another board layout

Ha, you are using PADS

We use that too and I often tease the CAD guys how crappy a tool it is

The SW is built in different steps, originally on Unix command line style, so that?s why the integration of functions is so bad

I like to use Altium, it?s really intuitive

Cheers

Klaus

We've been using PADS since its first DOS versions. It will still open the old files.

Some of the user interface is a little quirky, but we're used to it, so it's intuitive to us. It's fast and stable, and I really like the schematic editor. It can i/o Autocad (hence Solidworks) files, which is great for checking boards against boxes.

We don't auto-place or auto-route. The physical re-use thing is great; lay out one channel and it replicates the others.

I don't think another program would save us much time. We spend our layout time thinking and designing, not fighting the tools.

Everything, but everything, can be output and input in ascii. So we can do stuff off to the side, with an editor or Python or something.

One of my guys creates FPGA logic decals in Python. That saves days of typing and tons of errors. I've written a few PowerBasic programs to do stuff like validate PCB part numbers and values against our stock database.

It's a great tool. I've used layout software that wasn't.

I was just hunched over my screen, staring intensely, finger poised on my mouse, and Angie handed me an ice cream sandwich. Life is good.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Who willingly does layout without schematic? How badly do you hate your layouter?

Doesn't look like more than 4 layers to me, unless that BGA is more balls than pixels. Hard to tell at this scale.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

"Having the schematic in your head" means that you actually understand the circuit. That's a big help in getting the layout right. ;)

Having the same layout person for at least a few years is a win--you can develop some shorthands. For instance, round our shop we have "the pink treatment" (named after a pink highlight marker), which means that all traces so marked have to be kept inside a Faraday cage--a built-up shield on top, a nearly-featureless pour on the bottom, and lots of via stitching.

Along with toroidal inductors and minimum loop area, the "pink treatment" lets one put switchers within an inch or so of a 10Mohm, 3 MHz TIA (say) with no ill effects.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

What's cool about that is, so does everything else. Altium actually imports old OrCAD, Mentor, Protel, etc. very nicely. Newer files (say from the last

5 or 10 years, like PADS 6+ I think?) usually need to be exported in some usable (ASCII) format. Except for Eagle files, which it can open current versions directly without a hangup.

Every CAD tool worth paying money for, can read DXF/DWGs (and if not, what are you even paying for?)

Oh, wow, I just noticed in that screenshot, the toolbar buttons are all blown up. They look like it's 1998 again. Wait, on closer inspection, I bet they're under 256 colors too -- make that 1992. Yikes!

(Yes, yes, you'll complain back that the UI appearance doesn't matter as long as you can use it, which you've already learned how to use, so it doesn't matter. Well, that may be fine in the present case, but you obviously haven't tried, literally any E&M simulator in existence, for example. Physicists are *ATROCIOUS* UI designers...)

You must be using a very old version indeed if that still feels new. Even NI Ultiboard has it! Forget if Altium left it in Circuit Studio, maybe. If so, that's a /free/ PCB tool with it.

Certainly true of Altium Designer, as well as having internal scripting interfaces in basically every language. I suppose it's pretty well true of Eagle as well, with the ULP files and (probably fairly well RE'd?) native file format and stuff.

IIRC, PADS has a few user / "notes" fields on parts, and that's about it? AD has unlimited parameters, key-value pairs effectively, that you can add to parts. All are available for BOM generation, so you can one-click turn out a bill, including live supplier stock and cost. You can import it in Digikey and order parts within the hour.

The one thing I don't think you can get, at least not easily, is a spreadsheet that has live functionality built into it. I'm not sure if you can do Excel expressions in parameters (i.e., make an "Excel Expression" column, and put an "=SOMETHING()" expression in it, and have it work as a literal), or if there's a way to construct a template containing expressions that expand along with the contents (e.g., a row at the bottom showing total BOM cost, calculated from quantities and unit costs).

That would be easy enough to add with a postprocess script though.

Just saying, I'd take the sandwich over the tool any day. ;-)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/

That's adjustable. I have pretty bad vision, so the big icons suit me. How many colors do you need in an icon? Animation? 3D? Do they sing when you hover over them?

