Business models for having a PnP

[freaklabs]

When you get a PnP, people start asking if you do contract assembly. I always say no because I'm already busy with assembling my own boards. As designers, we pretty much want don't want to spend all of our time on assembly which is one of the reasons for having a pick and place. However there are some interesting possibilities with having a PnP. If you already have a storefront, then one possibility which is used pretty well by Sparkfun is that you can provide a service where you sell other people's designs in your store but you also assemble them. Its interesting because it addresses needs on both sides. On one side, sellers usually end up spending a huge amount of time on hand assembly and so you see a lot of people come out with a couple of PCBs, and then drop off the face of the planet because they determine that selling boards is much more work than it looks like. On the other side, people that buy a pick and place machine are looking for different ways to keep it occupied to pay it off. The majority of the components, as LadyAda mentioned, are jellybeans which are very cheap. The labor required to mount them by hand usually far exceeds the cost of the components. Ideally, if you combine the two, the designer of a board can just hand off the PCB to the pick and place owner with the BOM list. The PnP owner can manufacture a couple of the boards and sell them in their shop. The designer would get a royalty (or something like that) and the seller would get a margin that reflects the parts sourcing, assembly, and distribution. Both sides end up getting paid and the PnP is leveraged to carry the burden of assembly (except for the setup and mounting the specialized components).

I suspect that its pretty labor intensive though which is why I won't be doing this anytime soon. It is pretty interesting though and would be an ideal model for micro-manufacturing, or how to efficiently make products targeted at niches.

[freaklabs]

Hmmm...not so sure about that idea after re-reading it. Most likely, if you get a PnP, you'll end up designing more, not less. It's kind of a feedback loop where the extra time saved by the PnP gets spent on new designs which need to be assembled by the PnP. I'm sure there's an equilibrium somewhere, but at least thats what I'm finding with myself.

[ktownsend]

I was thinking about offering some sort of short-run assembly services (to help cover the payments on the PNP), but as a 1 man show I just can see it being feasible time wise. There's already not enough hours in the day to manage my own workload. If the interest is sufficient that you can hire someone to manage the assembly, then it's probably doable and maybe even interesting, but I suspect most of us who can't hire someone would rather be building stuff that we designed ourselves than other people's boards if you have to choose between one or the other with only so many hours in the day. I've decided to buy the machine to save time (and brain cells) for myself, and will have to add that into any prices charged for services or products ... if you need to develop an alternative business model around the machine to pay for it, prospective owners may want to hold off on buying one. That said ... there's definately a need out there for inexpensive short-run assembly if someone wants to do it.

[bootstrap]

Well, the other option is to find 2 to 5 people in the same general geographic area to split the cost of a pick-and-place machine, then share access to it. Or if someone already spent the money, perhaps they'd rent some time on his machine for a few of us not-so-lucky-or-rich. I would definitely be interested in either option. I'm in Nevada at the moment, but anyone within a few hundred miles (NV, UT, AZ, NM, CA), would be fine, since I only need to assemble prototypes once every few months. Anyone game for either idea? I could probably kick in up to $10K towards a shared system.

[charliex]

I've been thinking about how to work a cooperative in that respect. I'd really like a PNP but i just can't justify it at all, other than its awesomely cool. It was bad enough choosing my AccurateCNC. I'm in NV/CA area too. I had been thinking about doing something in one of the hackerspaces.

[mikeselectricstuff]

The other big issue is that for short-run jobs is setup time gets significant, and unless everything is on reels it gets fiddly - even then if you don't have enough feeders you spend a lot of time reloading them for different jobs. Not to mention setting up and cleaning stencils...
Another aspect is that as soon as you start doing work for other people you take on risks of bad boards, wasted parts etc.

[blogger]

It was bad enough choosing my AccurateCNC.

which is the 'bad' part? that it was a tough choice or that its bad?

[charliex]

choices, cost, how much i really needed it. the machine itself is great, no regrets once i got it.

[bootstrap]

I've been thinking about how to work a cooperative in that respect. I'd really like a PNP but i just can't justify it at all, other than its awesomely cool. It was bad enough choosing my AccurateCNC. I'm in NV/CA area too. I had been thinking about doing something in one of the hackerspaces.

What is "one of the hackerspaces"? What is a "hackerspace"?

If you and 1~2+ others others are willing to fund a cooperative venture, I am totally ready - NOW. I suspect we need at least $40K to get a minimal but quality setup (like the LE40V or equivalent).

Here's another idea. I need this equipment to bring a few products to market starting early next year. If my products sell okay, I'd be willing to gradually refund everyone's investment over time as my products generate revenue (and start consuming more machine time to build production PCBs). Over a year or so, everyone would get their money back, but still be able to build prototypes with the equipment.

If I knew 100% that my products would be successful, I'd just buy the thing myself right now - but I'm only 98.6% sure (plus or minus 1.3%). Hence, I'd like to limit my up-front risk somehow.

[charliex]

Hackerspaces vary from a couple of people with a common interest in making stuff, software/hardware and talks about it in a room, to larger scale locations that charge a fee to be a member every month. In equipment they range from nothing but a few chairs to metalworking, laser cutters and welding equipment.

Unfortunately the well established, well tooled ones seem to be mostly in Silicon Valley, there are some interesting startup types in socal i've seen, and not much of anything in Vegas/NV that i've found.

