Author Topic: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data  (Read 3874 times)

0 Members and 1 Guest are viewing this topic.

Offline bullet308

  • Frequent Contributor
  • **
  • Posts: 341
  • Country: us
  • Jack of All Trades, Master of None Related to EE
Just watched this...I am lead to believe that the board fab industry is sick to death of having to deal with either proprietary file formats from the CAD vendors or with antiquated and inadiquate file formats such as Gerber, etc and so they have been trying to impose some measure of standardization on the CAD industry for, what? 40 years? And 2581 is their latest and most comprehensive effort yet, and they are still running into resistance. Is this a reasonable interpretation on my part? Or maybe they are about to finally break out of this sorry state?

Even on my end, as the lowest level of newbie amateur just fiddling with this stuff, the whole issue of various CAD file formats and all the complexities that brings is annoying beyond words. I wish them well, but the commitment of the CAD vendors to their own CAD formats and their efforts at cutting their competitors at the knees is counterproductive for everybody, ultimately. The IT world of 2012 is built on a foundation of open standards, yet the people that build the software that makes the hardware that our modern world runs on is stuck in a mindset more appropriate for 1970. Not good. :-/
>>>BULLET>>>
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 27947
  • Country: au
    • EEVblog
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #1 on: September 13, 2012, 09:56:54 pm »
Yup, you nailed it.
The CAD companies don't want to change. They sign up to these things because they don't want to be "left out" and it's cool to promote you support these types of standards, but the follow-though commitment leaves a lot to be desired.
But really, how hard is it to write an exporter for a large CAD company?

Dave.
 

Offline firewalker

  • Super Contributor
  • ***
  • Posts: 2312
  • Country: gr
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #2 on: September 13, 2012, 10:06:34 pm »
But really, how hard is it to write an exporter for a large CAD company?

Dave.

Trivial.

But it's not only this. There was a case that a company sued a developer that designed an exporter to a common format. I can't exactly recall the case, or if was electronic CAD related.

"Our format is the best and should be used instead. But we want it to be secret".

Crazy things...

Alexander.
Become a realist, stay a dreamer.

 

Offline JoannaK

  • Frequent Contributor
  • **
  • Posts: 337
  • Country: fi
    • Diytao making blog
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #3 on: September 13, 2012, 11:19:39 pm »
One thing I noticed .. How everybody at that lecture hall seem to hate Mentor ...

Ah well, don't matter me much. Except they did purchase PADS and that *was* the pcb cad packege I used over 5 years.
 

Offline mikeselectricstuff

  • Super Contributor
  • ***
  • Posts: 11802
  • Country: gb
    • Mike's Electric Stuff
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #4 on: September 14, 2012, 12:24:08 am »
I think there's just too much installed base of different systems doing things differently to stand any chance of a reliable any-way interchange format.
The only other standard that comes to mind in this area is DXF, which is probably the most unreliable, inconsistent standard out there.
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7029
  • Country: us
    • SiliconValleyGarage
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #5 on: September 14, 2012, 01:10:15 am »
Whoa, hold it. Ipc2581 is a complex beastie.

Here is the problem that 2581 tries to solve.
We have schematics , stored in some proprietary format dpeneding on cad tool. This schematic references to components that will have to be bought somewhere. Modern cad tools link to procurement tools... The schematic may have additional pointers for layout. Like : this is differential pair , that is controlled impedance at 65 ohms. The link schematic to pcb is through a netlist. That netlist only ports over layout relevant info .. Who actually makes the resistor or where we buy it is irrelevant for the pcb layouter. And then the pcb tool spits out gerber. Further down the pipe it would be interesting for the bare board tester to have the netlist as opposed to have to extract it ( this is what odb++ solved ). Now we get into pcb assembling. We need info about part values. Pick and place files solve this only to an extent. For full cycle control you need to dig and scratch for more info.
Different software tools use intermediate file formats to do data exchange. And every cad vendor has his own. Move on to electrical testing and rework. It becomes an even bigger hairball. Test fails. Turns out a vishay part was placed instead of a tdk ... What was specced in schematic ? Was it overruled by the beancounters ? I s there a design variant ? Paperwork paperwork and intermediate files.... Coming for. Different tools , only exchanging partial info and leavingit up to the humans to 'glue it together' and p,ay sherlock holmes.

That is what 2581 attempts to solve. 1 format that holds ALL information about a board assembly. From part procurement , electrical interconnect, rules , everything. Using this 1 dataset you can get an anser to any question you have at any given point in the flow from schematic concept to a packaged , tested and ready to ship board.

