|
Ultra low cost pick & place?Views: 5241 |
|
Email This Page to a Friend
Date:
May 12, 2008 06:45 PM
Hi All, We are embarking on the design of an ultra low-cost pick and place machine for use by small manufacturers & would appreciate some input. Here are some points: Low cost - in the region of $6-11k (higher cost versions faster) Clearly any machine is built down to a price is not going to be massively rigid, we think we have got an 'on-the-fly' calibration system working that will allow the machine to calibrate out most errors once fixed in place (i.e. if you move the machine you will need to tell it todo a full calibration), and track/detect&correct errors as it goes along (i.e. temperature). Speed, we are targetting 5Kcph on the faster versions (lower friction bearings etc..) and 2-2.5Kcph on the standard machine. We are also planning on NOT selling via distribution, but 'factory direct', and support will be via a dedicated user forum rather than tech's (for cost reasons). Next day spare parts in the US/CA and EU. Software will be free for life self-install on what you PC you have lying around (XP at present) - if we achieive a certain critical mass then the software is likely to get very sophisticated and capable over time. We are interested in all comments (even 'don't do it') but especially interested in comments relating to good experiences with other p&p equipment in the fields of: nozzle design, mechnical vs powered vs 'intelligent' feeder issues and component alignment systems (e.g. cameras vs. laser). We aren't asking you guys to design it for us, and we don't expect this machine to replace the staples of the industry - our intention is to allow the little guy to make his small batches quickly and cost-effectively. Thanks all.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54718)
Date:
May 13, 2008 08:02 AM
Interesting thought. Among the many thoughts that came to mind while reading your post, one stands out for me. While making a cost effective machine is a great idea, it seems obvious that there will be some trade offs on quality. The cheaper the machine, the more likely it is that repair/replacement of failed sub assemblies will be. It may be cost effective up front, but in the long view, it may be more expensive.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54720)
Date:
May 13, 2008 08:24 AM
Anything that can handle the standard SMT packages of today has to be precision. The machines have to be built rigid, and cannot be made of sub-standard components. I have worked with placement machines since the late 80's. To simplify things...the machine needs to put the parts where you want them to go without having to screw around! If you build something that does not perform "as advertised" word will spread fast and your machine will get black balled. I cannot possibly image how you could effectively engineer, build, and market a machine in the price range you mentioned. Just my opinion....
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54721)
Date:
May 13, 2008 08:41 AM
Loco, My thoughts are that your market may be fairly limited. There are people and companies out there that can use low cost equipment. However, with your limitations of support you are basically asking an owner of a small company to output 11-15k on a machine that may or may not have tech support when needed. Major concern for most small owners. My suggestion would be to find 10-15 of these small sized companies and see if they will agree to be interviewed by you for market research. Also many of these customers buy lots of equipment and supplies from reps. They trust them to help if there are problems with suppliers. 1 less thing for the owner to worry about. I think it would be a major mistake to not sell through them. My suggestions would be following. 1. Get working system.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54723)
Date:
May 13, 2008 08:47 AM
Yes, this is a good point and really only time will tell just how reliable such a machine becomes. What I'm aiming for is to simplify and reduce the number of sub-assemblies. For example we are building the camera/vision processor and lighting into one small block on two PCB's, the top and bottom cameras are currently identical and only have a power (24v) and data (currently RS485). As this is a critical part of the product the PC has very little to do with the vision systems and the vision processors can be re-flashed in-situ (no eproms) as improvements are made. The CMOS vision IC is something like $13 and the DSP's doing the vision processing about $5 each, so as a complete sub-assembly the cost of replacing a camera will be $200-300 max. Same for the stepper motor drivers, these are located next to the motors and have LED's on them for diagnosis (power, data in/out) and take data&power only (4 wires). The concept being a blown driver board can be replaced in a couple of minutes. The whole principle is to reduce or eliminate wiring and avoid seperate fancooled boxes stuffed with boards connected by huge wiring looms. 4 common wires for everything (24v and DATA). This is based partly on the depressing feeling you get when opening up an older machine tool and finding huge amounts of wiring and really silly numbers of boards - great to keep service engineers busy, but really stomach churning if it's your $.... About 8 yrs ago I got a panic call from a collegue with a well-respected & fairly expensive CNC lathe, it had thrown up a PSU fault. He was quoted �1300 exchange and 10 days to get the exchange unit to him. I fixed a blown 1N4001 and BC182 on the unit in under an hour. So I guess I'm trying to say that the p&p machine is being designed to be fixed quickly by a descent engineer and a support forum should help find those problems we can't yet imagine.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54724)
Date:
May 13, 2008 08:57 AM
