven, shipping, spares,...)  is several thousand more than the PnP machine itself:

• $7,200 - $13,000 4 head QiHe TVM920 (Juki/Yamaha) embedded Windows
• $5,000 - $ 8,500 4 head CHMT 530P4 (Juki/Yamaha) embedded Linux
• $4,600 - $ 6,430 2 head Qihe 802B (proprietary nozzles and drag feeders) external PC/WinXP?

The smallSMT systems fit in there at the high end, with their new VP3000 systems looking to come in closer to $15-$20k
The Neoden4 (at ~10k+) looks to be the best hardware combined with crappy feeders and the worst software.

(Probably overly generalizing), I'd say the testimonials here from TVM920 and CHMT530P4 owners seem relatively happy and satisfied, while the Neoden and 802 class owners have more frustration with the software, feeders and component handling.

If the above isn't too overly simplified, the choice seem to really be between the CHMT530P4 and the Qihe TVM920, with the price difference mostly attributable to number of feeders supported by the 920 (56 -vs-30)

Is there anything in the EEVBlog collective mind's experience that would swing the pointer one way or the other, if feeder quantity wasn't a factor?

  -John