I am developing a multi-camera "robotics vision system" that has two PCBs (printed circuit boards) that contain components with very fine pitches and features (about 0.020") and hidden contacts (for example, BGA, QFN, iLCC and similar components). To assemble my prototype PCBs, I need to place components within 0.001" to 0.005" or so. This is extremely difficult to achieve manually - almost everyone says this is "impossible" to achieve reliably, and it sure looks that way to me.

Big companies with $50,000 or more to spend purchase "pick-and-place machines" designed for PCB assembly - a special purpose type of CNC machine. No way I can afford that.

So I invented a super-cheap pick and place device that is only designed to place these problematic components (fine-pitch components, especially those with "hidden contacts"). When I say "super-cheap", I mean only about 1% of the cost of the commercial automatic pick-and-place machines I mentioned.

I described the device I invented at this link:

http://www.cnczone.com/forums/showth...450#post805450

At least 50% the parts are standard off-the-shelf parts for building linear small motion systems. All the custom parts are simple and straightforward, with the possible exception of the vacuum pick-up and placement device.

I'm looking for someone with a CNC milling machine to make one or two of these device.

First we should brainstorm the details of the design a bit, in case you have cheaper and better ways to implement parts of the device. I haven't done much mechanical design, machining or prototyping for years, so perhaps you can suggest where to get the best price on the standard off-the-shelf components we need. Then, of course, I need the parts made.

If you are potentially interested, let's discuss the design via skype or YIM (yahoo instant messenger), then you can quote me for the work. I have a frugal budget for this project, so I cannot afford commercial machine shop rates. So don't quote if you're in that catagory. If you have a CNC milling machine in their garage or basement, and you take on odd projects now and then, you are probably a good match. I expect to spend about $200 for parts, and roughly $200 ~ $500 for machining, depending on the final design.

Presumably I will pay the actual costs for all materials, parts and any outside processes that might be necessary (like anodizing), plus some agreed-upon price for your machining work (and any assembly or testing we decide you should do before sending the parts to me, if any).

However, there is one other factor that might make this job more (or less) interesting for some of you. That is, assuming the device works as planned, I am seriously considering selling these devices.

Why? Because currently, there is a gigantic gap in the products that exist in the PCB pick-and-place market. Almost all electronics components entering the market in the past several years are in SMT (surface mount technology) packages. The most important and sophisticated of those components come in packages that have very fine pitches (distance between contacts) and have "hidden contacts" (arrays of contacts underneath the component, so a person cannot see the contacts on the component or the pads on the PCB when they must place the component on the PCB). The most common of these packages are called BGA (ball grid array) and QFN (quad flatpack, no leads), but several other packages with similar problems are also fairly common.

The difficulty of assembling PCBs with these components justifies the "automatic pick-and-place" market for PCB assembly. These machines cost about $40,000 minimum up to millions of dollars. Typically they are four axis machines driven by high-power servo motors with ~0.0005" increment glass linear scales on the 3 linear axes (and rotary optical encoder on the 4th axis) to assure components are placed within about 0.001" of the intended positions. The biggest difference between these machines and a CNC milling machine is the cutting-tool spindle is replaced by a vacuum nozzle and tips that lift the components from feeders and place them on the PCB. The 4th axis is rotation of the component. These machines also have 2 cameras, one looking down at the components to be picked-up, and another looking upward at the contacts on the bottom of the components on their way to being placed on the PCB. In practice, companies also tend to buy 32 to 128 component "feeders", each of which holds a different component to be picked-up and placed on the PCB being processed. Many companies spend several to many tens of thousands of dollars on feeders alone.

The above is what is required to assemble modern PCBs, both prototypes and production. The more expensive machines are typically faster, but otherwise equivalent to the low-end units that start at about $40,000 to $50,000 (with just a few feeders).

Another class of pick-and-place machines for PCBs are available, called "manual pick-and-place machines". These typically cost one to several thousand dollars, and have no motors, no encoders, and no "precision". In effect, they simply help the operator be a little more efficient in picking-up and placing components --- strictly by eye... by human vision. These devices are no more accurate than picking-up and placing components manually with a pair of tweezers. In fact, since the operator can move his eyes much closer to the PCB with tweezers, these machines tend to be less precise than manual assembly. Some people claim they help speed up the process slightly, and possibly delay the onset of repetitive stress injury from placing components. But they do not solve the problem of placing modern fine-pitch and/or hidden contact components.

While my device could in principle place all components, the device doesn't really provide any advantage over manual placement except for components with "fine-pitch" and/or has "hidden contacts". The whole idea of my design is to make it POSSIBLE for individuals and small companies to develop and prototype products with the latest and greatest components. Currently we cannot, for the reasons I have stated. And since most new (and virtually all seriously cool) components come in "fine-pitch" and/or "hidden contact" packages, innovation in electronics by individuals and nano-companies has been massively constrained, and is constantly getting worse.

Thus, the purpose of my device is to precisely place components that cannot be reliably placed with sufficient precision manually.

What makes this a pretty interesting "market" is the difference between the cost of automatic pick-and-place machines for PCBs ($50,000 up) and this device (about $500)... a factor of 100x. With some optimizing, it may even be possible to make versions that sell for less than $500.

The one and only part that is not simple and straightforward is the vacuum pick-up device. The basic elements are pretty trivial: something like a small aquarium pump, some rubber tubing, and a small metal shaft with a tiny hole through it (or large-diameter hollow needle). What may be less straightforward is toggling the vacuum line ON when it is time to pick-up the component, then turning the vacuum line OFF when it is time to place the component. Probably it is also necessary to generate a slight "puff" of positive air pressure at the correct moment to force the component off the tip. This is because these components are extremely light-weight (tiny fractions of a gram), and extremely tiny forces like molecular adhesion and static charge can prevent the component from falling off the tip when the vacuum vanishes. So, the toggling of the vacuum and doing so at the appropriate times, is pretty much everything non-trivial. I have a few ideas about these issues to brainstorm with whoever works with me on this project.

Anyway, the bottom line of the "incentive" is this. If the device is as convenient and reliable as I expect it to be, you could have ongoing work making parts for these units as they get sold. I have no idea how many to expect to sell per year, but anywhere between dozens (1 per week) to thousands (500 per week) are possible. Obviously if the quantity approaches the upper end of the range, we'd need to have parts made elsewhere unless you want to fill up your garage with machines! But let's not worry about that scenario yet, okay?

I'm willing to simply contract for you to do work, and pay for that work, then order additional sets of parts from you if and when I market them (which you can offer to supply or not, your choice). Alternatively, if someone wants to partner with me on this endeavor, I will consider that option too.

If you want to clarify the design and its elements, perhaps that is best done on this thread. If you want to contact me outside this thread to discuss quotes or terms, I believe this forum lets you send private messages to me.