No more need for intermediate files , no more data lost 'in translation'...

Note that 2581 does not hold the actual schematic , nor is it a fileformat to story polygons and tracks that make up a pcb layout. It is a database that tells you there is a part from vishay , this prtnumber, procured from this vendor , that has pin 1 connected using a 60 phm controlled impedance line that originates at coordinates so and so and terminates at coordinates such and such. The value of the part is 1nf with this tolerance. The part sits at coordinates so and so with this alignment.

Using this single dataset the bare board testers, the pick and place robots, the final board testers and any other machine or step required in the actual production of the board ( from the guy picking the prepreg and copper foils and needing to know the thicknesses of each to the final machine that verifies all parts are soldered down and the board powers up) have access to the data they need, and they also have the relation of that data to other data. You can extract pick and place data fro. This file , you can extract gerber from this file , you can extract netlist , bill of materials and many other things from this single file. The only thing not in there is probably what that component symbol looks like on the screen and its coordinates, so you can't extract the schematic . Although i think they even provisioned for that...

That is the attempt.
Note that they have been at it for at least a decade .... And we need to see the first tools that are feature complete....
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline bullet308

  • Frequent Contributor
  • **
  • Posts: 341
  • Country: us
  • Jack of All Trades, Master of None Related to EE
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #6 on: September 14, 2012, 01:45:16 am »
I was under the impression that 2581 has the capacity to store the schematic as well, but you had the option of including or deleting various parts of the package depending on who you were sharing it with. Or am I wrong?
>>>BULLET>>>
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7029
  • Country: us
    • SiliconValleyGarage
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #7 on: September 14, 2012, 02:55:52 am »
As understand it 'storing the sechematic' relates to connectivity information and parts information. not the actual graphics representation of it. the graphics could be 'extracted' from it. the same goes for the PCB. there is a textual description : track , 4 mil from x1-y1 to x2-y2 , rounded edge , solid. a graphics viewer can parse this infor and recreate the graphics. ( pretty much what a gerber viewer does, but with much more info in 2581 . like gerber only contains circles for a pad. 2581 has actual full description of the pad)

since a schematic symbol for a resistor is 2 pins with some squiggly lines between them you could also store this in a fixed textual format ( lines are defined by x-y coordinates. electrical hostpots on the symbol too. so yes you coudl store the parts symbols inside a 2581 object and link that object to placement information , procurement , connectivity (netlist ) , routing restricitons ( impedance control on signals, routing topology etc ).

This one dataset could be read by any 2581 compliant package and anything you need can be extracted form it. there would be only one dataset that is shared between all packages.

i think the toolvendors are going to fight this becasue they would have to compete on tool capabilites only. there would be no more lock-in due to being stuck to a sepcifiec fileformat. i used orcad to do this project so i need to keep orcad around in case a revision comes along. you wouldn't care. if a 99$ eagle version could open this whole project... kiss orcad goodbye..
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline ModernRonin

  • Contributor
  • Posts: 44
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #8 on: September 14, 2012, 08:33:17 am »
> That is what 2581 attempts to solve. 1 format that holds ALL information about a board assembly. From part procurement , electrical interconnect, rules , everything. Using this 1 dataset you can get an anser to any question you have at any given point in the flow from schematic concept to a packaged , tested and ready to ship board.

That doesn't sound like it's going to enable trivial copying of anything that gets sent out for manufacture.

Though, I guess you could argue, exactly how difficult is it to knock off things right now...
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7029
  • Country: us
    • SiliconValleyGarage
Re: EEVblog #349 - SMCBA Lecture IPC-2581 Open Standards for PCB Design Data
« Reply #9 on: September 14, 2012, 11:03:07 pm »
You would not give out this dataset of course. The tools would let you export only selected information.
The data that would go to the pcb maker wpuld be a subset. He doesnt need procurement or pick and place data , but he does need impedance control data .. Something that you cannot do today in one uniform set.
Same goes for the assembly house. These dont need impedance co trol data , but they do need placement and procurement data, and sometimes connectivity data... But not necessarily schematic or actual pcb layout. Netlist is fine.

So each of your subcontractors gets a filtered database. The dataformat remains the same , you only get the chunks you need so at no point is your design in jeopardy of being cloned. The full set remains in house. And you can control who gets access to what.

Schematic designers are not allowed to create a part symbol. That is for the someone with library clearance. When the part is made , procurement data needs attaching before it can be placed in a schematic etc ...
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf