DEX eval by free_electron

[nctnico]

The CAD software I use allows to manipulate a lot of data through spreadsheet like forms. It's very powerfull. I never place pads on a footprint using the mouse.

[free_electron]

@ntcnico : I Can do that too but that is not the point here. The goal is to find out what are the various graphical manipulations we can do. You want to take it back to numbers using spreadsheets. I want to stay away from numbers as much as possible by letting the machine figure those out. My method of working is: OK, computer , from here, i want something that is such and such; you are the mighty calculating machine that can do 4 billion operations per second: do it.

Like i said : datasheets are often severely lacking in the information you really need and you spend some time figuring out some dimension you need. The example i gave is before was : i really need the center to center spacing for two columns. Unfortunately the drawing mas made by some mechanical CAD guy and he gives met the toe to toe spacing and the spacing from one heel to the center. If you cad tool is flexible enough so it is not restricted to needing the center to center, but can work from random origin points this problem goes away. Simply set the origin to the heel of the pad , go in the given distance and use that as the origin point for the copy. Paste the copy, mirror it and paste it. No calculations required.

[zapta]

My method of working is: OK, computer , from here, i want something that is such and such; you are the mighty calculating machine that can do 4 billion operations per second: do it.

Ideally the such and such should be the exact constraints lifted from the footprint spec in the datasheet. In eagle pads are specified by center/width/height and rectangles by min/max x/y and I often need to first derive the values from the datasheet by doing side calculation. For me, this is the least fun part of designing a PCB.

[free_electron]

EXACLTY ! Zapta understands what i am talking about. Such drawings are made by mechanical guys and are HELL for PCB designers. For that reason you need a CAD program that has a rich set of tools. Some measurements are given relative to non-center points. So it is very important I can tell the CAD tool: for the next set of manipulations we will work from here, on this kind of grid. Sorry computer, i have only this number for you: it is that far from the edge of this, to the corner of that. One number is given in millimeters, another in mils. Be happy it is not in cows per acre.

Computer, you figure it out. This hole sits 34 degrees rotated counterclockwise from 0, with a radius of this number here. The next hole sits 12.5 degrees rotated further and 1 cm inward. That's all they gave me. I'll be damned if i am going to do polar to cartesian calculations, you are the computer: figure it out.

Computer, from: to compute. Meaning: person or machine that calculates. I will feed you the numbers I have and you should be able to handle it.

We live in a fast paced world. 3D Printers, maker spaces, laser cutters, flow jets, you can get anything online you can dream of. Things that were unimaginable 5 years ago. Thing you could not get your hands on.

Wonderful ! But when it comes to making a simple footprint we still are given a drawing with half the information missing.
And then one needs to try to get it into a CAD program. That better be one hell of a program that has the tools to work from the little data there is.

-Somewhat Off topic below - , not reflecting or criticism on DEX here . Just food for thought and idea's.
Not everyone may have access to the needed machinery and I do not expect DEX to cover the scenario below. It is too far out, or... maybe not ?

Inter-object relations is what parametric programs like Solidworks and Catia excel in. You can specify relations and constraints between anything you want and the darn thing figures out all by itself what that looks like. If you alter one of the relations later everything follows. You have given the rule set: A relates to B like this,  B relates to C like that. I Shift B and the CAD tool knows what to do. More, it tell's me if it can't be done. I'm sure you can find cases where you have drawn up something, go to a machinist and the guy looks at you and says : you figure out how to drill a triangular hole and i'll make it for you. Or: you buy me the 90 degree drill bit and i'll drill the hole. Such software will trap that.

The same holds true in the PCB world. You set up rules: Distances, widths, keep-outs. And then you throw a via in the middle. The software should be able to figure out how to move everything so this can be done. It has the rules. It can try billions of permutations per second until it finds one that works. It may not always work, but I'll be happy with a 80% success rate. I'll be very unhappy if it can't even attempt that.

I Don't expect it to be able to route the entire board. I May not have given it nearly enough information to do that. Mathematically speaking routing is one of these problems that are extremely difficult to solve. But, the software should be able to do some simple things based on some simple rules I gave it.

Like : A always needs to be this far from B. If i move A: you adjust B (differential  pairs). It is ok if you can't do it everywhere,  like going around a via or connecting into the actual connector. Best effort is fine. 80% Is better than zero. Another case could be: A needs to be the same length of B give or take this tolerance.(length matching)

You should see some of the connectors I have to work with. The mechanical drawings are hair pulling !

For example : http://prd.sws.co.jp/components/en/detail.php?number_s=60985503

Fortunately we can get STEP models for all of them.

Enter 3D mode. Create snap point from 3D vertexes , click on the center of the pins you want. The CAD software snaps to the center of 3D surfaces automatically. Once all points are marked: go to 2d mode. Now simply down pads on the created snap points. Done. That simple feature is a tremendous time saver.
The simple operation of being able to extract a coordinate from a projection of a 3D object. Piece of cake for a computer. Nightmare for a human.

I'm sure most of you have had 'technical drawing' classes at one point in your life. Like: take a cone, slice it here and there under this angle, now draw the side view projection. Didn't you just hate that ? And it had to be done in Chinese ink on vellum with those annoying Rotring or Staedler pens. When you were almost done you would move your hand just a but too early. Just before the ink was dry and smear it all over the page. There goes the entire afternoon, and it was due tomorrow.

Hooray for computers ! I Ritually burned those pens once I had a computer and a plotter.

I did a connector last week that was so scary we even exported a STEP file back from the PCB and had the board 3D-printed. Simply so we could stick the printed board in the enclosure and put a physical sample of the connector through the holes. Just making sure it would fit, before we ordered 40 prototypes of the PCB. 

This is a custom made connector and we were not sure that the STEP file we got matched the real deal. Since we had the actual connectors we ran the exercise.
We printed a plastic 1:1 model of the board, including all the parts on it. Every footprint we have has an accurate step model attached anyway. Put the 3D printed board in a 3D printed version of the enclosure and we were good to go. Only takes an hour on the printer we have. Better safe than sorry.

The bare boards are too expensive to play 'will-it-fit' games, and the design-cycle too short.

8 layers : 2 ounces of copper per layer, 3000+ via's. The board has 100 ampere running through it and has a bunch of back mounted flip-chip power transistors that need to press against the heat sinks in the chassis. Over 400 components in total. I needed 5 of the layers just to run the power in and out of the transistors.

You may think this is an extreme case, but, think about this:

Many hobbyists (maker is such a weird name) tinker with UAV's (drones, quadcopters etc...). How cool is it to pull in the chassis model of the drone and be able to extract a board contour and make sure everything fits in there. Then export the PCB design back to the body modeling software. To find out where the holes for connectors need to sit.
 I've seen this being done at the local Techshop. There's several drone builders there that already work like that. Purely for hobby.

So, these may not be so far flung idea's or techniques after all. They are already here and in use.

Like is said : food for thought.

[timb]

Try requesting it as a wish.
http://kov.com/Support/WishList

Dude, add it to the wish list yourself. You have people posting fantastic feedback here and you refuse to act on it unless they post it through the software/website. Obviously it goes into some sort of CMS/Ticketing system, so just add it there yourself.

It's like you go out of your way to be as unhelpful as possible. Reading through this thread and seeing your mood swings makes me seriously think you need to up your Lithium dosage. You're more bi-polar than a Tantalum Capacitor (and prone to explode like one too).


Sent from my Tablet

[Iliya]

Try requesting it as a wish.
http:/kov.com/Support/WishList

Dude, add it to the wish list yourself. You have people posting fantastic feedback here and you refuse to act on it unless they post it through the software/website. Obviously it goes into some sort of CMS/Ticketing system, so just add it there yourself.

It's like you go out of your way to be as unhelpful as possible. Reading through this thread and seeing your mood swings makes me seriously think you need to up your Lithium dosage. You're more bi-polar than a Tantalum Capacitor (and prone to explode like one too).


Sent from my Tablet

If you don't want to use the wish list then you are free to not use it. There is no need to be abusive. 

[nctnico]

My method of working is: OK, computer , from here, i want something that is such and such; you are the mighty calculating machine that can do 4 billion operations per second: do it.

Ideally the such and such should be the exact constraints lifted from the footprint spec in the datasheet. In eagle pads are specified by center/width/height and rectangles by min/max x/y and I often need to first derive the values from the datasheet by doing side calculation. For me, this is the least fun part of designing a PCB.

I usually print these kind of footprints enlarged and then calculate coordinates from one center point. For things like LED rings I wrote a small program which can calculate the coordinates and rotations based on the number of leds and size of the ring.
With that information I simply punch in the coordinates of the pads.

[mswhin63]

Dude, add it to the wish list yourself. You have people posting fantastic feedback here and you refuse to act on it unless they post it through the software/website. Obviously it goes into some sort of CMS/Ticketing system, so just add it there yourself.

It's like you go out of your way to be as unhelpful as possible. Reading through this thread and seeing your mood swings makes me seriously think you need to up your Lithium dosage. You're more bi-polar than a Tantalum Capacitor (and prone to explode like one too).

Sent from my Tablet

There could be potentially many people on multiple forums, including DEX forum that wish something. If you have ever worked in a business as an administrator then there is always a requirement for workflow.

A forum is not a place to get new ideas presented when there is an official section allocated. Iliya has already implemented and updated many aspects of this post, especially the legend for co-ordinates and angle and quite a few more.

What Iliya has done is put aside other wishlist to satisfy this forum post. There is a couple of very interesting wishes on the DEX forum that have been put aside or at least held to satisfy what would be considered more a rant than and legible post.

Your comments above prove how easy it is to ignore what you want. I would do exactly the same.

Try to pull your selfish head out of your arse.

[BradC]

I usually print these kind of footprints enlarged and then calculate coordinates from one center point. For things like LED rings I wrote a small program which can calculate the coordinates and rotations based on the number of leds and size of the ring.
With that information I simply punch in the coordinates of the pads.

I know this will sound like complete overkill, but for stuff like that I draw out the spacing in autocad (step and repeat, or some LISP if its complex). I can then export the DXF and load it into the PCB cad package allowing me to drop components on the right center points. For things like keyboards it makes it very easy to get the PCB layout dead on the mechanics.

[nctnico]

I follow a similar strategy but I usually (re)draw the mechanical outline on a mechanical layer in the PCB package for things like milled slots and other cut-outs.

[Iliya]

My method of working is: OK, computer , from here, i want something that is such and such; you are the mighty calculating machine that can do 4 billion operations per second: do it.

Ideally the such and such should be the exact constraints lifted from the footprint spec in the datasheet. In eagle pads are specified by center/width/height and rectangles by min/max x/y and I often need to first derive the values from the datasheet by doing side calculation. For me, this is the least fun part of designing a PCB.

I usually print these kind of footprints enlarged and then calculate coordinates from one center point. For things like LED rings I wrote a small program which can calculate the coordinates and rotations based on the number of leds and size of the ring.
With that information I simply punch in the coordinates of the pads.

Used snagit to copy from this posting and pasted into DEX.
I will list how to soon.

[Christopher]

That's not how you do it ! How do you know if the scaling of the drawing is correct?

What I do (with eagle, so only one grid) is redraw the suggested footprint on paper with all the dimensions one central point.

Iliya: I think what you're doing with this program is great but as you are in the infancy and you should take what Free_Electron says seriously but with a pinch of salt. It does seem like he is trying to shoot your program, but all of it is relevant advice.

Since I've started using Altium, I am falling in love with the STEP file import. All manufacturers use STEP. I don't even know what XDF or whatever format it is you're using. I now hate having to use primitive napkin software like EAGLE for my old designs. Being able to export a STEP file, with all the connectors mated with no hassles (Anyone remember eagle 3d?), into Solidworks is awesome.

[Iliya]

That's not how you do it ! How do you know if the scaling of the drawing is correct?

To scale the drawing.
1.   I measured the actual length of the 9.8 dimension (picked longest) Say this was 12.43
2.   Set the scale to 9.8/12.43 using the handy built in calculator in DEX.
3.   Lock the background image.
4.   Set units to mm
5.   Set snap to 0.1mm
6.   Set the origin to the center
7.   Added horizontal and vertical guides (construction lines), used dimensions on image to position them.
8.   Set snap to guides only.
9.   Added pads.
10.   Deleted guides.
11.   Unlocked and delete background.

