So, What's the future standard CAD tool for open source H/W

[Tabs]

Hi all,

I started a H/W project a few months ago and was trying to select a CAD tool to implement the PCB in.

I considered making it open source but never made a decision (still haven't) but didn't want my tool selection to hinder open sourcifying the project in the future.

I use Altium at work but it was too expensive. So my choices were:

kicad, eagle, design spark and circuit maker.

I rejected eagle (even though it seems to be the current standard for open source) because I feel the writing is on the wall for eagle. Having seen Daves CM review I believe this even more.
Between Farnell jumping in bed with Altium wrt circuit studio and Altium releasing circuit maker I don't see future funding for eagle to develop.
Infact, now that circuit studio pricing is available, I might have to eat my words. I agree with Dave, the pricing is way off the mark and eagle users have no reason to jump to circuit studio.
They do, however, have every reason to jump to circuit maker. Especially if they used eagle for open source h/w. In any case, eagle's still stuck between a rock and a hard place (business/funding wise)

Since circuit maker wasn't in open beta at the time i started my project, I made the decision to use kicad. I didn't choose circuit studio because - why waste my money, when I can waste my time with KiCAD.

Saying KiCad does everything a PCB designer needs is a bit like saying "some transparent overhead projector film and a black sharpie does everything a PCB designer needs"
yeah, sure you could design a 10 layer board with DDR3, PCIe and a truck load of other high speed interfaces, but would you? You'd need a stupid amount of time and infinite patience.
A bit like KiCAD then isn't it? 

The worst part of learning KiCAD was trying to figure out why my keyboard at work wasn't working the way it usually does with Altium.
Its only then, you realise your finely honed muscle memory of Altium shortcuts has been replaced by KiCAD shortcuts. 

Eventually KiCAD grew on me and I can now separate it from Altium at work. I actually like it.

After watching the review I found myself considering implementing the next two boards of my project in CM and moving the first from KiCAD to CM.

I decided to stick with KiCAD.

For me, CM in the cloud (Altiums cloud) is a bit risky. I'm not comfortable with the idea of Altium being able to lock me out if everything goes the way of the dodo.
Also, since my project is split over multiple PCBs (each PCB being considered as its own project in Altium) I would be over the CM limit of two private projects (assuming it works like Altium)
I know, I know, I haven't gone open source but I may in the future.

I think I secretly like the idea behind open source h/w. A rather romantic notion. Brings a tear to the eye as Dave says.

So for me, I'm hedging my bets towards KiCAD.
I use it with the daily updates automatically applied. Aside from frequent git library issues (now solved with a setting on my side), everything seems to run fine.

So what's your take?

Regards
Tabs

[EEVblog]

I've said it before, KiCAD will never become a standard until it has a stable release and looks like a real product that someone has responsibility for. None of this daily build rubbish.
But that's pretty close I believe.
The problem is Altium, are, well, Altium, and they always "almost" get it right whilst shooting themselves in the foot.
I also said this before, Altium (and CircuitMaker) could win the entire OSHW industry within months, bury eagle, bury KiCAD, and any other competitor, all they have to do is get the balance right 

[c4757p]

Altium will never "bury" KiCad, there'll always be some demand for an open solution, even if not much. Lots of commercial products with very good and affordable offerings have still not managed to "bury" their FOSS competitors.

Also - stable release coming soon, keep your pants on... O0

[EEVblog]

It isn't quite clear to me whether you don't think KiCAD is a "real" product just because it doesn't have a stable release that someone will support, or because it isn't fully baked yet.

I'm not a linux software geek, so any program that I have to download a "daily" build of gives me the heebie-jeebies. It gave me the hibbie-jeebies when I worked at Altium and it was company policy to use the daily builds. Heck, the hardware team I was in had to settle on and use a "known build" on the side we could trust just to get some real work done.
Last time I looked at KiCAD it wasn't easy to even download a compiled installable version, it was just on some guys personal website.
Sorry, but if you want a program to become an defacto industry standard, that just doesn't cut it, at least not for me.
I know this has changed a bit recently and is changing very quickly.

Dave. I think I'd be more interested in seeing you do a small project with KiCAD than with CM. But it would be good to see you do one with either (or both for comparison). Something needs to happen to drive change in the OSHW community. You are in a unique position to influence change considering your PCB design expertise and your established audience.

Yes, and I won't be committing to KiCAD until such time as it looks and feels like it's a real product that has a stable build that everyone gets behind, and it looks like it's a project that's being controlled. I believe this is close.
Note that this could in fact be just all about appearances, having a central official websites with a central official build that almost every uses. Oh, you want the source and nightly builds, ok, they are hidden over here.

Take a look at the website currently:
http://www.kicad-pcb.org/

Where do you download it? Links to a directory on some guys personal website, that's bullshit, no thanks.
They need a big fat download button on the front page to the Windows install executable. A huge community forum for support, and a stable version released maybe every 3 months or something.
At least give the appearance that this thing is a real product Joe Average should trust.

Several things need to happen to generate a critical mass around which a thriving community can coalesce. You can't make a stable build  appear but maybe you can sow the seeds of a need for one with your stamp of authority. It is time to generate some ripples in the cosmos.

I don't have the time nor the enthusiasm to help drive KiCAD to become an industry standard, sorry. Call me when it's ready.

There was a recent thread on the forum about a continuity buzzer/tester. Perhaps it would be a suitable project for an OSHW design. Your uSupply is another natural but it is more prone to delay because you have historical baggage attached within your personal investment in it.

The uSupply is currently being done in Circuit Maker. David2 just started the project last night.