I have used Sonnet, but and it was pretty bad, but that's not actually a physics program. If I ever had to do a serious multiphysics sort of thing, I'd hire a consultant. I design electronics.

Our material control system includes a folder for each part where we stash data sheets, notes, measurements, photos, web links, anything useful. That applies to all parts, not just things that go on PC boards. I wouldn't want to use a PCB layout program as our inventory database.

All are available for BOM generation, so you can one-click turn

PADS makes BOMs in our database format.

We almost always design boards based on parts that are already in stock. Currently, that's 6900 separate parts, total 3.3 million pieces. We discourage adding new ones unless they have some serious benefit. If I want a 6.3K resistor, which I do now, I'll put a 6K in series with 300 ohms. It will take 0.5 seconds for the pick-and-place to slam down one more resistor.

One of our rules is that we don't depend on anything that could go away next year, or anything that's stored offsite. We're shipping stuff that was designed 20 years ago.

Why are people so emotional about tools? I'm designing stuff that we sell quantities of, beautiful boards that work first time, and PADS works fine. That's what's important to me. An old wood-handle hammer drives nails just as well as a fancy USB fiberglass Bluetooth LCD-display-equipped hammer. Or better.

I never use Excel. I don't play with tools. I design electronics.

Too many engineers prefer to play with, or invent, tools instead of getting the work done.

I have one board that I did a few years ago. Concept, schematic entry, layout all in one weekend at home, with PADS. It's close to a million dollars in sales now.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Of course we have a schematic. But he has to look at it while he's placing and routing, and I don't.

A schematic doesn't tell everything, like the best place to put bypasses and terminators, or thermals, or trace widths, or aesthetic issues.

I said it was 6 layers. This fits into one of our standard boxes, and it's only an 11 sheet schematic, so density is low.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

e your

Knowing the circuit is not the same thing as knowing the layout. In analog the schematic is drawn to be clear in a single 2D layer. On the board it is often double sided and many layered. Not the same thing in many ways.

Digital is yet another animal where crowding of signals is often a road blo ck. Many times the best way to route does not come from the schematic whic h often is more like a net list with names on the wires on these pins and a ir connections by name to another sheet in the schematic. In those cases y ou won't know the layout issues until you try to lay it out. Trial and err or.

You are providing evidence of not needing to know the circuit design to do layout. The circuit designer can convey the useful info to the layout desi gner as you are illustrating here.

I do like doing my own layout, but it is amazingly difficult work when in t ight spaces. My board was highly space limited and I was barely able to ge t it routed without running the layer count higher. The next one will be m ore relaxed I expect.

Rick C.

I did a board recently that has some GaN fet gate drivers that float on the fet sources, and that node slews about 20 v/ns. An optocoupler feeds the drivers. Each pF of capacitance to ground jams 20 mA into a node. I sandwiched all the signal and gate traces between two copper pours that are nailed to the sources node, which is a planar Farady cage. Seems to work.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Yeah. So?

I don't do a lot of boards like that. Our stuff is 75% analogue, so the issues are circuit strays, oscillation, bootstrap gain, transfer function weirdness leading to settling artifacts, like that.

There are other items, e.g. not hanging bypass caps on pigtails. "Route the trace _through_ the bypass capacitor's pad, not merely nearby."

Enjoy. I don't much like doing layout, because it consumes gumption better spent elsewhere. The price I pay is having to go over the Gerbers super carefully.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Yup. Laplace's equation works in your favour in that sort of geometry. It's easy to get the unwanted strays down into the single-digit femtofarads like that. In my experience it's easy for a single 0.8-mm pad sitting over a very mildly bouncy supply pour to blow my measurement right out of the water. (That last happened in about 1995, but I remember it vividly--after the fact I calculated that the offending stray was about 50 fF.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

I wish I had tools to quantify this stuff. For instance, if a trace is just inside the outlines of my planar Farady cage, which has an open end, the "ground field" reaches in to my trace in some sort of fringing pattern. I guess I could run ATLC or something, which is a lot of work. Vias are even worse, 3D effects on multiple layers. I mostly guess.

If I did have some super EM analysis tools, it would maybe be more trouble to set up (and to learn how to drive it) than guessing and getting on with it.