[Christopher]

My point was the PDF datasheets will not always be scaled correctly and any designer worth his weight will not ever do this

[Iliya]

My point was the PDF datasheets will not always be scaled correctly and any designer worth his weight will not ever do this

I only use the background as a visual clue as to where to put the guides. I position the guides and then enter the guides precisely. (floating point)
Keeps needing to look at papers and is quick.
I know the image is inaccurate. The greatest error is anti-aliasing pixel errors when zoomed but it does not matter.
I can even add DEX dimension lines as checks.

[nctnico]

That's not how you do it ! How do you know if the scaling of the drawing is correct?

To scale the drawing.
1.   I measured the actual length of the 9.8 dimension (picked longest) Say this was 12.43
2.   Set the scale to 9.8/12.43 using the handy built in calculator in DEX.

That is a very bad idea. Pads must be in the right location within 0.05mm. There is no way you can do that by scaling a drawing (which usually isn't to scale to start with).

[Dave Turner]

I'd like to post a note of caution when copying and pasting data like this.I would definitely check that the stated measurements match all those that you measure in whichever PCB program you are using.

In the old days of drawing boards (I've kept my Rotring and Staedtler pens) there was good reason to write "Do Not Scale" on all drawings. Now, that most if not all drawings are produced by CAD by experienced professionals  , one should be able to rely on the accuracy of the drawings.

Me I'll check and recheck or most probably draw it again especially and definitely if the import method to my program isn't DXF or equivalent such that the individual elements aren't precisely defined.

I'm not - but if I were a PCB designer and just assumed that an import was correct when it wasn't without checking it; that could potentially cost big time.

[Iliya]

That's not how you do it ! How do you know if the scaling of the drawing is correct?

To scale the drawing.
1.   I measured the actual length of the 9.8 dimension (picked longest) Say this was 12.43
2.   Set the scale to 9.8/12.43 using the handy built in calculator in DEX.

That is a very bad idea. Pads must be in the right location within 0.05mm. There is no way you can do that by scaling a drawing (which usually isn't to scale to start with).

I use the image to roughly place the guides (construction lines) then I enter the exact x or y location. Then add the pads, they snap to the very accurately placed guides.
The image gives me a visual feedback that the arrangement of guides is correct.
When the pads are finally placed I see the also align with the images taking into account the errors of the image.

[Iliya]

I'd like to post a note of caution when copying and pasting data like this.I would definitely check that the stated measurements match all those that you measure in whichever PCB program you are using.

In the old days of drawing boards (I've kept my Rotring and Staedtler pens) there was good reason to write "Do Not Scale" on all drawings. Now, that most if not all drawings are produced by CAD by experienced professionals  , one should be able to rely on the accuracy of the drawings.

Me I'll check and recheck or most probably draw it again especially and definitely if the import method to my program isn't DXF or equivalent such that the individual elements aren't precisely defined.

I'm not - but if I were a PCB designer and just assumed that an import was correct when it wasn't without checking it; that could potentially cost big time.

As a double check you can add DEX dimensions.  I do not rely of a accurate image.

See 2 guides added at -4.9, and 4.9 and DEX dimension in the attached image.

[Iliya]

My point was the PDF datasheets will not always be scaled correctly and any designer worth his weight will not ever do this

I only use the background as a visual clue as to where to put the guides. I position the guides and then enter the guides precisely. (floating point)
Keeps needing to look at papers and is quick.
I know the image is inaccurate. The greatest error is anti-aliasing pixel errors when zoomed but it does not matter.
I can even add DEX dimension lines as checks.

See attached part.

[armandas]

Video Posted :

http://dl.dropbox.com/u/62657892/Making%20a%20PowerSO20%20package.mp4

The goal of this video is to show some basic cad manipulations like playing with reference points, grids , scaling and snapping to quickly create a relatively complex structure.

