advice on selecting new PCB design package

Right......... gimmee the names of some well known companies using gEDA on serious, mission critcal projects????

Prescott

OK Paul, are you talking from experience or what you think sounds right....???

You say "but people who have tried it (KiCAD) say it works well"....

what people...??

Tell us about the serious projects KiCAD has been used on..?? Can you give me the size of boards, number of layers, level of technology, value of projects...etc.

Prescott

I would like to jump in on this discussion. I have been frustrated finding an EDA program that is:

• relatively easy to learn/use (I don't do cards very often)

• Can do a EuroCard sized card (4 in. by 6 in. I believe)

• 2 signal 2 power construction, 512 nodes

• Schematic capture and PCB design

• Low cost to free (for moderate use).

My review of various programs has found:

• Kicad -- can't be used on WinME, have not tried Linux (and won't for now)

• FreePCB -- no schematic capture, but seems good otherwise.

• gEDA -- This group of well meaning people just don't have a clue! Just read the most recent Circuit Cellar. An hour to install on Linux! I don't think so. gEDA needs a consistent interface from program to program, an installer such as an open version of Wise, and a Windows version. Most desktops are Windows, so why not a Windows version?? A good user guide is also needed. This is a lot of work I know (this is what I do for a living, so I fully get what is involved), but is a requirement for gEDA to soar to the heights that I believe it is capable of doing. I would settle for a Knoppix CD with all of the gEDA programs and a user guide. The user guide can be an updated version of the articles in CC for a start.

• ExpressPCB -- propriety data format. Not acceptable!!

• Eagle -- Not the most user friendly program I have seen, especially for a "professional" one. The free version is very limited as well, but will do. This is what I'm using until one of the ones above offer a better deal.

Dave,

Ales Hvezda wrote:

You use WinME ?!? I didn't think anyone considered WinME a usable system, even for the simplest home use. I don't normally like it when people recommend upgrading your OS just to run a particular program, but in the case of WinME, I'd make an exception. W2K or even XP is a much better OS all round (assuming you want to remain with windows).

Incidentally, have you tried Kicad with WinME? I know the website says W2K or XP, but it might run fine nonetheless. Certainly the main tools and libraries (mingw and wxWidgets) are fine under at least Win98.

Yes, and it works well. Even for developing large VB6 desktop applications with database and multiple RS-232 ports going. Shock!

I didn't think anyone considered WinME a usable

I've given billionaire Gates all of the money he is going to get! My preference is for Linux once the desktop has as good functionality as WinME (here comes the holy wars). I want easy to use, consistent, etc. I know it can be done -- look at the Mac or Sun. I have used both and love them. For a number of reasons the PC is a better choice for now, and a Linux that works like a Mac or Sun would be great.

Yes, and I get unicode and other errors. My guess is that the developer uses WinXP. I found that doing development this way to be a big mistake. Better to use the lowest common denominator of O.S.'s and test with a higher level. Takes less time, offers a wider audience.

I know the website says

Newsgroup discussions like this are my own personal opinion, rather than those of my employer, etc., etc. But my company knows what software I use, and others here use lots of open source software.

I use a great deal of open source software professionally. So do most serious computer users who go beyond the limits of word processor, web browser and email. And everyone uses open source software behind the scenes, whether they know it or not - the great majority of internet infrastructure (mail routers, DNS servers, that sort of thing) is open source software, as are a substantial proportion of other servers.

Apart from "standard" open source software such as FireFox, Thunderbird and Open Office (which are to be found on many professionals' desktops), I make a great deal of use of open source utility software (such as cygwin), and I specifically choose open source tools for my embedded systems programming when such tools are available (and of a solid quality). If you want, I can go into detail as to why.

I didn't attack your logic - there was no logical argument but bold statements such as "common sense must tell you..." and "anyone who relies on software tools for the livelihood wasting time with them... No, of course not".

I don't disagree that gEDA or KiCAD is unlikely to be a sensible choice for a professional EDA designer. There will be occasional situations when such tools *are* the best for particular professionals, but not in general. In the case of EDA tools, open source software is not (yet) suitable for the mainstream.

What I strongly disagree with is your absurd claims that no open source software is suitable for any professional use. I believe in using the right tool for the right job, based on sensible decisions rather than knee-jerk reactions and PHB logic (such as "if we haven't paid for it, who do we sue when it stops working?"). That means using commercial software or open source software (or commercial implementations of open source software) as appropriate for the job.

You seem hung up on the idea of money. Open source means a great deal more than just money. Professional users of Linux, for example, are generally perfectly happy to pay hundreds of dollars for a solid Linux distribution from Red Hat or Novel/SuSE. If gEDA were brought up to a usable standard, they would have no problem selling CDs with basic support contracts and documentation for a few hundred dollars. While money is sometimes relevant (would I buy a compiler for $2000 when I can get one for free? Only if the commercial one was very much better in some way to justify the price), there are lots of other reasons for choosing open source software.

Let me give you an example of the use of open source software in a closely related field. If you use Altera's FPGA tools (or Xilinx's - they have a virtually identical setup), you'll find that most of the infrastructure of their tools is open source. They use open source languages like TCL and Perl substantially, and include a cut-down cygwin (posix layer for windows, along with a range of gnu and other open source utilities) installation. If you use their soft processor toolkit, it is based around Eclipse (an open source IDE), and the compiler and debugger are ports of the standard open source gcc compiler and gdb debugger.

By the way, if you want to use google groups for posting to usenet, it's helpful if you quote properly. I know the google groups interface is broken, but apparently the following works:

"If you want to post a followup via groups.google.com, don't use the broken "Reply" link at the bottom of the article. Click on "show options" at the top of the article, then click on the "Reply" at the bottom of the article headers." - Keith Thompson More details at: Also see

The PCB program in gEDA *is* 20 years old! But last year it got upgraded to Gtk, and over the last few months I've been segregating the GUI from the rest of the code, so we now support both Motif and GTK, and ...

... PCB's new API should easily support a native Windows and/or MacOS GUI. I think someone's already working on wxWidgets too.

I don't think gEDA is trying to compete with the big commercial packages. I think we're trying to compete with the small and medium commercial packages.

Money and time, yes - I fully agree! But I have plenty of motivation, and lots of people are providing direction, no problems there.

And we're certainly open to offers for contracting work on gEDA, if someone wants to provide the money so we can afford to spend more time on it. How much does it cost to have Orcad add a custom feature for you?

Does "common sense" tell you that you can't get an operating system for nothing? gEDA is still in what I'd call its early stages, but it shows plenty of promise and will -- hopefully -- eventually end up being as good a CAD tool as Linux is an operating system. That being said, while I think Linux is good, I also think that Windows is decent -- and worth the purchase price -- as well. Similarly, even when gEDA is a little more full-featured, that doesn't mean that commercial offerings won't be worth the money.

Uhh... you know about Open Office, right? It's not 100% compatible with Microsoft Office (in particular it doesn't support Visual BASIC scripting and the other forms of "automation" the office suite does), but for at least 90% of all Office users I'd defy you to find something Open Office doesn't do equally well (including opening and saving Office format files).

For that matter -- although it's something of a dying package -- TeX, LaTeX, etc. are considered by some as superior document preparation systems to the likes of Word, and these days most implementations are free or almost free.

---Joel

Please specify which clues we're missing. Saying it's bad is not constructive criticism, it's just criticism. Point out how we can make it better, and we'll add your suggestions to the list (we know about the ones below already). Please make sure you're using the latest version first, though.

That article was written a year ago, and a *lot* has changed in the software since then.

That's because the CD-ROMs have sources on them, not binaries, to avoid incompatibilities with your OS.

I think you can get them to work with Cygwin, but yeah, we know about that requirement.

I could say that about a lot of commercial packages, too ;-)

I think someone's mentioned a LiveCD with gEDA on it, on our mailing lists.

Dave,

Does this imply you're running WinME? I'd suggest it's *really* worth the money to pick up a used copy of Windows 2000 on eBay or similar and install it on your PC. :-)

Kinda reminiscent of the early days of Linux, no? :-)

I understand your frustation here, although it's not a show-stopper for me. (I used it once while working at Tektronix. Our group used it for non-critical test/prototype boards! They had enough bueaurcracy around and the "official" PCB layout department with their very $$$ tools took a dim view of engineerings wanting to layout their own boards -- even though going through the layout department usually have severe schedule impacts, i.e., a month to layout a 4"x6" double sided board.) My feeling was that ExpressPCB is targeted towards pretty simply boards anyway, so worst case having to re-enter the design into a better package was probably no more than a couple days work.

---Joel

Dave,

I want to be able to click on "setup.exe" and actually have the equivalent of start menu icons appears regardless of whether I'm running KDE, Gnome, or whatever else.

Given the development model of Linux, I don't expect there'll ever be just one desktop (as there is in 99% of cases for Windows, Macs, and Suns -- there actually are a few people hardcore enough to run alternative Windows desktops out there, but there of course are compatible with the Windows Explorer in terms of picking up the menus and icons from the right places) -- and that's certainly a good thing. Still, I don't think it's unreasonable to expect that all those guys get together to agree on a few more standards.

I hate to tell you this, but in 2006 it's a reasonable expection that Windows

2000 is the "lowest common denominator." Yes, I know, plenty of people still use Windows 98 or 95 (or even DOS!), but as far as new development goes, you just can't expect people to be writing software for anything over 6 years old, IMO.

---Joel

DJ,

...etc...

I'd just point out that -- while you certainly have a point -- better EDA tools that have keyboard re-mapping and automation features (e.g., a good macro language) can go a long way towards providing any new "feature" you want anyway, and yet you still retain compatibility with everyone else who's running the tool. Realistically you can't expect that anyone running gEDA will have any changes you've implemented other than those that were officially accepted into the "main line" distribution; as such configuration files and external (script/macro) interfaces can be much more "portable."

Let me guess... you'd ban Windows altogether "for security reasons" if you could? :-) I think your position is a little extreme -- with a decent network firewalling and automatic security patching, I'd guesstimate about 99% of security problems are with the end users (people consciously agreeing to & running a download, for instance) and not the software itself. I don't know what kind of company your the IT manager of, but in a company of engineering-level employees, it's downright insulting to try to dictate what web browser they might use. And of course many software companies require Internet Explorer to test their own products, since that's what their own customers will be using.

What, you don't like EDIF? :-)

PCB (well, the Xaw and Motif versions) has keyboard remapping, user definable menus, and limited scriptability already. I'm planning on letting you script mouse buttons too. I don't know if the Gtk folks will add that to their GUI.

The new API we're just finishing up offers a standard way to add modules to the core, without worrying too much about compatibility. Such modules at the moment are limited to GUI and export (print, gerber, png, etc), but there's no reason why we couldn't support import and "wizard" modules too.

Ok, now I'm thinking we could add dynamic linking support, and let you have your module as a .so or .dll. Sigh, more work to do.

I suppose you could add a script interpreter as a module, too.

I would hope that people who need changes bad enough to fund them, would help themselves and everyone else by getting their new features added to the official code set. That's how most of the features that are there got there.

But my point is you *can* choose that option. A more realistic option is to pay the gEDA developers to add what you want to the official sources, then you have it forever and don't have to worry about compatibility.

Oh my! Another open-source EDA flamefest of sci.electronics.cad! Dax's comments are very interesting, and he seems well informed regarding commerical offerings. On the other hand, the open-source discussion is giving me a feeling of deja vu.

I'll make a couple of points and then disappear again.

