Do tables of XY coordinates tarnish the souls of engineers?

[daqq]

I'm drawing a footprint from a datasheet. There's a recommended footprint - which I'll base my own on - and of course you have to calculate the position of the pads from the drawing. Because reasons.

Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads? OK, sure, if there's some weird shaped pad then OK, but this is a bunch of rectangles with easily defined centers. Why is this preferred? Is there something awesome about this that I'm missing?

[Siwastaja]

You are not missing anything. That has been requested for at least three decades straight. It's a tradition. Additionally, it's pure sadism.

Because every manufacturer does that, no one can vote with their money.

Also it's small enough problem so no one will base their competitive edge on such small detail.

Some managers and engineers fear change; some may actually get satisfaction out of bullying their customers. In any case, it sucks, and continues to suck.

Actually, you don't even need an xy table to implement what you suggest: sanely chosen reference lines (at part origin) and measurements drawn to this line suffices. This strategy could be used to satisfy those who require things not to change, yet provide a readable drawing. The fact that 99% of the datasheets do not do this IMHO proves that it's just a case of bullying for no other reason than satisfaction stemming from sadistic soul.

And it's not only about how to represent the data. In 97% of datasheets, part of important data is missing, namely the origin, which is highly important for SMT components. It's the place where the P&P nozzle is supposed to touch. The drawing should have a crosshair there, all measurements shown relative to this. Now it's up to the designer to guess where the P&P origin is, and up to the CM to use their experience in modifying their customer's files to increase the yield based on testing. This is absolutely unbelievable.

[Benta]

"Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads?".

Yes there is, but you need to know a bit about mechanical engineering to understand why:
Dimensions need to refer to measurable points. An imaginary "center" is not one. You also need a reference point for the dimensions (there are several ways to define this)
Edges, corners, diameters, angles etc. can all be measured physically for tolerance and quality control. A "centre coordinate" can not. You can derive it, but that's not the same.

[Kremmen]

"Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads?".

Yes there is, but you need to know a bit about mechanical engineering to understand why:
Dimensions need to refer to measurable points.
[...]

Maybe so but this only shifts the responsibility from the one manufacturer to the numerous individual users. Because in many EDA programs you just cannot use the dimensioning style chosen by the manufacturer. So such strict principle is quite useless in practice.

I fully agree with the previous sentiments, although i would add, quoting Hanlon's razor: Never attribute to malice that which can adequately be explained by stupidity.
If not actual stupidity or sadism, then certainly hidebound tradition and inability to listen to the paying customer.

[tom66]

I think it's a mechanical engineering tradition.  In my day job I deal with several mechanical engineers and none of them like these diagrams.  They're drawn to have only one set of measurements for every possible dimension, which eliminates rounding/precision/typographical errors (in theory) but they can make finding some dimensions incredibly difficult.

[sleemanj]

If your PCB package supports it you could import the image, scale it so that a horizontal and a vertical direction are correct and then draw the pads directly on top.

If it doesn't then you can import the image as a reference into an appropriate cad package, scale to two of the dimensions and sketch on top then you can just measure directly any point or dimension that you wish. 

Of course this assumes the original drawing is to scale, which it probably is but sensible to check against the dimensions printed on the drawing.

[Brumby]

I think it's a mechanical engineering tradition.

This.

Drawings have been created like this for decades, beginning in the days of T-squares and set squares.  Upgrades with mechanical aids did nothing to change the outcome, just the speed of creation.

[gnuarm]

I see many references to the center of pins and such. 

[Brumby]

No-one is saying centres and axes don't get used - but look closely at where they are used.  There will usually be some very obvious reasons why.

Pin centres are easy.  This defines pin pitch - and easily allows for different pin diameters without complicating things too much.

[Siwastaja]

"Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads?".

Yes there is, but you need to know a bit about mechanical engineering to understand why:
Dimensions need to refer to measurable points. An imaginary "center" is not one.

Why not? You just came up with a completely arbitrary rule.

Your post is an excellent example of the sadism I described.

- We need thing X. X is easy to do. X solves our problems and costs nothing to implement.
- Can't do X.
- Why??
- Because can't.

And of course centerpoint is measurable and non-imaginary. Heck, in mechanical engineering, holes are referred to their midpoints, not to the hole walls, all the freaking time! There's nothing different in pads.

[filssavi]

I think it’s more lazyness and tradition than anything else

The MCAD workflows are designed to automate a certain typical style of diagram, which is the traditional one, for anything else you probably would have to fight with the tool, or at least take three times as much as you would do normally. This would be worthwhile if the creators of said diagrams were also the users, however I am willing to bet  in a vast majority of cases the mech engineers never even spoke with a PCB designer, let alone tried himself.