Usually a pretty good list going on here.
http://hackerspaces.org/wiki/List_of_Hacker_Spaces

[bootstrap]

The other big issue is that for short-run jobs is setup time gets significant, and unless everything is on reels it gets fiddly - even then if you don't have enough feeders you spend a lot of time reloading them for different jobs. Not to mention setting up and cleaning stencils...
Another aspect is that as soon as you start doing work for other people you take on risks of bad boards, wasted parts etc.

I'm not sure I understand your points.

To me, the point is this. If we (product-developers, engineers, inventors, hobbyists) want to develop electronics gizmos with modern devices (0201s, 0.50mm pitch QFNs/BGAs, etc), we NEED some way to develop our PCBs --- or we're "out of business". In other words, to have a pick-and-place machine is a "make or break" distinction. Either we CAN develop these kinds of PCBs, or we CANNOT.

Unless I misunderstand the goals of others who need access to a modern pick-and-place machine, setup time is nearly irrelevant. Why? Because without the machine, our projects would take infinite time to complete (as in "never"). To complain "we need a whole day to install parts on the machine" simply means "we waste 1 day". So what! I've already wasted 2 weeks trying to figure out how to assemble my prototype PCBs! And I'd pay $10K to share a machine just to assemble these PCBs... even if I never made any more later on.

The primary goal is to make prototypes (and maybe later, modest-size production runs) --- NOT to start a "PCB assembly business". Even if some of us did decide to offer such a service eventually, the existing PCB outfits do not guarantee "no waste". In fact, the cheapest outfit I found asks their customers to send 20% extra of every component if they can! Talk about waste!

Actually, I'd be more inclined to "offer/rent time on the machine" than "assemble PCBs". Probably in this case they'd watch one of us owners assemble their PCBs, and be paying for our time more than a specific service. If they were a hobbyist, perhaps we would charge nothing, but require they arrive when we have the time and inclination. If they were someone like me, trying to create a new product to start a new business, perhaps we would charge something now and/or something later on (when they went to market).

Anyway, can you explain more clearly what you were thinking? And do you know PCB assembly businesses that guarantee the PCBs they assemble work? Really? I wish!

[mikeselectricstuff]

The point I was making is that having invested in a P&P, going into the subcontract assembly business is unlikely to be a sensible way to help pay for it. Time rental/sharing is probably a more viable approach, but I think possibly still fraught with potential issues - maintenance/consumables costs etc., but as most machines in developer environments are likely to spend a large proportion of their time idle, it may well be viable in the right circumstances

[scsi]

And I'd pay $10K to share a machine just to assemble these PCBs... even if I never made any more later on.

The primary goal is to make prototypes [...]

Have you looked at Madell PnP's at all? I'm using their vision/software on my own machine and I can see how it may very well work for what you need it to do - 0201's and 0.5mm pitch components.

I don't think they can use local fiducials for positioning correction, but there is a feature that allows you to use the down looking camera to correct the placement position based on the contact pad image. In other words, it can look at the PCB and correct any offset before it places the part. This is in addition to the global fiducial correction.

-scsi

[ktownsend]

The other big issue is that for short-run jobs is setup time gets significant, and unless everything is on reels it gets fiddly - even then if you don't have enough feeders you spend a lot of time reloading them for different jobs. Not to mention setting up and cleaning stencils...
Another aspect is that as soon as you start doing work for other people you take on risks of bad boards, wasted parts etc.

After a few weeks of having my own pick and place, that's pretty much my own thoughts reading this thread. It just isn't worth the hassle to me. Because I can't afford 50 reels, I find myself optimising my designs to a certain number of common components that I always leave on the machine. It takes me about 3-4 hours to program a board in and run through one or two test runs to make sure everything is setup properly, ASSUMING that I already have proper, tested part definitions for all the components (which means pouring over every datasheet for evey chip, connector, etc.). It's a lot more work than just pressing the 'assemble' button, not to mention dealing with the solder paste and cleaning stencils, etc. Large ICs are actually the least work ... it's all those discretes that take forever since you need to verify the package dimensions on all of them, etc. Unless someone was using the same parts as me, I just can't see it being worthwhile to offer any sort of assembly service. It's fabulously boring, manually-intensive work and it isn't that much fun to do it for yourself ... I can't imagine anyone happy doing it for someone else for a couple hundred dollars given the amount of time and effort involved.

[bootstrap]

And I'd pay $10K to share a machine just to assemble these PCBs... even if I never made any more later on.

The primary goal is to make prototypes [...]

Have you looked at Madell PnP's at all? I'm using their vision/software on my own machine and I can see how it may very well work for what you need it to do - 0201's and 0.5mm pitch components.

I don't think they can use local fiducials for positioning correction, but there is a feature that allows you to use the down looking camera to correct the placement position based on the contact pad image. In other words, it can look at the PCB and correct any offset before it places the part. This is in addition to the global fiducial correction.

-scsi

I guess the main question is... what kind of built-in precision do they have [over short distances]. The problem is, even if they exactly sync the pick-and-place-head against a fiducial mark or via on the PCB... they might instantly void that precision the moment they move the head, unless they have linear encoders on the X,Y tracks. Anything with gears, belts or other schemes between the pick-and-place head and the encoders (or stepper motors in my nightmares) can go out of sync even before it moves the head 0.001" due to drive-train backlash.

I have looked at Madell pick-and-place units, but I've read a few horror stories on the internet, and I'm too chicken to risk more than a modest expense with them --- and pick-and-place machine costs are light-years beyond the "risk" dividing line for me. Lucky for you, the item you purchased from them works well. I may buy a few of their cheaper items on their "rework" pages.