• D> Right......... gimmee the names of some well known companies using

Don apparently has a memory loss problem since he asked *exactly* this question last year on s.e.c. Here's the thread:

The point made to Don then (and now) is that this business about "mission critical projects" is a red herring. (Remember, Don?) GEDA competes against low to mid level EDA packages. Nobody uses low- to mid-level EDA packages for "mission critical" projects. That is, nobody will use circuitmaker, pulsonix, ExpressPCB, Eagle, Kicad, or gEDA for a 20 layer router board with 622MBps busses requiring matched-length tracks & 2.5Gbps diff pairs to an optical transceiver. If you're worried about "mission critical" stuff, then fork out your $20000 and use Allegro.

On the other hand, if you need to bang out a 4 layer test board in a hurry, then give gEDA a try. Or if you're a student, hobbiest, independent consultant, or professional engineer who needs to do a reasonably simple board in a hurry, it's worth it to try the open-source alternative. You can easily do a 6 or 8 layer board with a couple of hundred components.

Yes, gEDA lacks a couple of features (such as backanno from PCB to gschem), but this is well known and documented in the gEDA FAQ:

Open-source stuff is completely open about it's features and drawbacks. Is commerical software?

• Dave Boland says:

Folks often assert that gEDA (or other open-source software) is "difficult to use". This is a naked assertion made without any support; unfortunately, it creates the same effect amongst uninformed potential users as intentional, vendor-created FUD. Dave, can you please call out specific problems in gEDA holding it below a "usable standard"? I am particularly interested in hearing about stuff you notice while trying to use gEDA. If you haven't actually used it, then let us know.

GEDA is a collection of programs written by a confederacy of different programmers who have banded together to create a suite of tools useful to everybody. There are certainly UI differences between the different tools. OTOH, at the level of file formats and design compatibility, the design flow in gEDA is pretty seamless. No Perl duct tape is needed anywhere. This isn't true of lots of ASIC flows, but nobody every complains. Think about how different all the Xilinx FPGA tools are (when run from the command line)! Why do people beef about gEDA, but ignore Xilinx? Answer: the complainers don't have that much experience with EDA tools. Just MHO.

Here in New England, the most common board flow (in my experience, anyway) seems to be Viewdraw -> Allegro or Viewdraw -> PADS. In both cases, there is a large difference in the UI experience between the two programs. Like gEDA, the schematic capture program was developed by a different group from the layout editor. However, Viewdraw can export flawless Allegro and PADS netlists, and nobody every asserts that this UI difference is a problem. The same is true of gschem, which can export over 20 different netlist formats. Maybe 5 or thereabouts are layout netlists -- for open-source as well as for commercial (e.g. Protel) tools.

This topic is discussed at greater length in the FAQ linked above.

As DJ pointed out, this is (average) time required for the install wizard to run from begin to completion. The install wizard simply automates the process of compiling the software from source. You don't need to be there while the wizard is doing its thing; it runs itself, unlike a Windows install wizard.

Installing the software from source was chosen for the CD distribution since it eliminates many of the dependency problems associated with installing binaries onto unknown systems. It is also consistent with the GNU/open-source philosophy. If you want to install binaries, the following package types are also available on the gEDA website: Red Hat RPM, Debian DEB, and Mac Fink. Check them out here:

BSD packages are also available elsehwere.

This is an interesting one. It is certainly true that there is no good Windows version of gEDA. There are (at least) two reasons for this:

1. Nobody has stepped up to the plate to make this happen. Most of the gEDA developers are Linux geeks, and they are not very interested in producing a Windows port. "If you want to use gEDA, get a Linux box" is the reigning philosophy. If a new developer does a Windows port, he will be welcomed into the fold.
2. Some current gEDA developers also feel that a Windows version would open the floodgates to totally clueless newbies, who would then flood the e-mail discussion lists with dumb questions and support requests. GEDA is created by volunteer developers; they don't want to spend 10 hours/day holding the hands of idiotic and ungrateful newbies. If you want an example of what can happen to a totally-open EDA project running on Windows, read some of the idiocy posted to the LTSpice Yahoo group, or read the constant begging for free cores on the OpenCores development lists.