And for the silicon vendors, there is absolutely no economic incentive to change things

[Rerouter]

I think its more a case of its one of the last steps, so slap down enough measurements to make it reproducable, but no real thought as to making it easy to model,

Atleast on the up side, more manufacturers are starting to produce footprints for various CAD packages. even if you still have to verify them,

It seems USB connectors / SD card connecors / Sim card connectors are some of the worst offenders as they have different offsets as that keeps all the pins the same shape in manufacturing.

Kicad hs some plugins that let you quickly rebuild it doing the intermediate steps for you, but I haven't used them myself.

[penfold]

When you are verifying the footprint, it is much easier to measure horizontal and vertical distances to the edge of a pad or feature than it is to guess an arbitrary centre or make two measurements to average (maybe your cad program has the option to place the pad origin away from its measured centre or perhaps one might be doing this in Gerber format... the component manufacturer doesn't know).
There may even be some valid engineering reasons to offset a pad or alter its dimensions slightly in which case you are perhaps more interested in verifying that you haven't ridiculously exceeded the recommendations of the edges of the pad w.r.t an origin that's easy to measure to or even a creepage/clearance constraint, again which need to be measured to an edge.

Circles and holes are also a special case for so many reasons, but I believe they're a Gerber primitive or most Gerber viewers can pick up the drill file to place centre markers.

I would imagine the reason they have been presented for so long in such a manner is that manufacturers have never been able to find a compromise between the "information you need to build the footprint" and "information that's more helpful when measuring" so they simply give you the information that they would use to verify your footprint should you complain to them.. and leave it at that to avoid confusion.

[Berni]

Yeah i absolutely hate these kind of drawings too.

Id say its just laziness. The thing likely exists as a drawing in a 3D mechanical CAD. When turning it into a drawing the designer can easily click edges and plonk down dimensions. So he just clicks enough edges to make it unambiguous. Drawing centers would require extra work to do.

But yeah some of it is probably that mechanical engineers find this form of measurements makes "more sense" because its how its usually done, and that way tends to work better for simpler drawings of most mechanical parts.

[Brumby]

The documentation stage in any IT project has to be one of THE least tasks to be given appropriate time allocation - especially if the project has made it to implementation.  A management decision.

I can see the same "logic" occurring in other disciplines.

I'd bet there are many engineers just grimacing at the drawings they've been forced to leave because they're "good enough".

[VK3DRB]

Some footprints are disgusting, most are medicore, some are excellent.

To be frank, many footprint drawings are made by people who have never done PCB design. Altium is by far the world's most popular professional PCB design tool. Better footprint dimensions that are Altium-friendly should start from a datum reference point, such as the centre of Pin 1 and everything is dimensioned from that point. Manufacturer's should realise helping the electronics designer helps sell their product as does a decent website.

I tend to avoid vendors who provide crap datasheets or don't think you are worthy enough to have a 3D step file for their component. 3D models help me verify footprints and mitigate risk from mistakes. You can get a hint of the bad brands when Digikey has no link to the 3D models. For some 3D models, you have to register or even email some sales department. I won't use a part (especially connectors!) if there is no 3D model, whether it be from the OEM,  SnapEDA, 3D Content Central, Octopart or elsewhere. There is no excuse not to use 3D component models in a PCB design these days. Building crappy extruded models for components is a poor man's approach and is a recipe for errors. It tells me the designer was time desperate or is lazy. Besides, PCB layout is ARTWORK that should be created by a skilled craftsman with pride in their work, not by a slap-dash tradesman.

I often modify recommended footprints for improved reliability or manufacturability. I NEVER use those Altium default footprint library component from last century, but I do sometimes use their footprint generators for things like BGA's, but I always use a small filled triangles for pin 1 markers. The triangle is superior to the round dot in a number of practical ways, especially on high density boards. There is no valid argument why the old round "fly-vomit" dot should be used over a triangle for a pin 1 marker.

Why a triangle? It is an inherent arrow head that points to where pin 1 is, reducing ambiguity, unlike a dumb arse dot or circle. For those with poor vision, a triangle stands out out clearly even if a lot smaller than a dot, where a dot looks like a via, fly vomit, a spec of dust or a solder ball. A triangle helps debug technicians and manufacturers easily verify pin one without resorting to a separate a drawing or using a magnifier. Everyone I know who has used triangles never go back to circles unless they inherit someone else's library. Not having pin 1 markers or hiding IC orientation information under a populated IC, or omitting a ground testpoint could almost justify a sentenced of installing Windows 95 floppy disk version ten times a day for one year onto an old PC with a noisy fan.

[Rerouter]

I kind of agree there, 3D models for connectors, and any large throughole components are a strong contender for if I will consider a part or ignore it. footprints I am still more leniant seeing as there are so many different cad packages, and they won't have an easy means to export them, but they will have 3D models in house, and those can be easily / automatically exported

[ConKbot]

I'll take a table if they stop sticking with mechanical drawing tradition of no redundant information. Connectors are especially bad at this. Have to find how to get from one feature to another via adding, subtracting, dividing 3-4 other dimensions to get the one I need. Just mark the convenience ones as reference only, so the tolerance stack up can still be unambiguously calculated, and everyone is happy.

