Page 1 of 1

Posing a PCB Panelization Question for the Community

Posted: Tue Feb 15, 2005 10:35 am
by sparky
Hello all,

The PCB pooling program is taking off. People are really excited and we are very happy to provide great PCBs for a good price.

Unfortunately we have been on the fence with this program since the inception about a month ago.

Here are our problems - I'd love to get your guys' feedback on ideas on how to make things better.

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Basically, even at $2.50 an inch, we are loosing money because of the time it takes. Customers are sending zip files that are failing DRC, formats are wrong, layers are mirrored.

We have to import them, find out what's wrong, email the customer. The order has to be marked as hold. The customer has to try again. Sometimes the files are fixed the second time around, sometime we have to do everything over again with no success. This is very time consuming.

Once we have valid files, we still have to spend a considerable amount of time (1-2 hours) panelizing the designs. I'm not going for anything crazy efficient, I just get them onto the panel. But combining all the layers, exporting the files take time. Making sure I didn't combine a top layer of one to the bottom copper of another - there is plenty of room for error!

My question : Who knows of some good software to handle CAM file or gerber file manipulation? We are using Protel CAMtastic which is ok, but not great.

Does anyone think it is feasible to setup a PHP script? Java Script? for automatic DRC, panelization, etc? This would make a superb computer science senior project and we'd be willing to pay for a fitting solution. www.freedfm.com is awesome, but we are not Advanced Circuits by any means (>$13 million annually). I am a hardware guy - I don't have the slightest idea about server side applets or coding on a large scale.

I imagine there is all sorts of programs out there but I'm just not in the loop or know what to look for... My only solution at the moment is to throw cheaper labor at the problem but this is dangerous as the panels will be more prone to errors.

Thoughts?

-Nathan

Posted: Thu Feb 17, 2005 12:52 pm
by Moore Logic
Sparky,

offer a two tier price schedule:

1. A "high touch" service for first time customers that includes a file review, DRC, email interactions to resolve minor issues etc @ $5/sq-in. This would be a preferred/required approach for first time users of the service.

2. A "no touch" service for experienced customers that builds the boards exactly as they come in, all fubar or not @ $2.50/sq-in. This is the preferred option for experienced customers that want rock bottom pricing and are willing to take full responsibility for their files.

Jamie

Posted: Thu Feb 17, 2005 2:27 pm
by Caffeine
Moore Logic wrote:Sparky,

offer a two tier price schedule:

1. A "high touch" service for first time customers that includes a file review, DRC, email interactions to resolve minor issues etc @ $5/sq-in. This would be a preferred/required approach for first time users of the service.

2. A "no touch" service for experienced customers that builds the boards exactly as they come in, all fubar or not @ $2.50/sq-in. This is the preferred option for experienced customers that want rock bottom pricing and are willing to take full responsibility for their files.

Jamie
Sounds good to me!

$5 a square inchis still outstanding, and once people get the hang of it and get to know how to put a board together properly, then they can submit to the 'no questions asked' $2.50 price point.

Posted: Fri Feb 18, 2005 9:24 am
by sparky
What a fantastic idea! Thanks guys! You're right, once I get customers past the expensive hurdle, things run nice and smooth on the following orders.

I will think some more on how to get it in-line with the website.

-Nathan

Posted: Sun Feb 20, 2005 3:02 pm
by Vraz
One thing to keep in mind is that the cost of submission mistakes (your time to deal with customer data which is incorrect) is not really tied to the size of the board. If you went with the $5/$2.50 model, then the most cost effective approach for the customer is to layout a 1x1 board for $5 to "learn" how to do it. That continues to expose you to issues and I doubt an extra $2.50 total is going to offset your time.

I think the first-time "high touch" service idea is good, but would simply make it a flat-fee per-design extra on top of the $2.50 sq/in. That protects you and continues to make the service affordable.

