I appreciate you input, and yes it is ambitious. Here are some of my personal motivations behind it. I usually travel a lot. I also have a handful of hobby boards that I produce. The problem is that when I am home I do not want to spend the time building a bunch of boards by hand when I could be with my family.
So, I looked into what it would cost to have these boards assembled for say 100 or 1000, and the cost was on the order of $10-20 per board. The boards only have 10 components on them, and they can sell for about $15, and have $5 in BOM costs. This makes it financially not viable.
But if I have a machine that costs me $500, and in my first round of 100 boards, this means that the cost to me was only $5. This is an acceptable cost, and gets paid for in the first round of production. At 20 boards, it is at least break even with having it outsourced. So from a cost perspective, it makes sense.
Now that the cost perspective makes sense, the time also falls in line.
For a person that only does 1 small board per year with under 20 components, then this may not be for them, but if they do a single board with 200 components, then there is value in having a machine that is more likely to place the correct components more accurately.
I've read through all of the various posts and comments so far, and learned quite a lot! But I gotta say: a fully automatic desktop pick and place with the kinds of parameters you're talking about is a pretty ambitious way to go about addressing the need to populate a few prototypes once in awhile. While there would certainly be a "cool!" factor having such a machine, perhaps a more practical approach would be to simply improve the method of manually populating boards so that it is less tedious and more efficient. I researched this challenge a couple of years ago and came across a great product from Abacom called EZpick, which we now use. In fact we use it for low-volume manufacturing as well, in combination with low-cost laser-cut mylar stencils from Pololu, and a retrofitted toaster oven for reflow.
I have come to peace with both systems of measurement. Each has their advantages. As you mentioned, you can get a whole lot of common hardware in Inch. Many scientific calculations end up being easier to do in metric except when it comes to derived units, then it can be just as hard because you never remember what the derived units actually are in their proper combination of units.
McMaster is one of the greatest stores ever. Here in the states, they have great service and a lot of things at a reasonable price. There are a handful of other sources I use, but they are a big one.
Some may say I'm a big fan of metric, but you can't beat UNF & UNC for quality hardware. I've just spotted on the solar blog http://csirosolarblog.com/page/6/ , about 3/4 way down
26.Each heliostat and its components are held together with 55 bolts – for a total 24,805 bolts in Solar Field 2 alone.
Most of these we airfreighted in from McMaster-Carr, including fasteners you just can't buy over here: 10-32UNC Stainless Hitensile thread forming hex head screws to precisely align the "origami" frames. 3/8UNCx1/2 flangehead ZP bolts with matching serrated nuts. 3/4"UNF x2" ZP for the bearings. And AN526C1032R8 for attaching actuator brackets.
It sounds like you too have had a lot of fun in your day. With your frequet references to astronomy, I am guessing that you work in a rather specific field. Though it seems too that you may have some not just ground based hardware, but space based hardware to your name as well.
We are going to do a poll about the build area size. I was going to pretty much use a cut down list from Seeed Studio as it seems to offer a rather extensive set of size options.
The AS5030 encoders were used to retrofit cheap satellite actuators for the heliostat field at CSIRO , each mirror has two actuators, as bought they have a 4pole-pair ring magnet and a reed switch, we take these out, glue in a small magnet, and mount an adapter PCB using the reed switch screwholes:
With an actuator BoM cost of $25ea, we could point the 200lb+ mirror assemblies to a few percent of the suns diameter even with strong winds.
I also worked up a design for angle sensing through a fibreglass bulkhead in a marine situation with the same family. There is some really neat stuff in this encoder family like UVW to simulate hall switches, as well as the quad outputs you mentioned. (And the zebra stripe magnets for linear position sensors).
I had actually known about the AMS site for some time. They have a great app note about using RFID tags under water in salt water. It was pretty impressive. They also have an interesting line of noise canceling chips.
The thing that impressed me about this chip is that it seems to be a full quadrature encoder replacement that is programmable. This is something that is pretty ideal. The cost was not too awful, and the other thing about it was that it essentially applies a predictive value to compensate for lag while maintaining pretty high angularity accuracy and precission. It seems like a chip that I need to look more into. I like that it is a drop in replacement for an encoder and that it can be programmed with different scaling factors. It can also output the valueas a digital word value, or as just a quaderature style output.
On belts, I am specifically referring to the toothed timing belt style. These are not bad, and are used in a lot of systems. It is not so much the belt that I am opposed to, just that they are sometimes notas clean to integrate. They are effective solutions.
Yep , You have the gist of the motor comparison. Each sort of motor has its weaknesses and strengths.
I see you have found the Austria Micro Systems = AMS website ! , these guys were relatively unknown about 6years ago, we had to go through all sorts of weird purchasing arrangemengts to get the stuff from Austria, now you can buy from your favorite distributer.
Selsyn: just a kind of Synchro , check them on wiki , too hard to explain without pictures. They were antecedents of the resolver. Quite common on 1940's ships and aircraft for relaying the position of say the rudder to the bridge (some still in use on DC10 era aircraft for fuel gauges). You just wire two together, put AC on each end , when you turn one shaft , the other turns, it's bidirectional, great fun for kids of all ages. The other neat thing from back then was the amplidyne: before the days of high power servo amplifiers, you could, for example, attach a synchro to the hour hand of a wind up clock, and a 1000ton radio telescope would follow the sun across the sky, just like leading a bull with a ring through the nose.
Belts: Do you mean toothed belts like a timing belt? or multigroove like an automotive fan belt. Belts have the desirable property of adding damping, most stepper motors without micro-step drives are unstable if a high inertia load is directly coupled to the shaft (That's why you see so many timing belts on stepper driven items). Timing belts are more accurate than you might think (if they are tight enough) and used on most PnP machines of 10yrs vintage.
"Closed loop control dramatically improves holding accuracy and stiffness. Closed loop control can double available accelerations, but no effect on limiting speed."
Though the speed will be present by the very nature that you are using a bldc motor over a stepper. Though the fact of adding closed loop to the bldc motor will not change its nature in speed, but only increase its accuracy and low speed qualities.
On the though of going closed loop and needing an encoder. I just saw this product announcement on a competing website.