OK. You're right. FINE. Autorouting sucks (but KiCad is pretty nice)

What does KiCad have in the way of component libraries? My personal take on the value of most any free/low cost tool is that it is only as good as what you can do with it without having to spend a lot of time extending the vocabulary...

I just spent a couple of weeks with kicad too. It is definitely "coming along" and has some very powerful features (3D and the new manual router in particular). I am now following development and it is likely that I will switch to it in a year or two. It seems to have a bright future with ongoing major contributions from CERN as well as a busy developer community.

There were just a few too many usability issues for me at present with respect to (re)creating all my library parts. But it is the clear leader in open source PCB cad and everything seems to be moving in the right direction with it.

The one usually suggested is freerouter.org. It is very powerful as autorouters go, and can be used for intelligent manual routing too (push and shove etc). It does have a problem with scaling, getting very slow with more complicated boards. The original developer has stopped working on it due to legal threats from his former employer (Zuken). But it is open source and downloadable.

The other one is Electra, which is about as good but does not have the scaling issue AFAICT. This is a commercial product in the $500 to $3000 range depending on layer count.

The new CERN manual routing engine in KiCad is a great feature here, makes autorouting much less attractive by comparison.

Den torsdag den 12. februar 2015 kl. 20.38.40 UTC+1 skrev Tim Wescott:

haven't tried kicad in a long time, but I remember they used to use an online router, it's no longer online but you can donwload the source and run it

-Lasse

There's a fair collection - some native (typically supplied with the KiCad distribution), quite a lot ported over from Eagle or other libraries, and some contributed (the latter two being "you download them and install them")

It's not as extensive or consistent a collection as you'll find with a $$$ professional package, but neither is it chopped liver.

It's helped along by the fact that the libraries (and board files) are represented in an ASCII format which isn't terribly difficult to edit or generate. With a bit of scripting in whatever language you prefer you can spit out all sorts of parts variants into a new library.

Interesting, Thanks. (I'm still using Eagle 4.xx) Thinking out loud is there some reason we can't share component libraries between software? The (mechanical) CAD programs are able to talk to each other.

George H.

when I tried kicad long time ago there was scripts to convert eagle schematics, boards and libraries, seemed to work ok

-Lasse

I found this aspect frustrating, it is what stopped me switching to it recently. There are a lot of libraries, with some great features like 3D models. But I still ended up needing to make a lot of parts myself, and the facilities provided I found frustrating to use. Simply drawing an

By the way it can import Eagle libraries I believe.

Autoplace sucks-squared.

The reason of course is that all the commercial companies want to protect their investment so it is not in their interest to be able to export to a standard format.

There are some "international standards" I think for this but these do not seem to be widely supported.

KiCad does import Eagle libraries although I have not tried it myself.

John

That is not uncommon, you'll often have to move silkscreen around anyway and most PCB houses will mask the silkscreen with the soldermask so you don't get silkscreen on pads

yes and also whole schematics and boards, at least it could when I tried it

-Lasse

but it is in their interest to be able to import other formats so it will be easy to move their program

-Lasse

Expensive? Considering your board complexity you should be able to get by with the standard version. This has six layers max and 160mm by 100mm routing area. If you buy the whole chebang with layout and autorouter it'll cost $820.

You can do all those things with Eagle, some of them like the grouped BOM even with V4. I always did Eagle BOMs that way. V7 has the extra attribute fields in parts and (finally!) a hierarchical sheet structure. The latter was the reason I bought V7. For the same reasons I had been sitting out V5 and V6.

It isn't. However, while the autorouter runs you can go mountain biking.

On very dense boards most autorouters stall out a lot.

Ok, I am one of the guys who said it and still stands by that opinion. But ... ... when buying V7 I also bought the autorouter. The reason is a medical project where I likely have to whip out some intermediate boards in quick succession so the software engineer has something to sink his teeth into. A client of mine uses the V6 autorouter all the time and it does a decent job for simle stuff. I prescribe the RF layout in great detail, he followed that meticulously by hand and afterwards for the mundane stuff he just hits the button. Not pretty but works and time is always of the essence.