On a related note (avoiding bad submissions), it would be great to have the DRC rule set for Eagle in electronic form. That way I could run it in advance and be certain my design passes. As it was, I just entered in the values by hand which increases the possibility of error.

Posted: Mon Feb 21, 2005 10:12 pm
by Moore Logic
Yup, this thought crossed my mind that it is not based on sq-in, but I was then thinking that on average it would work out. Also, a large board has more potential for problems than a tiny one in terms to profile, hole sizes, DRC failures etc...

Jamie

Posted: Thu Mar 03, 2005 9:11 am
by cbatters
How about $2.50 per support email required to get a job right. If it is a quick problem, your costs are low and the penalty is low. If it takes 6 email responses you have earned every cent of the extra $15.00 fee.

Perhaps you can keep track of the time required on a "problem submission" and see if this covers the cost.


Clint

PHP / Java Script

Posted: Thu Mar 03, 2005 11:41 am
by lucky
Hi Nathan,

I know nothing about CAM files etc but why not attach to a post some files:

- A set of files that you would expect to receive from the customer in a Zip file.
- A complete full Panel CAM file that is made up of the single PCB from the Zip file repeated over and over until the Panel is full.

Then us softys can look at the files and see what we can think of.

Also - what kind of testing do you run and do you run it on the CAM or on the indevidual Gerbers?

Posted: Mon Mar 14, 2005 9:24 pm
by wiml
It seems to me (me being another optimistic software guy) it shouldn't be too hard to make some kind of simplified freedfm-like interface. Just enough to catch the easiest 80% of user errors. Panelizing might be harder...

What kinds of things do you have to take into account when panelizing? Just the outlines of the boards? Other stuff?

PHP / Java Script

Posted: Tue Mar 15, 2005 1:53 am
by lucky
Hi wiml,

I was thinking PHP to upload the files and check for errors including drill file problems then the PHP will send the good files that passed to Sparky via sendmail() email.

This solves the checking problem but as stated before I would need to look at the makeup of a CAM file before I can think about how to automate that. It too could prob be done in some clever PHP / Java Script.

We need to see:
- A set of files that you would expect to receive from the customer in a Zip file.
- A complete full Panel CAM file that is made up of the single PCB from the Zip file repeated over and over until the Panel is full.

Posted: Tue Mar 15, 2005 2:49 am
by leosedf
Why dont you use OrCAD Unison suite 10 which has great gerber software and the design is very easy?

Posted: Fri Apr 15, 2005 8:08 am
by pittuck
being a php person, how exaclty do u need drc files to be checked? If possible i would work on something like this for the price of a 3inch square PCB + UK cheap postage ;)

I think that $2.50 per square inch is great, but $5 would lose its appeal to first timers.

Perhaps, $2.50 per square inch + $5 support (option on buying PCB). U check em for both support and no support options, BUT without support option u just get an email back with there is a problem somewhere it will be $5 for checking or ask on the forums...

Posted: Fri Apr 15, 2005 8:44 am
by Jay3
I, personnaly, think that:

1) You need a way better faq-style page. Having to search a forum for questions make room for customer "mis-understanding", and much Sparky wondering what the hell went wrong with a particular design.

2) You would need a freedfm style applet. I know all files form gerber and such are text-based, making them so easy to script-process. Searching for clearance, hole size, and border (with .gko) violations would be real easy to catch. Automate the process so the faulty user receive a mail "design contains error X at Y".

3) Maybe an overcharge around 5$ for wrong design sent while fault explained in FAQ page. This said, every time an error comes from something unexpected, you add a topic to the faq concerning said error. In a month, said FAQ should cover all the infos one could ask, and designs would be easier to do, and to check. The 5$ will only make newcommers more aware of their responsability to toroughly read the FAQ.

I really think this service is great. You could even charge more. Do whatever it needs to keep this kind of service availlable. Being forced to pay a full 20inch.sq pannel for a 1"x0.8" design is an aberation.