It will be rigid, but not as rigid as something weighing 600lb's. We can't afford to cast/machine parts for a low cost machine. But we think have found techniques that can calibrate out any small distortions in the unit. Your comment about a decent software interface is valid - we certainly aren't planning a complex interface to begin with, our aim is to get a bunch of demo units out there and see what the customer comments are and expand on this. We aren't planning on marketing the units in the conventional way - anyone with $100k to spend on a new machine is not likely to looking at this.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54725)
Date:
May 13, 2008 09:30 AM
Thanks for the comments. Tech-support is one the big savings we think we can make and the customer will have to install the machine themselves - we will make a comprehensive series of videos and tutorials. Calibration will be semi-automatic and the unit will be supplied with all the necessary accesories to do this. In the end we hope that the installation and training won't take more than two days from signing for the crate from UPS... I'm kind of set on selling direct, it really is a great way to keep down costs - dealers/reps do add value, but can add alot of cost. We are engineering the product to be quickly repaired in the field by the end user - certainly any part should be able to be replaced in less than 30 mins. We will make money on spare parts for sure, but we are not designing in any obsolesence into the machine; our bearings shouldn't wear any quicker than anyone elses. I read somewhere that this forum is 70% CM's and perhaps this isn't the #1 market for our machine, but if we can build something that produces CM acceptable results we will be more than satisfied.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54726)
Date:
May 13, 2008 11:40 AM
Helo
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54729)
Date:
May 13, 2008 12:15 PM
Harry, thanks for your comments. Certainly circuit diagrams where appropriate and mechanical schematics of complex parts like the head would be useful to the end user. We are working hard to eliminate relays and switches from the unit, and things like safety interlocks will chosen to be 'off the shelf' and from RS/Digikey/Mouser etc.. It's possible that bearings specs. and part numbers will be included in published material to enable the user source parts more economically. I should point out that I'm designing this machine after many years experience in electronic/electromechanical design and manufacture - whilst I have no experience as a CM (which would be useful), I am the end customer of this machine. We will look for perhaps 6 test sites where people will be happy to run the machine in a CM enviroment. We looked at trying to elimate air from the design, but there doesn't seem to be a way round it at this price.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54730)
Date:
May 15, 2008 04:15 AM
Take a look at the Essemtec machine it basicallly does what you wish to acheive but at a much higher price. It uses a standard camera, laser alignment of components which would probably be essential if you wish to achieve your stated speed. Could sit on a bench but comes with a stand. Uses a PC has a fairly good peice of software, but could be better. reagrds sarason
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54752)
Date:
May 15, 2008 06:56 AM
Thanks Sarason, Laser alignment of components was actually our first attempt - it certainly offers speed and accuracy, but purchasing cyberoptics units would blow the budget & getting round their patent not really an option. I've seen the Essemtec machines and they are well made & expensive. Our aim is to build a whole new (lower) market tier. We have chosen bench mounting for cost and transport issues - the client will have to do a little bit of thinking & work for the large savings. I would really interested in the laser alignment vs. camera alignment discussion.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54755)
Date:
May 15, 2008 11:13 PM
The Essemtec has a very cheap to manufacture feeder which you could adapt, or try one of the feeders that remove the tape by "ploughing" it off. Can't recall the brand right now. They are well made, but very cheap to make. The laser gets you speed. The other way to approach that issue, which I don't believe anyone has tried yet would be to have a similar setup to the laser and just take pictures with a camera and do some vision processing to arrive at the same sort of result. This would be more engineering expense but should still be do-able. Another thing that may help with your laser approach would be to go and visit an engineer/company that builds supermarket scanners. They would be the experts in this field. regards Sarason
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54770)
Date:
May 16, 2008 04:09 PM
Hmm, interesting. My Philips CSM84 weighs 1600 Lbs or so, and it bounces around quite a bit when moving. It doesn't really move very fast, maybe 2 M/sec tops, and 3600 CPH is about all you really get unless the feeder is right across from the component location. One other thing is the number of feeders possible. I originally thought the CSM84 was total overkill at a theoretical maximum of 84 feeder locations, but things are creeping up, and I am using more locations all the time. Will this machine have auto-change nozzles? My CSM doesn't have that option, but has 3 heads. I can do just about anything with the right selection of nozzles on the 3 heads. What motion drive will you use? For low cost, probably toothed belt linear drive? How do you deal with belt stretch? Anyway, I already made my choice to buy a well-used machine, but I would have been quite interested in your model at $6K, if it could do the kind of things I'm doing now. Jon
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54775)
Date:
May 27, 2008 08:03 PM
Insane, inane and a waste of everyone's time. Buy a soldering iron at Target and have it. A cheap machine cannot do what you plan. If it could, China would already be selling them.'
