NEW!! We just introduced a neat little manual pick and place machine: SMT Caddy!

[tszaboo]

Not hard, but this is less than 5% of what it takes to really sell a product into all states in Europe. How do you handle support? How do you handle service? Do you translate the manuals? Are there any specific health and safety concerns for those given states? What about the EU laws on merchantability and fitness? Can you be sued in a random state in Europe now? Will you need to offer customer support by phone during the European workday?

You are applying your thinking to the Europeans. It doesnt work that way. In the EU you dont sue companies, if your microwave kills your cat. Also, in EU, people have a basic understanding of maps,and realize if a phone number is from the USA.
You can let a company like Amazon handle all the logistics. I can order the EEVBLOG BM235 and the HVP70 from amazon.de. A single person was able to solve all these problems for products, that are required to go through a bunch of certification.
It's just... You know americans USA people think the rest of the world doesn't matter. We get it. So come up with a lot more poor excuse please.

[mikeselectricstuff]

A cheap affordable pick and place machine, great...

I was watching the video and thinking "nice, now he is teaching the machine how to do the first pick and place..." then after some repeat placings I realized it was manual??

I'm sorry, but it seems a bit silly to do all that engineering into what is seemingly well constructed mechanics and then skip the the last step of putting in step motors and control to make it automatic.

My suggestion, add the step motors and control, ditch the custom screen in favor of a mobile application that can be used with any Android device and you have a winner product.

Automating something like this isn't a trivial task. It could easily double or triple the development time and costs.

Using a phone would be a terrible idea, as the latency would be poor

[rx8pilot]

The other hand has nothing to do anyway The foot-pedal cord can become an issue and your foot keeps losing the pedal, unless you use a massive, captive housing flavor ... matter of preference I think ...

Just before I got my P&P line together, I was manually assembling a LOT of boards. I had made up a lot of bit and pieces to help the process and at one point - put together a manual picking machine similar in concept to this one. The first lesson I learned is that two-handed operation was dramatically more accurate and fast than just one hand. I had rigged the system so that the x/y position could be held and fine-tuned with one hand while the other operated theta and drop. Since the system moves so freely, it was really important to have one hand effectively operating as a brake for X/Y movement as you lower the part, make the final alignment, and release it.

Personally, I do not like the foot pedal either. Pushing my foot down would put just enough pressure on my body that I had to compensate with the position of the part. I was in a typical rolling and swiveling lab chair and that may not be a problem if I was more locked down - but I like my moveable chair.

On the business end of the conversation.....

I probably would have been a target customer a while back when I needed a stepping stone to a full P&P system. I lost a lot of nights, weekends, and holidays trying to get a dozen 500 part boards assembled manually. With that in mind - I may have just purchased this without any further thought to ease the painful situation. I started my business by restricting sales to Los Angeles only so that I could personally participate in the customer experience and fix problems as they came up. My customers knew that I was making a new product and they bought into it because I was close by. The feedback and experience I got from that went right back into the design and it improved dramatically and quickly. I slowly increased my reach as I gained confidence in the product and I now sell worldwide - having product on every inhabited continent. Starting off with planned geographical boundaries is not a bad idea.

My issues would have been the number of parts you can get into the machine at a time and how quickly they can be swapped out. My manual process had a combination of loose part trays and cut strip holders that were arranged in the order of the BOM so I just went down the line picking parts with a placement cue system on a monitor to remind me what and where things go. To be honest - I was pretty quick. The parts for an entire board were all on a tray so I could put one tray away and grab another and boom.....ready for the next PCB design. The position of the parts to the board made it easy for the way I moved and I literally free handed the pickups and drops with a vacuum pen and a stereo microscope. This allowed 0402 and .5mm QFN's even after a big cup of coffee. The downside (and it is a big downside) is that I could not pass that process on to someone else very easily. It was a setup based on my own personal preferences for sure.

I am curious if your system would be much faster or easier in the end. It definitely looks less intimidating than what I did in terms of actually placing the parts.

[SMT Caddy]

... I am curious if your system would be much faster or easier in the end. It definitely looks less intimidating than what I did in terms of actually placing the parts.

The one hand vs. two hands went through a war of wits here before we settled on this approach! Our first prototype utilized a two-handed head, similar to a CIF machine (if you're familiar with their products). I agree re. the accuracy with two hands, but, reach becomes a concern since your whole torso tends to travel rather than just the arm. And, longer P&P sessions become rather tiring with the two-handed head; feels like you just came back from the gym! It wasn't an obvious choice to go one direction or the other; it became a democratic vote - this approach won by far better margin than our current president

I think most folks will be sitting in a roll around chair, at least business folks, which makes the foot-pedal quite annoying.