That's what the two articles in Circuit Cellar (March & April 2006) are about. There is also a lot of documentation on the web site, including official docs, tutorials, and an ever-expanding, wiki-based FAQ. See it all here:

GEDA is not difficult to use if you have some passing familiarity with EDA programs, or you take the time to read the docs. (Unbelieveably, we have seen total newbies try it without doing any reading, or having ever run any other CAD software before. That individual whined a lot on the e-mail list, but apparently got his project done.) Nor does it look like a DOS program. (What prompted that assertion?)

GEDA is available for free download and has been used (by me, for example) both at my day job as well as at home for my own projects. Unlike e.g. Eagle, the stuff you download for free is full-featured -- there is no crippled version of gEDA. I suggest that interested folks download it for themselves and try it out -- warts and all -- rather than believe FUD spread around on usenet.

Stuart

::gEDA needs...and a Windows version.

I appreciate the

*steering business toward Redmond is evil* mindset (or whatever makes a Windows port less attractive to developers).

The big hurdle you guys have to jump is getting folks to install Linux. ISTM, a bootable Knoppix-like release of gEDA would push things in that direction. People who already use Seamonkey (or Firefox/Thunderbird/Nvu), OpenOffice.org, mplayer / VideoLAN Client, the GIMP, gcc, GAIM, etc. under Windoze would get used to yet 1 more piece of utilitarian software and maybe that would further convince them that OS migration isn't going to be that big a deal anyway as far as lack of apps.

I'm hoping you guys give this notion more than a passing glance.

I don't think that's the issue. As a gEDA user I've gotten to know the developers (even gone out to dinner with some), and as far as I can tell they just aren't Windows developers. Personally, I use MacOSX on a PowerBook for portability and Linux on midrange commodity hardware for heavy lifting: gEDA works well in both those environments (needs Fink on MacOSX), so it suits me very well. gEDA's modular toolkit approach suits me well also: my customers want netlists and schematics (they have other contractors they use for board or chip design).

Somebody *did* port gEDA to Windows a few years ago, but nobody maintained the port. If the Windows EDA community wants gEDA, somebody has to step forward and do the work.

--
John Doty, Noqsi Aerospace, Ltd.

I won't speak for the others here, but my opinion is that open source software on proprietary platforms acts as a lever to convince more people to consider open source software in general, and specifically to consider OSS alternatives to their existing applications.

Thus, the goal still isn't to steer business *towards* Redmond, but to convince people (by example) to steer away from it. A program that

*only* works on Windows would steer business to Redmond. A program that's available on multiple platforms lets the user choose the OS if they can, yet still use OSS applications if they can't.

A Windows user who prefers Firefox to IE, may choose gEDA over Eagle as a "first try" because of a bias towards OSS from Firefox. Eventually, they may choose to replace Windows with, say, Linux, assuming it's going to be better for them like all the other OSS apps they use. Etc.

Of course, this assumes that (1) they *can* use the applications on their OS, and (2) they find them to be as usable (for less cost), or better than proprietary alternatives.

So, I have no personal opposition to Windows (heck, I wrote DJGPP, remember?), I just don't happen to use it. If I were to develop any Windows apps these days, I'd have to install a MinGW cross compiler and Wine.

I'd used DOS PCAD back in '88-'91 and came back to it in '98. By then it was Windows based ACCEL-PCAD V13, and Tango was mixed in there sometime. Was it Tango that was orphaned, or did Tango become PCAD?

Anyway now its V18, Altium PCAD 2004. The 2004 version (which really didn't come out 'til early

2005) has been pretty well maintained and they say there will be a 2006 release. It continues to suffer from rumors that it will be merged into Protel (Altium Designer DXP, whatever).