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54882)
Date:
May 29, 2008 01:01 PM
That's what they said about Christopher Columbus. Don't you know that the world is flat - you're going to fall off the edge. If Chistopher Columbus had listened to that advise then Europe would be a lot more crowded today. Bert
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54902)
Date:
May 30, 2008 12:41 PM
I say that there is a market... The question is this: Yes, I concur that with the advent of better and more readily available software and sensors you can probabaly make a low cost 3-4 axis gantry that senses and makes up for poorer tolerances with better software. A good example of this is the Zenbot router being sold on Ebay which actually uses an injection molded UHMW frame in place of the usual alloys and castings. If you do decide to pursue this and even eventually produce such a product without going bankrupt, your real challenge will be sales and marketing. If your like most engineers you've probabaly already grossly underestimated those costs as evidenced by your inferred statement that giving a sales rep a 10% commision or finders fee is somehow expensive when in fact when you don't have sales it's quite actually a bargain and I'd predict you'd pay it willingly. Manufacturers today are relying on old business models and there is always room for risk takers and innovators. All of that being said if your already cash strapped and risk adverse at this stage of the game your not liable to withstand the real physical and emotional costs of building a business... especially against the kinds of odds you start at while selling a disruptive technology into existing markets.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54927)
Date:
June 01, 2008 06:31 PM
I like the possibilities. Couple thoughts; all good equipment is solid, don't try and design a machine around modular aluminum extrusions, they're junk. At the very least use a heavy steel table with a 1/2" base and 4" square legs, they can be bought for $300-$500.
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54937)
Date:
June 04, 2008 07:15 AM
Hi LoCostPP, Sounds to me you are moving towards an Open Source development model. What you might want to do is come up with a robust framework (hardware & software) to which anyone who wishes to take part can develop (and market) their own plugins and modules. Contibuters can be commercial companies, but also customers that have the engineering capabilities. I think this way you are most likeley to minimize development costs while maximizing your market-potential. Base
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54973)
Date:
June 04, 2008 03:06 PM
Hi Base, I guess your right - due to the complexity of the software we plan on keeping the source code under our control BUT we will not be dongling or protecting it in any other way and we will be publishing the program on our website. We will publish the file structures/key so that anyone can write there own front end software - and we are looking at some quite flexible I/O functions for unplanned future options. We won't be providing a PC with the unit - at present a single USB cable between a PC and the machine is all thats needed; the vision system cameras are proprietary and the bulk of the number crunching goes on in the camera - so no need for anything fast PC wise; and as the software is 'free' the customer can download it onto anything running 98-XP (hopefully vista when launched) - i just plug in an old laptop. At present we have a small box with the PSU in it - this drives the p&p, the feeders and supplies vacuum. The whole unit can lumped around by a single person and put in the boot of a compact....
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54991)
Date:
June 04, 2008 04:04 PM
How about involving (technical) universities in the process? They might be willing to supply some manpower and more importantly knowledge for low to no cost, (but you probably already thought of that).
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=54992)
Date:
March 13, 2009 12:42 PM
By way of an update. We have the locost machine running. Plenty of stuff left to do BUT we should have it available for Q3/2009. The basic spec is much the same, we have spent *ALOT* of time developing the vision system and calibration mechanism. Feeders have also been interesting. We still intend on being 100% factory direct, forum & email tech support, and at present around $5 to $8k (including feeders).