I can tell you that after many similar long hours hunched over a microscope with tweezers in (very cramped) claws, sitting behind the SMT Caddy is a sheer joy! The full panel you see in the video is similar to a 500 count board, and any of us can put one together in roughly an hour (more or less, depending on experience). There are 0.5mm pitch leadless packages and the passives are generally 0603 (0402 doesn't make it much different). I hired a few college student summer interns a couple of years back and trained them to build a batch of boards the old-fashioned way. The yield was at best 60% for functional boards, and maybe 80% after rework of obvious problems. We rarely lose a board in fab with SMT Caddy.

The reason we stayed with 20 resident feeders is primarily reach. e.g., I'm not a big guy, 5' 8", and repeated reaching beyond the farthest feeders will probably make it a bit uncomfortable for me. Feeders swap out in seconds, and, setting up a new tape in a feeder is basically a one minute task. So, batching components and swapping feeders is really pretty quick and convenient, while keeping the machine footprint small. If you're a power user (a PCB prototype assembly job-shop, e.g.), you'd do well to have a bunch of feeders pre-packed with the most common components to reduce the number of tape changeovers.

I've noticed that the arm-rest is used differently by different folks. I like to rest my forearm on the rest and keep the wrist floating; my lefty machinist keeps his distal wrist joint on the rest, etc. Generally speaking, couple-hour-long sessions behind the P&P are quite reasonable. If you need to run with it the whole day, a break here and there is called for. Not having to deal with microscopes and magnifying glasses, and being able to switch between natural view and magnified view by just changing focus, dramatically reduces strain.

A bunch of folks have compared this to an auto P&P; it really doesn't apply! If you're building a few of one board and a couple of another, as everyone in the business does in development or for one-off needs, you'd sit behind SMT Caddy any day of the week before even thinking about approaching your auto P&P! The setup overhead is just not worth it. You need various tools for various needs; this one addresses the particular prototyping need.

The comparable big company systems in the market, selling for similar dollars and way beyond, don't offer a fraction of the features SMT Caddy does. I looked at Novastar, CIF, Essemtec, Manncorp, etc., long and hard and couldn't convince myself to buy any of them, though many businesses do. That's why we decided there was a hole in the market which needed fillin'! It has been a lot of work and investment, and although there's always room for improvement, we're pretty happy with the end result.

[jmelson]

You have very interesting meaning of word "affordable". $2800 is not affordable since many automatic PnP machines are in the same price ballpark.

* asmi went back to manually place parts...

OK, just a point of reference.  10 years ago, I bought a 10 year old high-end pick and place machine, with a bunch of feeders, for $3600.  It handles anything from 0603 up to 20+ mm square ICs.  it is "old school" with no vision, but many machines of that vintage had broken vision systems.  It will do over 3600 CPH on the passives, and slows down to about 600 cph when it needs to use the alignment station for the larger chips.  I've done several thousand boards with it.
I have had to do really minimal maintenance on it, an optical sensor broke, some hoses cracked, that was just about it.

I can't IMAGINE doing all that with a manual P&P, and I didn't end up spending a whole lot more than your unit.  (I DID have to wait for a good deal to come my way, I admit that!)  And, now the Chinese are making a number of full-auto P&P machines at crazy low prices.

Jon

[lundmar]

Automating something like this isn't a trivial task. It could easily double or triple the development time and costs.

That is true. However, I don't think it is as difficult as people might think. I realize that a big part of doing an automatic pick and place machine is creating the control software but there are ways to reduce the software development effort using open source components. As a professional sw engineer my view is of course biased.

Using a phone would be a terrible idea, as the latency would be poor

That is true if you were to use the mobile phone for controlling the picker/placer head position. However, for simple hand-eye-coordination as it is used today, I think the latency is low enough using a decent modern mobile phone. Saving the display would help keep the cost down. But now that they already include a display lets keep it.


I really love the mechanical design of the machine - its a clean an simple industrial design. I also think that a 2nd generation or variant of this machine screams for an automatic pick and place feature.