Nice video, got some good ideas from it. The only thing I can add is that if you use the Heads-Up display, you can see the delta distance from your last click/action and don't need to count grid lines any more.

[Dave Turner]

All - my point is not about DEX, or indeed any other CAD program, but any assumption that a drawing not created by oneself is both accurate and precise.

One may scale an image in both X & Y dimensions to be apparently correct for the any of the given dimensions but that is not a guarantee that the other dimensions are in proportion or in the correct relationship to the rest of the construct.

I comment on this because in another field entirely I was almost caught out by a similar issue.

[nctnico]

That's not how you do it ! How do you know if the scaling of the drawing is correct?

To scale the drawing.
1.   I measured the actual length of the 9.8 dimension (picked longest) Say this was 12.43
2.   Set the scale to 9.8/12.43 using the handy built in calculator in DEX.

That is a very bad idea. Pads must be in the right location within 0.05mm. There is no way you can do that by scaling a drawing (which usually isn't to scale to start with).

I use the image to roughly place the guides (construction lines) then I enter the exact x or y location. Then add the pads, they snap to the very accurately placed guides.
The image gives me a visual feedback that the arrangement of guides is correct.
When the pads are finally placed I see the also align with the images taking into account the errors of the image.

Still being able to just punch in the coordinates works much faster than placing pads using the mouse. If I need to place things accurately I usually end up using the arrow keys on the keyboard.

[Mechatrommer]

Dude, add it to the wish list yourself.

There could be potentially many people on multiple forums, including DEX forum that wish something. If you have ever worked in a business as an administrator then there is always a requirement for workflow.

the requirement for a workflow is important if you have multiple departments. for one man army like this, i agree with timb, the admin can add it himself. not being selfish but thats sensible... if all the wishes are ever going through the workflow, the one man army bandwagon will be overwhelmed, so its the same story. otoh this thread is not just about wishlist...

That's not how you do it ! How do you know if the scaling of the drawing is correct?

To scale the drawing...

we know that classical trick, but the point is (regardless if datasheet drawing is in scale or not), thats not how engineers do their things, bitmap scaling is more suited for artists.

To scale the drawing...
3.   Lock the background image.
10.   Deleted guides.

just to name a few, you forgot to use the spell/grammar checker here (point right back at you) jk

[CatalinaWOW]

Got away from this thread for a couple of weeks (as I said earlier this is a hobby for me) and was amazed at the blowups.   Found a few things that seem worth a comment.

1.  Many complaints about the user interface.   Complaints like this are of some value, but recognize that people are different.  Good for one doesn't work so well for another.  And what works for one is heavily dependent on your own background/training.  Example is Microsoft's Office.  I got very proficient on the Office 97 style interface.  Hated the new ribbon interface to the point I that I switched to Open Office to avoid it.  Then got a job which required use of the new version.  After a few months of forced learning found that it wasn't so bad.  Even better for a some things.

2.  Dimensional woes.  I have to add a vote to those that pointed out that we are stuck with mixed units for a long time to come.  Program support for this would be very valuable.  Perhaps in the form of default units that can be overridden by adding unit definitions.  This scheme would still allow Mars lander type mistakes, but would at least let the computer do the work when asked.  Anyone who is in danger of making billion dollar mistakes probably should spend more than $100 bucks on their software.

3.  Complaints about clumsy or difficult ways to select, move or modify elements.   Yeah, sometimes it is clumsy, but in every case I have found it is possible.  Would I like better.  Sure.  If I liked it enough I would pay the big bucks for one of the high end programs.  When you pay rock bottom prices you can't expect Ferrari features.  Some of the comments liken these problems to having square wheels.  To me it is more like doing auto-maintenance/repair without a hydraulic lift, a refrigerant capture machine and the manufacturers diagnostic analyzer.  It is harder, and some jobs just have to be deferred to the pro shop, but we can't all afford a complete pro tool set in our garage.

4.  Lack of a paste mask.  This is really a substantial flaw, and I haven't figured out a viable workaround.  This problem wasn't obvious to me since I do all my paste applications manually, but it really needs to be there for anyone who sends boards out.  It is almost inconceivable to do even hobby scale boards these days without surface mount.