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=58345)
Date:
November 04, 2009 05:47 PM
Q3/2009 been and gone. We are moving forward with low cost p&p, quite a few things have changed. We are applying for a number of patents for the feeders, placement head design, servo design, calibration system , etc... We now expect the price of the machine to be significantly lower than $8k minimum target (with basic feeders). Feeder design has changed and we are doing longterm reliability testing on them NOW. Each feeder is now an intelligent single lane unit (alot of intelligence), can index 2/4/8mm. Offline setup using $250 USB station using the free software (barcode compatible). Price per feeder is still on target for $70-100 each. Placement rate is around 3k/hr - although we have run the unit at 5k/hr, for about another $2k we expect to get closer to 10k/hr. Posistion accuracy is better than +/- 15um at present (we cannot measure better than this here, and are looking for a 2nd user laser inferometer at present). Direct sales is still the model, and due to the design, the end user can service the unit 100% themselves with 24hr parts, forum support, videos, guides etc. Short of throwing the unit out of a truck it is intrinsically calibrated or self-calibrates. The user can download (for free) the latest software (and easily rollback to a previous version) to drive the machine - a single USB cable connects what-have-you to the machine. Unit is movable by 1-2 people (depending on your back...) and fits in the boot of a car. Air is self contained, so you just need a strong and level bench, mains socket and a windows PC (development units running XP, release will be Xp/Vista/7 compatible). Should run on a netbook also. It's also VERY quiet. You can work underdisturbed next to it...
(http://www.smtnet.com/distributed_forums/Index.cfm?CFApp=8&Message_ID=60317)
Author:
locostpp
Subject: Ultra low cost pick & place?
Two cameras (upwards fixed, downwards flying)
Vision & lighting system (in-house for cost reasons!)
Zero installation (sits on sturdy bench or desk, self calibrating)
Manual or automatic feeders (with vision check before pick) - very low cost.
Free for life software/upgrades/enhancements.
Single USB cable from P&P to any PC, shop air and AC.
Finepitch as standard, smaller than 0603 as goal and enhancement.
Author:
Avalancher
Subject: Ultra low cost pick & place?
Author:
Sr.Tech
Subject: Ultra low cost pick & place?
The machines that work the best are built on rigid platforms and have "decent" software interfaces.
Author:
fishingfool
Subject: Ultra low cost pick & place?
2. Find COGS ( base your system price on what you spend to build the system.)
3. Work with reps to find a market.
4. Have a small support team and if sales go off you can then contract with 3rd party service providers to do support.
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
Value Machine Sales
Subject: Ultra low cost pick & place?
Since we had designed a screen printer, we have a good idea of the process. There is one thing that we did that you may want to consider. We were going to sell the printers with a complete list of the components and where the customer could buy them at. I felt that that would make it easier to get the machine on the market and the customers would be encourged to buy if they knew that they would be able to buy the replacement parts with out high mark up. Just thought this idea may be of some help. Would be interested in knowing how this developes. And by the way our printer used no input air and was very quick. We felt that a ex field engineer could design a better machine than a design engineer that probly had never been in a production site.
Harry
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
sarason
Subject: Ultra low cost pick & place?
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
sarason
Subject: Ultra low cost pick & place?
Author:
jmelson
Subject: Ultra low cost pick & place?
Author:
canuck
Subject: Ultra low cost pick & place?
Get a clue.
Author:
Bert
Subject: Ultra low cost pick & place?
Author:
severs
Subject: Ultra low cost pick & place?
-How big is the market?
-Is it worth going after?
-Are you the person who can do it?
Cost is everything to some people. There will always be a certain segment of buyers who will buy a product based on lowest cost. Fortunately many of these types already know what to expect at a certain price point and will actually take ownership and responsibility. It shouldn't surprise you that most won't. So the real question becomes just how many of the first types are there who are ready to buy and how can you market to them and close the sale?
Author:
noleadsplease
Subject: Ultra low cost pick & place?
The servos needs to be closed-loop, you can save money on the torque specs and give up speed.
Good luck!
Author:
Base
Subject: Ultra low cost pick & place?
(or am I being stoopid now...)
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
Base
Subject: Ultra low cost pick & place?
You may not want to rely on them for the initial development (too slow), but improving an already working prototype or machine is a great challenge for (software) engineering students as a graduation piece.
Author:
locostpp
Subject: Ultra low cost pick & place?
Author:
locostpp
Subject: Ultra low cost pick & place?
We will publish the file specifications to encourage anyone to write add-on file translation code. We expect to release 1-2 versions per month (depending on bug-fixes, improvements, etc...) We anticipate evolution rather revolution.