I noticed that the display device is featuring a raspberry pi linux system. The way I would upgrade this machine to an automatic pick-n-place machine would be to roughly do the following:

Add 2 stepper motors for xy plane position control
Add 1 small stepper motor for picker/placer head height adjustment (z axis height)
Add 1 small stepper motor for picker/placer head rotation adjustment (rotation in the xy-plane)
Add a calibration pad with colored background between reels and PCB table
Add a PCB with:
 - 4 channel stepper motor controller
 - I/O for simple control of reel wheel motors
 - Suitable microcontroller
 - USB interface for connecting to RPI

I would install opencv on the raspberry pi for computer vision control of the picker/placer. I believe the control software can be done relatively simple:

Learning procedure:
1. Register button press to start
2. Register button press to capture pickup position (xyz position)
3. Register button press to capture position on top of colored calibration pad
   - Use openCV to trace the contours of the component to calculate component center in the xy plane.
   - Save outline of contours (for later calculation of required rotational offset, if any)
4. Register button press to capture placement position
   - Use xy-position of calculated component center as target (opencv tracking)
   - Save z-axis height
5. Register button press to end pick-n-place procedure

Pick n place procedure:
1. Pick up component using pickup position
2. Move picker/placer head to calibration pad position
  - Use opencv to trace the contours of the component to calculate xy center offset and rotational offset
  - Rotate picker/placer head if necessary
3. Move picker/placer head into place position in xy-plane
  - Adjust xy position by previously calculated xy-offset
4. Lower picker/placer head to final z-axis to place component

Of course, the procedures can be improved and refined using opencv to e.g. recognize PCB solder pad placements and adjust accordingly but I think the above basic procedures will suffice.

There would also be a few additional steps to control the reel wheel motors to automatically feed components but I believe that is solvable with opencv too.

Just my 2 cents worth

[Kjelt]

Thats all great but steppers and the rest brings the price to $4k or even higher and there are already many chinese and other machines in that marketsegment.

[mrpackethead]

I have a dima FP-600 in my workshop, which has done a LOT of work and it still gets used along side our PNP machines, for various tasks.. ( sometomes an odd part that just can't be machine placed.. )

Having watched your video it seems really slow to use this machine.. The 'double' touch when your picking up just doe'snt seem quite right..     i regulary place quite fine pitched parts and dont' need a camera. Just good lighting.. 

[mpi]

I always get a laugh out of the hand-wringing about "Exports! Compliance!". Meanwhile those of us who do sell globally just figure it out and do it.

Most of these hurdles are based on FUD, "worst case scenarios", or just plain wrong, and (IMHO) just serve as a handy excuse for those who don't want to go that last painful step -- selling (I was one of these people for the longest time).

As for the SMT Caddy -- too expensive for what it does. The moving display makes sense though -- like having a magnifying glass on the end of the tweezers. That, combined with a separate display showing the PCB and placement (something like this) and an LED indicator on the feeder showing the next part to place would make it very interesting. Oh, and make it $199.  Seriously though, under $1K.

[Psi]

The ideal price point for this sort of thing is in the $300 range.

Nonsense, the BOM would be around $100, for that you can not even commercially build something with a camera , vacuum and 1 electrical feeder.
IMO if you build this yourself from scratch and pay yourself $10 an hour it would cost you more than $800.- being a conservative estimate.
I am not sure what your wage is but I bet it is more than $10 

Sure you could. The quallity wouldnt be as good. But it would be good enough to be very useful. And it would open up sales to hobbyists and small home labs.

I really think you would make more money selling higher volumes of a cheaper unit.

Just my 2c

[jmelson]

You have very interesting meaning of word "affordable". $2800 is not affordable since many automatic PnP machines are in the same price ballpark.

* asmi went back to manually place parts...

$2800?  YIKES!  I have a Quad QSA-30A that I bought at auction.  Including shipping, repairs and accessories, I've got just a little more than that invested so far, and it is up and running.  It will do over 10K parts/hour, fully automatically, with vision registration of both the board fiducials and part alignment.
I just CAN'T imagine why anyone would pay this much for a machine where they still have to place parts by hand!

Jon

[sokoloff]

To some extent, it's a different tool, even though both place parts.

It's very common for small, desktop or benchtop milling machines to cost more than much more capable used Bridgeport mills.
Bridgeport buyers can't understand why someone would pay so much more for so much less capability.
Desktop mill buyers can't understand why they should buy a much larger machine if all they need is to mill flats or small pockets on small parts.
Both could quite easily be "correct".

[ar__systems]

$2800?  YIKES!  I have a Quad QSA-30A

Please stop bragging about stuff you got for cheap. Not everyone has time to spend on
repairing old equipment. It is a very poor reference.

[Mechatrommer]

agree, just leave it be, its been 3 years, no point necroing this thread, whatever will happen, will happen the force will find its balance... btw i just use my $100 34MP 1080P 60FPS HDMI USB Industrial Electronic Microscope with 28" monitor to do stuffs.. good for like 0.5mm pitch RGT package and down to 0402.... if i can sell my stuffs and collect $2800, i'll find the one that can do automatic, or else build one myself if i got more time...