Surface-mount pads usually have tiny areas hence low capacitance to the next layer, or to the universe. But traces make things much worse. I can't bury the parts inside the Farady cage! I can punch a hole in the ground plane so the parts see the top layer of the cage below.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

alog the schematic is drawn to be clear in a single 2D layer. On the board it is often double sided and many layered. Not the same thing in many way s.

Schematics are just connectivity. Layout is topology. Not the same thing at all.

block. Many times the best way to route does not come from the schematic which often is more like a net list with names on the wires on these pins a nd air connections by name to another sheet in the schematic. In those cas es you won't know the layout issues until you try to lay it out. Trial and error.

So 25% digital isn't enough to pay attention to?

an

nk

a

do layout. The circuit designer can convey the useful info to the layout designer as you are illustrating here.

Isn't that layout 101? I recall dealing with a layout guy who didn't want to run a trace to a resistor voltage divider that was measuring the power s upply voltage. It had been drummed into him that power planes were *never* branched out like that. Rather than try to get past his bias, I had him move the resistor so the long trace was *after* the resistor and so not a p ower plane pour. lol

in tight spaces. My board was highly space limited and I was barely able t o get it routed without running the layer count higher. The next one will be more relaxed I expect.

Not sure why your supply of gumption is so limited. I'm pretty sure if I h ad paid someone to do this layout I would have ended up with extra layers o n the board or a poor design in some other ways. I almost gave up myself o n the 2.0 revision. Eventually I found I could move some parts so a bit mo re pads were opposite other pads freeing up a bit of space for through hole s. I wonder if auto-placement considers that sort of thing.

Rick C.

The selling point back then was that it needed a lot fewer mouse clicks to do anything, and I found that to be true. (The DOS version was the only one I used because it was free.) Does it still take fewer clicks? The newer (free) tools I've used seem to have copied PADS a little in that respect, but not completely.

I haven't used other software much lately, but I do like the PADS schematic entry. It really understands that the thing that joins two nodes is a connection, not just a bunch of line segments that might get lucky and intersect on their ends. Making a connection in Pads is very different from other ones that I've worked with.

Mike says that LT Spice originally used Orcad for entry, and that it still suffers from that legacy. Its schematic editor is barbaric compared to Pads Logic. It will leave line segments anywhere, and make apparent connections that don't actually connect. It needs a coarse grid and the page-size cursor to be usable at all. [1]

The layout part of PADS works fine. I can't imagine a newer package saving us much time, or making better Gerbers.

The layout connectivity checking could be better. It will, for example, show an error on a ground connection that's actually OK, when the actual error is far away. One missing ground via creates a connection error, which shows up as *two* flagged errors, the real connection error and another at the other end of the rubber-band connecting node or something. Both ends of a theoretical connection. One gets used to that.

I remember hand-taping layouts on layers of mylar. A board check could take two people two days. Now it takes a few seconds.

[1] In LT Spice, sometimes I start a wire and get one immovable x-y giant cursor frozen on the screen, plus the movable one. That's really annoying.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Why so? Yours is too. I didn't say the supply was _low_, just that I have better uses for it than PC layout. Layout is slow and boring.

I'm pretty sure if I had paid someone to do this layout I would have ended up with extra layers on the board or a poor design in some other ways. I almost gave up myself on the 2.0 revision. Eventually I found I could move some parts so a bit more pads were opposite other pads freeing up a bit of space for through holes. I wonder if auto-placement considers that sort of thing.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Laplace is your friend here. A hole or a slot lets a bit of field through, but it decays as exp(-2 pi z/w), where w is the feature width.

So it goes away very very fast.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Hmmm... if supply is not limited then you can spend it *everywhere*. I don 't find a problem with using my time and effort to do board layout. Then I know the layout is the best it can be for my requirements and every level of trade off. That's often important in my work. Rather than trying to do a *lot* of jobs, I try to focus on the ones with high rewards. So far tha t has worked very well. :)

Layout can be hard. I find I have a bit of a talent for it. I enjoy it, n ot boring at all, rather like a puzzle. I also do puzzles for fun and ofte n build things other than electronics.

Rick C.

Well, if you like doing layout, at least it's more productive than the NYT crossword.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com