If you could automate one part of your design workflow, what would it be? We are asking e14 in our Join, Share & Win Competition

• versioning
• generate BOM
• create FAB files from (Ki)CAD
• automate physical testing as soon as board starts up
• API documentation generation
• firmware build and unit test
• release creation

if security / IOT related

• certificate management

Getting the coffee to the desk. 

The thing I would most like to automate out of existence is: Kitting parts to build new pcbs.

I make lots of pcbs, average > 20 per year. 

Each may have between 50 and 200 unique parts. The build process is:

Design - using pcb CAD

Make purchasing BOM - PCB CAD + Excel + VB app that groups parts together

Buy the bits

Using the pick list (created by more VB app that munches a BOM into a per side pick list

SPEND THE NEXT DAY COUNTING BITS OUT OF BAGS, TUBES, AND LITTLE DRAWERS INTO LITTLE POTS FROM WHERE THEY ARE PICKED TO GO ON THE BOARD

Stencil print solder paste onto bare pcbs

Reflow the board

That picking of parts is not automated at all - the time it takes is independent of the number of boards and mainly a function of the number of unique parts. Almost everything else is computer assisted but that picking process is grim. It isn't helped by the way some suppliers (and I do mean Farnell) put parts in bags folded over, done up with two bits of tape and then the label stuck over the tape and where th bag folds over. So you can't get to the bits without ripping the label in half !!!! (I'm getting quite cross about it now !!).

Anyway - automating that horrid process would improve my life. So far I've been developing better ways of keeping parts so they can be indexed and locate quickly but it isn't the total answer.

MK

michaelkellett said:

That picking of parts is not automated at all

You'd almost want to farm this out to a part picking company. They use lower skilled staff to unpack assorted items and prepare them for more automated/organised processes.

Because this is hard/expensive to automate with software or machines, you have to rely on a service that does this reliably, with staff that's not engineer graded.

I build quite a few boards each year (250+).  I spend almost as much time in bringing up and testing the boards as I do in fabricating.  The process of testing the boards (visual inspection, power on testing and processor ISP tests), followed by configuring the uProcessor and USB devices (programming fuses, initializing EEPROM, loading bootloader and operational firmware) followed by functional testing (comms testing, performance testing, calibration and other required tests/setup) is quite burdensome.  It would be nice to automate these steps through the use of some combination of in-circuit testing and automation of programming the boards (across multiple software packages) and finally automating the necessary test equipment (programmable power supplies and loads, sending and receiving USB commands, and data collection). This would definitely speed up theses time-consuming tasks and give me more time to pursue other engineering for products and projects.  

That solution would be (possibly) viable for small volume (batches of 100 or so) production. 

Typically I make between 2 and 10 boards in a batch so the cost of using a third party would be very high. 

And of course as soon as you involve any other company in your production flow it becomes a new bottleneck with handovers and extra documentation.

I've found that using third party board assemblers always involves more work  - they (not unreasonably) can't make decisions about substitute parts so every time there is a query or an out of stock part you end up with more delay (and often paperwork).

One option that is gaining traction is using the board supplier to put the parts on but the nature of my work means that my boards usually involve at least some unusual parts. And as soon as any parts are soldered on then stencil and reflow can't be used on that side of the board.

What can work (but is horribly expensive) is to put parts into an computer controlled carousel type bin system - but I'd need maybe 2k bins. I have seen a system with electronic drawers that might help (but I can't find it now).

MK

michaelkellett said:

Typically I make between 2 and 10 boards in a batch so the cost of using a third party would be very high. 

You are in that unhappy space where you cost too much, but involving anyone else costs too much too.

Not really unhappy, because you have the business. But scaling up the customer base per board could bring savings.
But then, if the customer wasn't niche, they probably would not knock at your door?

michaelkellett said:

What can work (but is horribly expensive) is to put parts into an computer controlled carousel type bin system - but I'd need maybe 2k bins. I have seen a system with electronic drawers that might help (but I can't find it now).

I think the costs would outweigh the benefits in your business model?

(also: the customer seems to want to pay what you charge - so savings would benefit you)

I'd like to say "PCB routing", but despite how frustrating it can be, it can also be very satisfying.
A slightly sillier answer would be "finding and picking up small screws and parts that fall to the ground". That happens to me *SO* much while assembling prototypes or final boards, and it can either be a 2 second task, or a full 30 minutes all-hands-on-deck "search and rescue" operation. It doesn't help that SMD packages are very good at literally disappearing and being gone forever the moment they leave the bench.