Wrong unit when importing ECAD, ecad lib?

[don.ordo]

Hello!
1 I am trying to import ecad (idf 3.0(emn,emp)) from Cadence Allegro to ProE Wildfire 2.0 and I do get the blocky boxes as components, but somehow I always end up with inches as units. Both Allegro and ProE is set on mm. Is there anything that I could have missed (probably..)?

2 I am also looking for any kind of library of pre made electrical parts such as resistors, caps, connectors, IC's etc. Any ideas? I know there is a website selling a package, anyone tried that? Dont know if I am allowed to write the address so I block some letters out www.simplified*******.com

This is my first post on my first job so be kind

 

[Dell_Boy]

In brief because this is a BIG topic to try to cover

Units for components are set in the .emp file exported out of Allegro

.ELECTRICAL
3216 CAPT10U_10V_3216_20P THOU 80.0
0 -105.0 -45.0 0.000
0 -105.0 45.0 0.000
0 105.0 45.0 0.000
0 105.0 -45.0 0.000
0 -105.0 -45.0 0.000
.END_ELECTRICAL

you may be able to override this via the config.pro setting for

template_ecadpart

In reality it shouldn't matter whether the imported extruded block part are in metric or imperial. As long as they are the right size they should import correctly into the assembly and you should only treat them as a stepping stone until you can replace them with accurate parametric parts anyway.

If you use someone else's parts they won't have the origin c_sys in the correct place or orientation. There is also a good chance they will be any of:- overmodelled, undermodelled, incorrect or have datums for Africa.

With simple parts such as chips Rs and Cs there is far more work linking the ECAD to the MCAD database than there is in the modelling and for ONLY these components I would suggest having a single MCAD component to represent all 0402 chips etc.

For more complex parts I would recommend having a seperate MCAD part for each ECAD part even if it is a visually identical member of a family table.

Overall I would not recommend using someone else's database of
components. By all means use whatever you can get as a starting point
but tailor them to your requirements and rebuild them if necessary. It can be a lot of work to do it correctly, especially the initial set-up but it will save you a lot of grief in the long run

B.T.W.
Molex have a lot of Pro/E parts on their website and if you are a corporate customer of AMP/Tyco you may be able to get access to their connector modelling service but beware because even these sources can have important errors in their CAD models.


DB

Edited by: Dell_Boy

 

[Zestje]

Hello,

If you have 'inch' dimensions instead of 'metric' dimensions after importing an ECAD file' (e.g. every mm is only 1/25.4 mm), you can change this from:

<ul>[*]'edit'[/list]<ul>[*]'setup'[/list]<ul>[*]'units'[/list]<ul>[*]under 'system of units', choose the units you need and then click 'set'[/list]<ul>[*]now make sure you choose 'convert dimensions', to really convert the dimension values.[*]then confirm until back at the main screen[/list]
hope this helps...


Edited by: Zestje

 

[don.ordo]

Thank you both Dell_boy and Zestje!
I have followed you advise Dell_boy, I now started model my own electrical parts and I think it is going quite good!
And I now import everything correctly with mm!
Thanks again.

 

[3DPCB]

Dellboy,


After about a dozen years of doing ECAD/MCAD imports and opening my company which focuses on ECAD/MCAD integration, I disagree with many of your points:


The key to having the correct orientation of components with respect to ECAD data is to follow standards on the ECAD side. IPC-7351A orientation standards are the way to go(for surface mount components, pin 1 in the upper left corner with the coordinate system in the center of the component...for thru hole, pin 1 is also in the upper left corner with the coordinate system through the center of pin 1). If your ECAD data is oriented using this standard then you can use off the shelf models like those at www.simiplifiedsolutionsinc.com


There is also a good chance they will be any of:- overmodelled, undermodelled, incorrect or have datums for Africa.



I agree with this, however, it is quicker to start with something overmodeled and remove rounds than have too little information. In general, PCBs do not get that big because they use the same components over and over.



With simple parts such as chips Rs and Cs there is far more work linking the ECAD to the MCAD database than there is in the modelling and for ONLY these components I would suggest having a single MCAD component to represent all 0402 chips etc.



This is a really bad generalization. 0402 chips come in many different heights. In one case, the component may be 0.5MM, in another it may be 1MM. Unless your design allows for this variation, this is an extremely bad idea...and most of today's products are so dense, it will not be an ideal scenario for most designs.


For more complex parts I would recommend having a seperate MCAD part for each ECAD part even if it is a visually identical member of a family table.


Again, I disagree...one to one correspondences are a bad thing unless the ECAD naming convention accounts for it. For example, if the ECAD_NAME is so14...the MCAD part can have a number of different height variations. If the ECAD_NAME is so14h155 (with a height of 1.55MM) you can get away with the 1:1 scenario.

Overall I would not recommend using someone else's database of components. By all means use whatever you can get as a starting point but tailor them to your requirements and rebuild them if necessary. It can be a lot of work to do it correctly, especially the initial set-up but it will save you a lot of grief in the long run


In general, I would agree, however Molex has excellent data on their site, and Simplified Solutions, has extemely accurate models with an easy to use interface that also creates your ecad_hint.map file.


Regards,


Keith Richman ([email protected])

 

[Dell_Boy]

A clarification:

My posting was a list of suggestions and recommendations based upon about 10 years of importation. They are NOT rules because there is no chance they will suit everyone.

