Feeders, new hardware (split from Head wiping tapes)

Started by Gopher, March 25, 2011, 08:00:39 PM

Previous topic - Next topic

fcb

Quote from: Gopher on March 26, 2011, 02:59:17 PM
<£10k ? that's not even an operator! at this price point it's is surely not something you can sell for profit, it's collaboration for mutual benefit or nothing. I have wondered why it has not yet being attempted either for AOI or p&p, certainly there are closed source lab/homebrew projects (they look rubbish).
It is profitable (else I wouldn't do it) - but there isn't the sort of fat in the project to make it attractive to the big players. I have seen the car-parks of some of these companies.. neither will our sales team drive Porsches (you know who you are..).
We plan on providing training but this will be at our factory, we will also offer to install your machine - although you can do it yourself.

Mike

Quote
<£10k ? that's not even an operator! at this price point it's is surely not something you can sell for profit, it's collaboration for mutual benefit or nothing.

Not sure how true that is. How much does the hardware really cost, if well designed for low-cost manufacture?
Quote

Being open source brings advantages and disadvantages

Pro's

It can keep going if its main backer fails or drops out
Anyone can contribute
Tweaks can be made for one off requirements
Interesting ideas can come in and you have a nice array of beta testers
Buying one would be less fraught with doubt than a new entrant who might go titsup 2 years later
Documentation through wiki, so all those questions get answered only once.

Con's

If its main backer fails or drops out they might well take the expertise needed to keep going with them.
This can happen with a closed-source solution too. Or a small company. But if O/S the info is there and it can be taken up by others. With closed source you're screwed, like with the RV.
Quote
The target market might not have the skills or time to help with development.
Only a proportion of it needs this, and it is likely that even those that don;t contribute to code or design can provide valuable ideas and feedback based on their experience and expertise
Quote
Most or indeed all open source projects have a few core contributors and a whole bunch of leeches.
So what? As long as there are enough useful contributors the others don't matter
Quote
The physical machine can be knocked off cheap by the Chinese who then undercut you by ignoring the original ethics and aims of the project.
This is a common retort to the concept of open source hardware, however this can just as easily happen with closed-source hardware (I just bought a 'Segger' J-link JTAG interface for £30 on Ebay from China), and I believe there have yet to be any significant instances of this actually happenning.
Quote
I would assume such a project would require at least some income form the sales of hardware and support if only to fund development
There are a lot of OSHW companies doing very nicely thank you  a few with >$1m turnover.

With P&P there would be plenty of opportunity to sell support packages to people who
want it - companies like Red Hat have made decent busnisses out of this model

QuoteOpen source?
Probably not. I'm quite sure that opening the code up to the wider world would have certain benefits - but it isn't a model I'm comfortable with. Mainly the fact that there is a risk of damage to warranteed hardware if the user can tinker or load a non-approved application.
That can fairly easily be dealt with - it's only the lowest level motion-control stuff that has risk of breakage, and wrapping that up in a 'do not touch' DLL, or in embedded firmware wouldn't be a big deal, while still offering the ability to tweak things that people want to.
Quote
In fact the idea behind the code is to license it so that you would pay a small amount annually for the latest major revision.  This appears to be a model with a proven track record (I have used Easy-PC for many many years like this, and pay them about £60 per year for the latest version).  The incentive to keep supporting older machines in the market with new features then exists.
Paying for upgrades is one thing, as long as it's not the sort of insidious model where the software stops working after a time period  I would never consider buying anything that worked like that.
However it does have the potential to get messy - some people will not upgrade, meaning there will be a mix of software versions out there, and if you need to release a fix for a serious bug (which users will not be happy paying for),  things can quickly get messy.
Another thing I always fell leaves a bad taste in the mouth is where companies sell hardware that is cripped and you then have to pay extra to use kit you already own (Agilent Scopes come to mind - paying for extra software functions is one thing, but having to pay to enable all the memory is taking the piss.  
As regards development funding, Kickstarter is an interesting concept for open source type projects.


Gopher

Quote from: fcb on March 26, 2011, 03:43:05 PM
I have set a basic specification and spent a lot of time thinking about the potential market.

Sub-contractors/faster machines
I'm not targeting sub-contractors specifically, mainly the 'little-guy' who wants to bring his low volume assembly in-house. The price point of the machine is going to be very important - we have a target of <£8-10K with a 'sensible' number of feeders and feeder mix.  I don't know what the market size is - I'm not sure anyone does really, as it's quite likely this machine will set a new price point/quality.

Ok if I go back to where we were 10 years ago when the RV machine did one or 2 things for us and one other customers products so effectively similar to a "little guy" in house line.  Forget the feeders, the niggles are not rocket science to get round and you could even argue that perhaps batch sizes here warranted most stuff going onto passive strip feeder holders. I am going to shrink the theoretical company because at 14-20 the real one is too big and a lot of it is full turnkey.

£60/year sounds low it may even be low enough to work against you, your target may well be paying more than that for anti-virus updates and I agree with Mike here you could end up with a confusing array of versions. Our accounts package has this model and we did indeed stop paying them because the new features were of no use to us and the support that came with it was woeful.

Now I transplant it to 2011, It's the machine itself that would concern me, I'm small low volume and specialist.

There's me and maybe one or 2 others to design order pack test hand solder etc. So I'm going to use SM wherever I can as I don't have a £30k flo-wave machine? This however ups my lanes count, but its cheaper than another pair of hands.
I assume this means what I'm doing is quite clever or some generic Chinese tat would be being modded putting me out of business. So I'm going to want use all sorts of parts because I have limited choices, quite possibly including fine pitch bga, 0402/0201 in 2mm index tapes etc. This means 60 odd 8mm lane slots and somewhere to put a waffle tray.
(based on the typical stuff I get asked to build these days of up to 30 8mm and a good scattering of 12 16 and 24mm tapes; having said that vivo say they have customers who only need 20 lanes because their designs are digital).
Probably a pipe dream but: As I'm low volume I'd rather not be spending £150 a go on a stencil, I want the machine to do it or at least it could be an option. ( Or indeed £1500 on the pasting system and £800 on the frame) Granted I might consider the £350 prototype stencil system.

Software, I'm small so my staff are few in number but skilled, perhaps not as great a need for muppet proofing as might otherwise be the case, so technical terminology is ok but it needs to be intuitive and a close to load->place as possible so I can do something else while it works.
It needs to take standard pick and place files.
I would be nice if it had a good way of handling complex panels, where one panel has multiple pcb's perhaps in different orientations.
Inventory control would be handy, or at least an end of job report detailing any parts attrition.
It has a camera, I want a software module that does basic (orientation etc) AOI just to make sure. But I can wait, that's cake icing.
It needs a small footprint because space is expensive, possibly I work in my garage or large shed.
It needs its own air supply
Be nice if it was quiet
It needs a support forum (because its 2011) so the community can help itself
A loan system for feeders we only need for one job
A service exchange program so a ready repaired working part can be sent out when something breaks
Offline programming so a new project can be setup and ready to go, or indeed a processing footprint and modus operandi that means the control software isn't on a dedicated machine but instead is on mine.

Am I now your target?

Gopher

Quote from: fcb on March 26, 2011, 04:33:36 PM

It is profitable (else I wouldn't do it) - but there isn't the sort of fat in the project to make it attractive to the big players. I have seen the car-parks of some of these companies.. neither will our sales team drive Porsches (you know who you are..).
Well my weakness here is I know sweet fa about how much mechanical stuff costs to make.
I might have seen that car park, middle men are a significant issue in this market. And I suspect are the reason the low end machines look so poor against the bigger ones certainly you get way more bang per buck with the expensive machines.

Quote

We plan on providing training but this will be at our factory, we will also offer to install your machine - although you can do it yourself.

On site training means the training can be done on an actual product, it also means its cost covers the expenses of one trainer not multiple trainees. Maybe the option of both would be good, I could see a lot of people wanting a quick introduction alongside installation. The right government grant can often mean this bit is paid for by someone else.