Supposedly the autorouter is different and better in V7 but I haven't used it yet. I'll likely only ever need it for this one project.

afaict on the free version the 100x80mm limit is only for components and such, you can put copper and route outside

-Lasse

I can't use any free version because I use it for profit, same for Tim. Normally routing makes only sense if you also can place parts there. Or very few where leaving a chunk of bare trace suffices. But for folks who are into rigid antenna designs being able to route beyond the limit can be very useful. As long as the Eagle CAM processor will properly generate the Gerbers.

my point was that the pro version might also allow copper and routing outside the limits

could be useful if you need to fit in a standard box

we use some nice extruded boxes that takes a 160x160m board, and it needs a bit of empty space on two edges to slide in anyway

something like this that takes 100x100

if the free version will make gerbers for a 100x100 that would be nice

-Lasse

I can't get my kid to build components for me without a whip and a chair. Unless I can ask him "could you please generate component files for a bunch of these connectors, from three pins to thirty?", in which case he'll happily lose himself in Perl for a day, and cough up a bunch of components for me.

Wow. I didn't know that. I'm going to have to try it.

I've never had a tool I didn't have to tweak, and I haven't found building components to be that big of a pain.

A guy sent me a board design done in KiCad, so I downloaded it and tried it out. I did manufacture his board from KiCad, and everything went OK. He did make a prototype himself, and the PCB fabricator screwed up the copper pour. Old versions of CAM350 would show one pad as the board was (mis)made, with the pad bloomed out to short to the surrounding ground pour. It was never really figured out what happened, but that Gerber file would print differently on different versions of CAM350. NOT a good thing to have happen! Not clear this was in any way related to KiCad, but just something that had us worried.

I normally use Protel 99 SE, which works QUITE well. It has a number of bugs, but if you know them, you can just avoid doing the things that set them off. Nowadays, I run Ubuntu 12.04 on my main desktop, and run Protel under WinXP using VirtualBox. VirtualBox now actually works better than VMware! (And, of course, even if VirtualBox had no support, they couldn't be worse than VMware's support, which actually proves you can actually provide negative support! Their abominable support drove me away from their product.)

Well, back to the point, I think KiCad has a WAYS to go to reach the level of ease and functionality of Protel 99. One thing is that I know the DRC on Protel is totally flawless. I've done several hundred boards, and never seen a case where the Gerbers had a problem, if the schematic was correct. (Hmm, OK, there is ONE bug in the netlister, if you accidentally have a schematic component with the same pin number appearing twice, it SILENTLY screws up the netlist. Never actually got bit by this one, but a couple times have come close.) Autorouting on Protel is so laughably bad I have never used it on anything serious. Every once in a while I try it on a really SIMPLE board, and it is a total disaster! it tries to route all the traces around the outside of the component area until it gets stuck, and they stalls. Also, if you command it to do horizontal on the top layer and vertical on the bottom, it totally ignores you. So, I can't imagine anyone could actually make a worse autorouter, anyway. Have not tried KiCad's, however.

But, Protel is so EASY to use, I feel more comfortable with 2-key shortcuts and a bunch of programmed key bindings (some customized) for the things I do the most often. KiCad seems to be stuck in an ALL graphical icon mode, and I can never remember what TINY little icon is the one I am hunting for.

The biggest thing, however, is Protel's global editing features. This stuff REALLY WORKS, and saves enormous time. It is hard to explain to somebody that has never used Protel, but all through the whole suite of EDA, it has global editing modes where you can set match conditions, and then every entity of the right type will be compared to that condition and altered by what you are changing. You can set many match conditions at one time, and only the things that match ALL the conditions will be changed. if you forgot something and it changed some item you didn't intend, just click undo!

So, I'm fairly impressed by KiCad, it really DOES show promise, but I'm certainly not ready to switch, yet. I DO hope they continue to develop it, and will keep monitoring it. Eventually, I may not be able to get WinXP to run anymore, and Protel may not work on Win7 and above.

Jon