Standards are great which must be why there are so many of them. Just because you use IPC-7351A doesn't mean that everyone else in the world will and even then a particular standard won't work for ALL situations. Hell, even the "pin1 upper left" is ambiguous for most 2 pin devices, it needs to be qualified with something like top first, left second. Some leaded components get used in SMD fashion, some SMD components don't have defined centres, some placed components don't have pin 1s. Some components don't even touch the PCB. Even some manufacturers have trouble with standards; the AMP/Tyco cPCI connector pin numbering does not match the cPCI standard so which pin 1 do you choose - the standard or the datasheet.

In all my years of dealing with AMP/Tyco models I cannot recall ever getting one with the coordinate system in the right place/orientation; many did not even have z in the correct direction. I have had VENDOR-supplied models with important mistakes and others that are difficult to simplify due to inappropriate creation techniques and others that can't be because the "Pro/E" model is an imported STEP file. If you can use them; great but don't bank on them being always that way.

At one stage I was dealing with 6,000 component motherboards and running interference checks with two 2,000 component daughterboards. I think these would qualify as large and substantially strengthen the argument against overmodelling.

If the height difference between Rs and Cs is important to you, go ahead and have different models, it was only a qualified suggestion. I used to have a single model for Rs and a single, thicker one for Cs modelled at the maximum allowable thickness before they were susceptible to tombstoning. After all, Rs are usually isolated on the upper surface and Cs are normally live which in itself is enough of a reason to treat them separately.

Different ECAD systems may force different mapping requirements. With something like UniCAD the ECAD_NAME and the ECAD_ALT_NAME are almost always the same. With Allegro ECAD_NAME is the package name and ECAD_ALT_NAME is the schematic name and they are almost never the same. My general preference for Allegro was for the instance name to match the schematic name (beware illegal characters) and the generic name to match the package name I'm the first to admit there were many exceptions to this rule

From your comments on on 1:1 naming, I think you have mis-read a section of my post and with a vested interest like yours, I think I can see why you appear to have taken an skewed view of my other suggestions.



DB


P.S. do you want to buy some of my database of ECAD models.


Edited by: Dell_Boy

 

[3DPCB]

Dellboy,


You obviously have a lot to offer with regard to this topic. From years of working in this area, it is obvious that individuals working on a particular product line can best define a process that works for best them. Our business makes suggestions and fills in the missing pieces. This is definitely not a one size fits all scenario.


You seem to have perfected a process that works in the development of your product...like you mentioned, a computer PCB has much different requirements than a handheld-device PCB, or a PCB in a military product.


With regard to IPC-7351A orientations, most companies seem to be working their way over to it on the ECAD side. Only ourlargest customers are not migrating...this is mainly because of the cost of migrating the large amount of legacy data. The standard calls out orientation for two pin items and just about any other type of component. In my opinion, the orientation standard (and new naming convention) is very effective when utilized correctly leading to a huge cost savings on the ECAD side.


Regarding Amp/Tyco...I agree, use their online information at your own risk, or at least understand how the models differ from the spec. In contrast, Molexhas extremely accurate data and I can not remember a Molex part that I encountered where the dimensions on the spec do not match the database. With any downloaded model (except ours

) you will have to modify or create a new coordinate system.


As far as overmodeling, I say overmodel...PTC created Simplified Reps, Family Tables, and the supress featurefor this reason. You can always "dumb down" a model quickly. Adding detail can be time consuming but the data will be there when you are in a crunch. I can't tell you how many times I had a potential customer that encounters an issue because they were working with "dumbed down" or "outright on the fly data" Almost every time, components have incorrect coordinate systems or "on the fly components" have the incorrect height. The scenario typically works out with a resistor (say 0805)...the PCB may have 100 different resistors with the name res0805 many with different heights called out in the emp file. The first item is built from the emp file and has a height of 0.5MM...consequently all parts in that res0805 family are imported at that 0.5 same height. Meanwhile, several of the components actually have a height of 1, 1.5 or 2MM...which is not reflected in the PCB. If the individual used the name res0805h200 as the ecad name, the 2MM resistor would be imported on the fly.


With regard to the 1:1 naming convention, I have no issue with it. However, the problem lies with process mentioned above. Most companies use simplified naming conventions for ECAD Names that do not include height. If this is the case, 1:1 naming conventions often results in improper height data regardless of whether the Engineer uses "on the fly" data or replacement 3D models. There are very few companies out there that have good processes/naming conventions with respect to ECAD/MCAD integration. From a business perspective, those that haveprocesses in order do not subscribe to our products, consequently, we gain nothing by promoting a lack of 1:1 match.All of our business comes fromefficiently assisting those that don't have full time employees dedicated to this process. There is no hidden agenda here...the only goal is to provide an excellent process or service to our customers.


Feel free to contact me offline about your last question.


Keith

 

[Dell_Boy]

My personal dread was radial electrolytic capacitors as the same "footprint" could relate to parts ranging in height from medium to skyscraper. Relating the MCAD name to the ECAD_ALT_NAME bypassed these sorts of problems caused by ECAD database people not considering height variations.


I would call the practice, sufficient modelling which means capturing all the important information AND suppressing the stuff you don't need 100% of the time.
E.g. a compact PCI connector as it is typically imported with 31 features.

View attachment 3296

And the same connector with all the features resumed in order to check the PCB hole pattern and pin alignment with the mating connector (and for pretty pictures for the marketing department) with 97 features

View attachment 3297

And as a contender in the overmodelled category weighing in at a mere 8,4MB and slower than a wet week in HLR we have a Molex 74332

View attachment 3298

As it was only available as a STEP file it is highly impractical to simplify.


DB

Edited by: Dell_Boy