Mike

Quote
It needs to take standard pick and place files.
which of course don't exist, so it must use open file formats to allow someone to write a utility to convert the output from any random PCB software to work. You could probably cover a reasonable range with some sort of scripting facility where you could tell it which fields to look at, which way the orientations work etc.
Quote
It would be nice if it had a good way of handling complex panels, where one panel has multiple pcb's perhaps in different orientations.
That's just software  - if the system is open source, maybe a user will write this for you.
Quote
It has a camera, I want a software module that does basic (orientation etc) AOI just to make sure. But I can wait, that's cake icing.
And if that camera can find and orient loose parts, even better. And find the pick points on feeders. Similarly, on an Open Source system , someone else might do this.
Quote
It needs a small footprint because space is expensive, possibly I work in my garage or large shed.
It needs its own air supply
small is good, definitely table-top. Ideally no air requirement, which also helps with 'quiet'.
Quote
It needs a support forum (because its 2011) so the community can help itself
A loan system for feeders we only need for one job
A service exchange program so a ready repaired working part can be sent out when something breaks
As well as parts and info readily available for user repairs. Schematics would be nice
Quote
Offline programming so a new project can be setup and ready to go, or indeed a processing footprint and modus operandi that means the control software isn't on a dedicated machine but instead is on mine.
a no-brainer. Should have some graphic simulation/representation of board and parts so you could have high confidence the programming is right

Mike

QuoteOn site training means the training can be done on an actual product, it also means its cost covers the expenses of one trainer not multiple trainees. Maybe the option of both would be good, I could see a lot of people wanting a quick introduction alongside installation. The right government grant can often mean this bit is paid for by someone else.
Why do you need on-site training? A system that is designed to be easy to set up & use to start with, plus a few good videos and a user forum should be all that's needed.
It's pretty important to be able to sell something niche like this to a worldwide market, an you want to avoid the cost increase of using local dealers, so something that can be shipped in a sensible sized box, installed, serviced and maintained by a typical end user means a lot of support infrastructure just isn't needed.
By all means sell a Skype support option...

fcb

QuotePaying for upgrades is one thing, as long as it's not the sort of insidious model where the software stops working after a time period. I would never consider buying anything that worked like that.
That drives me nuts.

QuoteHowever it does have the potential to get messy - some people will not upgrade, meaning there will be a mix of software versions out there, and if you need to release a fix for a serious bug (which users will not be happy paying for), things can quickly get messy.
One would hope that serious bugs would be spotted earlier in the cycle. Some sort of paid-for-annual-upgrade allows the development costs of later features (such as AOI) to be amortised across a wider group of machines, and avoids having to 'knobble' useful future features for some iterations.

QuoteAnother thing I always fell leaves a bad taste in the mouth is where companies sell hardware that is cripped and you then have to pay extra to use kit you already own (Agilent Scopes come to mind - paying for extra software functions is one thing, but having to pay to enable all the memory is taking the piss.
I heard Intel were doing something like that also with the lower-end processors. I'm not sure it's criminal but it leaves a very bad taste in the mouth.

Mike

Quote from: fcb on March 26, 2011, 05:31:47 PM

However it does have the potential to get messy - some people will not upgrade, meaning there will be a mix of software versions out there, and if you need to release a fix for a serious bug (which users will not be happy paying for), things can quickly get messy.
Quote
One would hope that serious bugs would be spotted earlier in the cycle. Some sort of paid-for-annual-upgrade allows the development costs of later features (such as AOI) to be amortised across a wider group of machines, and avoids having to 'knobble' useful future features for some iterations.

Hope isn't exactly a great strategy... and you can never be guaranteed you won't get tripped up  by things like when Microsoft breaks something in a new version of Windoze or a security update.

fcb

QuoteOk if I go back to where we were 10 years ago when the RV machine did one or 2 things for us and one other customers products so effectively similar to a "little guy" in house line.  Forget the feeders, the niggles are not rocket science to get round and you could even argue that perhaps batch sizes here warranted most stuff going onto passive strip feeder holders. I am going to shrink the theoretical company because at 14-20 the real one is too big and a lot of it is full turnkey.

£60/year sounds low it may even be low enough to work against you, your target may well be paying more than that for anti-virus updates and I agree with Mike here you could end up with a confusing array of versions. Our accounts package has this model and we did indeed stop paying them because the new features were of no use to us and the support that came with it was woeful.

It would be more than £60, that is the figure I pay each summer for the latest version of Easy-PC.

QuoteNow I transplant it to 2011, It's the machine itself that would concern me, I'm small low volume and specialist.

There's me and maybe one or 2 others to design order pack test hand solder etc. So I'm going to use SM wherever I can as I don't have a £30k flo-wave machine? This however ups my lanes count, but its cheaper than another pair of hands.
I assume this means what I'm doing is quite clever or some generic Chinese tat would be being modded putting me out of business. So I'm going to want use all sorts of parts because I have limited choices, quite possibly including fine pitch bga, 0402/0201 in 2mm index tapes etc. This means 60 odd 8mm lane slots and somewhere to put a waffle tray.
(based on the typical stuff I get asked to build these days of up to 30 8mm and a good scattering of 12 16 and 24mm tapes; having said that vivo say they have customers who only need 20 lanes because their designs are digital).
Probably a pipe dream but: As I'm low volume I'd rather not be spending £150 a go on a stencil, I want the machine to do it or at least it could be an option. ( Or indeed £1500 on the pasting system and £800 on the frame) Granted I might consider the £350 prototype stencil system.
Feeder mix, types and qty's is always going to be an issue with smaller machines. Feeder rental/loan could help when really odd mixes are an issue. 0402 is definite, 0201 is possible with the design - but I haven't run 0201 yet.


QuoteSoftware, I'm small so my staff are few in number but skilled, perhaps not as great a need for muppet proofing as might otherwise be the case, so technical terminology is ok but it needs to be intuitive and a close to load->place as possible so I can do something else while it works.

It needs to take standard pick and place files.
I would be nice if it had a good way of handling complex panels, where one panel has multiple pcb's perhaps in different orientations.
Inventory control would be handy, or at least an end of job report detailing any parts attrition.
It has a camera, I want a software module that does basic (orientation etc) AOI just to make sure. But I can wait, that's cake icing.
Software..

QuoteIt needs a small footprint because space is expensive, possibly I work in my garage or large shed.
Yes

QuoteIt needs its own air supply
Be nice if it was quiet
Yes, yes

QuoteIt needs a support forum (because its 2011) so the community can help itself
100% agree.

QuoteA loan system for feeders we only need for one job
Yes. I can think of a bunch of jobs that I've done recently where that would have been VERY useful.

QuoteA service exchange program so a ready repaired working part can be sent out when something breaks
Yes. Spares support with next day delivery - I had envisaged that since day1.

QuoteOffline programming so a new project can be setup and ready to go, or indeed a processing footprint and modus operandi that means the control software isn't on a dedicated machine but instead is on mine.
The machine connects via a single USB port to a PC. The code is not locked to the PC. The plan is that off-line programming is effectively 'free' and you will be able to use it on any machine you like. You will need a USB feeder interface/PSU block (~£250) to do full off-line loading/setup of feeders at your desk, full support for barcode scanner from day1.

QuoteAm I now your target?
Quite possibly - the funny thing is the projects code internal codename is "little-guy".

fcb

Quote from: Gopher on March 26, 2011, 05:14:53 PM
Quote from: fcb on March 26, 2011, 04:33:36 PM

It is profitable (else I wouldn't do it) - but there isn't the sort of fat in the project to make it attractive to the big players. I have seen the car-parks of some of these companies.. neither will our sales team drive Porsches (you know who you are..).
Well my weakness here is I know sweet fa about how much mechanical stuff costs to make.
I might have seen that car park, middle men are a significant issue in this market. And I suspect are the reason the low end machines look so poor against the bigger ones certainly you get way more bang per buck with the expensive machines.

Quote

We plan on providing training but this will be at our factory, we will also offer to install your machine - although you can do it yourself.

On site training means the training can be done on an actual product, it also means its cost covers the expenses of one trainer not multiple trainees. Maybe the option of both would be good, I could see a lot of people wanting a quick introduction alongside installation. The right government grant can often mean this bit is paid for by someone else.

Middle-men. We don't plan on having any middle-men in the UK. Period.

On-site training is more relevant if you train more than one person - good point.
Can be done on customers product - good point.
Can be paid for by grant - again good point (and not one that had crossed my mind, thanks).

Gopher

I'm not sure I NEED onsite training but I suspect someone might say I had to have it likewise the actual product, everyone thinks their baby is a special case. Personally an IRC channel & videos would be fine but everyone is different.

Perhaps I should have said common not standard pick and place files.

+1 to the loose parts finder

Database driven, making interfacing it with other software/things easy, potentially.

Yes even if fully closed source, schematics are important, without them we have to fall back on the support of the manufacturer who if small and hoping not to spend their lives helping us will be busy - selling new ones!

Its probable they would need to stock parts, taking the RV pump example its 4 times the price on short notice as going straight to source and waiting. Or the machine could come with a spares box option.

Would tabletop not create restrictions especially WRT feeders? If it's on is own pedestal then the waste tape can collect inside, no trip hazard. If it could flat pack shipping shouldn't be an issue.

Any commercial machine needs a  guinea pig, someone who can use it and provide feedback without a vested interest or the desire to make excuses for its shortcomings. Perhaps this is what went wrong with the RV shipping with those weird software bugs.

Don't get me wrong I love open source, I run CentOS as our server and Scalix for email, but there you have prime examples of it failing.
Am I paying RedHat? No it's big players who pay their bills which is why RedHat spend a lot buying the toys big boys want.
Am I paying Scalix? (ok that's pretend open source, you can't fork it and end up with.. well anything really)
Where is Centos 6? Can people join in and help it happen? (no it takes to much effort to teach them how)
Where is Scalix 11.5? For that matter where is parent company Xandros now Asus deprecated their distro from the eepc?

With a machine this complex I can see those who have the skillz becoming distracted or busy and moving on, its certainly a trait I see with developers I come across. These people are presumably consultants who will often have more lucrative work. I suppose the software could be split into sections so that other interested projects find it easier to cross develop, say vision, motor control, ui, etc some of these could be used for other open machine projects, I did see a totally dead looking AOI using canon camera project on sourceforge....
Windows or Linux?
How many people here or that you know of could help? I saw a similar project mooted elsewhere and all they got was don't be silly it can't be done, go away. In that instance it was a commercial project with very similar ambitions and target pricing to fcb.

Mike

Quote from: Gopher on March 26, 2011, 06:22:39 PM
Yes even if fully closed source, schematics are important, without them we have to fall back on the support of the manufacturer who if small and hoping not to spend their lives helping us will be busy - selling new ones!

Its probable they would need to stock parts, taking the RV pump example its 4 times the price on short notice as going straight to source and waiting. Or the machine could come with a spares box option.
..of course everything should be as off-the-shelf as possible to minimise custom parts  e.g. USB camera(s) - in the above case I think putting the RV pump in the arm didn't really make sense- it constrains size, increases weight and it doesn't really need to be in there.   
Quote
Would tabletop not create restrictions especially WRT feeders? If it's on is own pedestal then the waste tape can collect inside, no trip hazard. If it could flat pack shipping shouldn't be an issue.
maybe - I was just thinking for both ship size and flexibility of installation - obviously depends on feeder design how practical this is. Maybe base it on an Ikea table that can be sourced locally ;D
Quote
Any commercial machine needs a  guinea pig, someone who can use it and provide feedback without a vested interest or the desire to make excuses for its shortcomings. Perhaps this is what went wrong with the RV shipping with those weird software bugs.
I doubt you'd have problems finding a few beta users
Quote

With a machine this complex I can see those who have the skillz becoming distracted or busy and moving on, its certainly a trait I see with developers I come across.
I'd dispute that it actually is all that complex. There are ready-made libraries for the trickiest part, the vision  everything else is a bunch of individually simple tasks strung together. 
The crucial key is getting the software architecture right to begin with, and making it nicely modular so different parts can be done independently.
Quote
Windows or Linux?
Probably Windows, because it is more standardised and more people know it. Anyone who is religiously anti-Windows can always buy a dedicated Netbook for 200 quid. Fortunately I doubt many people would want to run a P&P on a Mac, but I bet you'd get some requests...
How many people here or that you know of could help? I saw a similar project mooted elsewhere and all they got was don't be silly it can't be done, go away.
Never listen to anyone who say things can't be done unless they can give some really good arguments. It usually means they lack imagination.