[Nominal Animal]

"Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads?".

Yes there is, but you need to know a bit about mechanical engineering to understand why:
Dimensions need to refer to measurable points. An imaginary "center" is not one.

Why not? You just came up with a completely arbitrary rule.

Your post is an excellent example of the sadism I described.

This is ubiquitous.  Humans love arbitrary rules.  Benta is not evil or sadist, just being an ordinary human behaving as normal humans do.

My favourite example of this is how browsers nowadays no longer let you choose UTF-8 as the default character encoding.
Why? "Because it is not a legacy encoding." Who decided that? "It's obvious, isn't it? Everything except Unicode is legacy."
Yep, the reason boils down to someone making up a new rule, and now that it exists, it is wrong to break it.  All current browsers now follow this rule, even though it is a rule the browser users have no use for, and that only causes friction and troubles, solving nothing.

[Terry Bites]

It's good that there is a reason why manufacturers like to be difficult. Why don't they just convert their mechanical production drawings into simpler formats that are engineer friendly? I've got to say electromechanical are the worst offenders of all in this racket. ARRRGH!

[daqq]

Thanks, I assumed as much. I'm familiar with mechanical drawings and I understand that there it makes sense. But these are drawings that have one purpose - to be fed into a PCB design tool, the bulk of which simply work with centers.

It's not a huge issue, I can work with that, but it's an annoying issue that would cost nothing to fix.

/rant

[VK3DRB]

...But these are drawings that have one purpose - to be fed into a PCB design tool...
It's not a huge issue, I can work with that, but it's an annoying issue that would cost nothing to fix.

But like I said, those who made the footprint know no better. They need to be made aware, ie: education.

Maybe a new format needs to be created where the footprint dimensions are in a table, something looking like an Excellon drill file. Then, irrespective of the CAD package, the footprint dimensions are simply imported with one click and converted to a footprint with accurate dimensions.

[nctnico]

I'm drawing a footprint from a datasheet. There's a recommended footprint - which I'll base my own on - and of course you have to calculate the position of the pads from the drawing. Because reasons.

Is there any reason manufacturers prefer this way rather than giving you a table with the XY coordinates of the centers of the pads? OK, sure, if there's some weird shaped pad then OK, but this is a bunch of rectangles with easily defined centers. Why is this preferred? Is there something awesome about this that I'm missing?

Like others said: it is how mechanical engineers work. Sometimes I get drawings like that for PCB outlines 
BTW: your example isn't the worst footprint drawing I have come across. Some can be real puzzles in order to reference pads to eachother and the package outline.  Having a CAD package which supports 3D models is a big plus. At least you can fit the 3D model on the footprint.

[CatalinaWOW]

I worked in a mixed engineering environment.  Those of us with EE degrees were called sparkies by those with ME degrees, who we called tin benders.  Both kinds were sure they knew how to do the others job better.  Me too, possibly with more justification than most since dad was an ME, and I switched from ME to EE after sophomore year at university.  Weird to think that I thought that contour integrals and Poynting vectors were easier than Hamiltonians and stress tensors.

The point of this is that it is easy to disdainfully say the other guy is being lazy.  That centers are easy, that is how centers of holes are specified, the nominal place that it is.   But there are real problems involved, and I suspect that verification of performance occurs far more frequently than creation of CAD models.  How do you measure the center of a real hole.  If the drilled hole is actually round there are geometric ways to infer the center, but real holes have imperfections including taper, barrel and higher order diameter vs depth variations and many types of eccentricity.  How do you define te center of an egg shape hole, or one not orthogonal to the surface?  It makes sense to make the more common job the easier one.  Supplying both is an option, but then you get the issue if which one is the controlling document.  Would you like a set of X, Y with an accompanying note with weasel words like "typical" or "refer to ... for true dimensions".

The suggestion on pick up crosshairs seems a good one, but I can also imagine problems with process variations. 

Finally, there was a move as I retired to move away from paper based specifications to model based specs.  Would a model of a part (a very fancy list of x,y s) make your job of designing footprints easier?  It promised to make the mechanical side easier.

[gnuarm]

I don't really mind too much the goofy ways parts are notated.  But what I really hate is when the drawings were taken from a CAD package that thinks lines are infinitely thin.  Seems the "official" PDF rendering is to draw them as 1 pixel wide no matter how much you blow them up.  The result is often lines that are so faint they can barely be seen. 

Some PDF viewers allow you to set a minimum line thickness in some way.  Others feel this is a problem with the author of the data sheet.  Obviously the people making the data sheets don't use the same PDF viewer I do.