fcb

QuoteI'm not sure I NEED onsite training but I suspect someone might say I had to have it likewise the actual product, everyone thinks their baby is a special case. Personally an IRC channel & videos would be fine but everyone is different.
When I bought a new machining centre about 18months ago it came with 2 days training at a training centre - by the time I got round to going I needn't have bothered.

QuotePerhaps I should have said common not standard pick and place files.
What are the file formats people prefer.

Quote+1 to the loose parts finder
An advanced feature that will have to come in later code versions.

QuoteDatabase driven, making interfacing it with other software/things easy, potentially.
My prototype uses a simple CSV file for the placement locations. Component images are stored for debugging (like the RV).

QuoteYes even if fully closed source, schematics are important, without them we have to fall back on the support of the manufacturer who if small and hoping not to spend their lives helping us will be busy - selling new ones!
Mechanical exploded drawings are a must. Circuit schematics/harness drawings too.

QuoteIts probable they would need to stock parts, taking the RV pump example its 4 times the price on short notice as going straight to source and waiting. Or the machine could come with a spares box option.
Agree - main spares are really only seals, belts and perhaps pumps. And every part will have ex-stock spares.

QuoteWould tabletop not create restrictions especially WRT feeders? If it's on is own pedestal then the waste tape can collect inside, no trip hazard. If it could flat pack shipping shouldn't be an issue.
The only reason not to go 'table top' is the feeder exits. I wanted to avoid the expensive of a pedestal. This will have to be revisted later I suspect.

QuoteAny commercial machine needs a  guinea pig, someone who can use it and provide feedback without a vested interest or the desire to make excuses for its shortcomings. Perhaps this is what went wrong with the RV shipping with those weird software bugs.
Beta-testing, yes valid point. Nearer the time I guess we will have several machines available to do beta-testing.

QuoteWith a machine this complex I can see those who have the skillz becoming distracted or busy and moving on, its certainly a trait I see with developers I come across. These people are presumably consultants who will often have more lucrative work. I suppose the software could be split into sections so that other interested projects find it easier to cross develop, say vision, motor control, ui, etc some of these could be used for other open machine projects, I did see a totally dead looking AOI using canon camera project on sourceforge....
Perhaps another reason not to go OS.

QuoteWindows or Linux?
Windows

QuoteHow many people here or that you know of could help? I saw a similar project mooted elsewhere and all they got was don't be silly it can't be done, go away. In that instance it was a commercial project with very similar ambitions and target pricing to fcb.
By it's nature, building P&P appeals to the kinds of people that need P&P. I feel that the difference is that I could have afforded something like the TWS Quadra or similar, but never felt that it was value-for-money.

Gopher

so openCV for vision
I suppose you can borrow heavily from open source robots and robot wars for the gantry control?
I don't see an open source gantry/cartesian robot, so that you have to make, as with the feeders?
The frontend, not so much complex as requiring the most thought, it needs consistency and the bare minimum of dialogue boxes and weird data entry windows.
project management.
Camera's: cheap usb webcams probably won't cut the mustard (unless there are known good ones) their focal lengths are not exactly rock solid and the models go out of fashion quicker than H&M. You need something you can rely on to stay the same, unless the software can calibrate itself to whatever is fitted. I think you can still buy RV cameras even now, tho' you likely wouldn't want to.

Looks like you are itching to get started!

Gopher

We went and looked at the Quadra and it was a "oh, is that it?" kind of feeling especially with its price tag, at the time Blundell's other option was an M4, which has the opposite response, "what the hell would I want that for?"

If it can be delivered at your target price I could see sub-con people buying it to stick in the corner for things that do not warrant loading the 12k cph machine.

Good luck, when do we get to see it?