[EEVblog]

Altium will never "bury" KiCad, there'll always be some demand for an open solution, even if not much. Lots of commercial products with very good and affordable offerings have still not managed to "bury" their FOSS competitors.

Of course, but the current defacto standard is Eagle, by many order of magnitude over KiCAD.
If Altium made Eagle redundant, well, KiCAD remains were it is.

[c4757p]

Take a look at the website currently:
http://www.kicad-pcb.org/

Where do you download it? Links to a directory on some guys personal website, that's bullshit, no thanks.

Bullshit? Why? The old stable release links to the university where the creator of KiCad works - that's not good enough for you? - and the new release is still upcoming.

They need a big fat download button on the front page to the Windows install executable.

Agreed. I'll suggest it tomorrow.

A huge community forum for support, and a stable version released maybe every 3 months or something.

Perhaps we should see about joining up with Chris's forum [https://forum.kicad.info/]. I'd like to see that. Perhaps we can work in a nice fat link to it on the homepage too. Hmm...

[EEVblog]

BTW, Chris write a good article on the essentialness of the Internet for design:
https://contextualelectronics.com/learning/simple-plane-experiment-kicad-vs-circuitmaker-vs-upverter/
Thus making the internet-tied nature of Circuit Maker (and others) a little bit less of an issue (but still it's major one).

[EEVblog]

Bullshit? Why? The old stable release links to the university where the creator of KiCad works - that's not good enough for you? - and the new release is still upcoming.

Oh come on, really?
It's a very bad look, it does not instil a lot of confidence. I as Joe Average have no idea that guy is the creator, nor where he works etc, all I get is a link to some unknown dodgy sounding file server:
http://iut-tice.ujf-grenoble.fr/cao/
If oyu want a program to be taken seriously by Joe Average then it's got to be better than this.

[LukeW]

The amount of pre-existing time that you've invested in one of these software packages makes a big difference.

If you've spent many, many hours learning how to use KiCad well, or Eagle or Altium or whatever, then the value of that experience and the time that would need to be spent re-learning something different makes a big difference. If the software you're already experienced with is suitable for what you want to do, why change?

If you're paying for person-hours at any sensible rate for an EE in a professional or semi-professional context then suddenly paying $1000 for a software package does not seem so expensive compared to the time spent learning to use it to do the job efficiently, and that learning curve applies to any one of these tools.

[codeboy2k]

Bullshit? Why? The old stable release links to the university where the creator of KiCad works - that's not good enough for you? - and the new release is still upcoming.

Oh come on, really?
It's a very bad look, it does not instil a lot of confidence. I as Joe Average have no idea that guy is the creator, nor where he works etc, all I get is a link to some unknown dodgy sounding file server:
http://iut-tice.ujf-grenoble.fr/cao/
If oyu want a program to be taken seriously by Joe Average then it's got to be better than this.

I agree with Dave's point of view on this.

All they need to do is put the big fat download button for each platform.  Behind the scenes it can go to the creator's personal directory on his University's server, no one will care. 

I like Kicad, and once you get used to it, it's pretty decent for free.  It's got some weird-ass behaviours that don't always make sense, and it really needs some polish.  I'd like to donate some of my programming time to it, if I didn't think the devs were such assholes.  As it turns out, I'll probably just do the fixes/improvements for my own use, if it ever comes to that.

CircuitMaker will take some market away, but there will always be those who never want to use a product that is not open sourced.

[c4757p]

I'd like to donate some of my programming time to it, if I didn't think the devs were such assholes.  As it turns out, I'll probably just do the fixes/improvements for my own use, if it ever comes to that.

I find the current "leadership" easy enough to get along with. No comment on whom I think you might be referring to.

[HackedFridgeMagnet]

Bullshit? Why? The old stable release links to the university where the creator of KiCad works - that's not good enough for you? - and the new release is still upcoming.

Oh come on, really?
It's a very bad look, it does not instil a lot of confidence. I as Joe Average have no idea that guy is the creator, nor where he works etc, all I get is a link to some unknown dodgy sounding file server:
http://iut-tice.ujf-grenoble.fr/cao/
If oyu want a program to be taken seriously by Joe Average then it's got to be better than this.

Lol I'll translate the host name:
 University of Joseph Fourier, Grenoble, France.

i think the subdomain name iut-tice may be a campus.

I'll admit that the name Fourier may mean nothing to Joe Average but it would mean something to you.

I kind of think it adds a bit of distinction having your builds served from a University bearing the name Fourier.

[EEVblog]

Lol I'll translate the host name:
 University of Joseph Fourier, Grenoble, France.

And how is Joe Average expected to know that? You have to actually go to ujf-grenoble.fr to figure that out first.
Oops, no you can't I just checked http://ujf-grenoble.fr/
nothing!
You have to put www in front to get anything.

"ujf" could mean anything and indeed looks random, as does "iut-tice" and "cao"
http://iut-tice.ujf-grenoble.fr/cao/
Come on, without any knowledge you can't expect people to think that's not a dodgy looking address.

It's a just a joke for a product they are trying to get people to take seriously.
Damn, every time I get an idea, I get a domain name for it and host the files on it, and at the very least make it try and look serious. Yet they haven't done some simple stuff like this after how many years?
It's not rocket science - stable build everyone uses, hide the damn source code (99.999% of otential users don't care), install executables on the main page with a big download button, and a user forum on the same site.
Chris Gammell had to go set up his own forum!
This stuff needs to happen at a minimum for this to become the successful defacto tool. It's the small things that matter.

[HackedFridgeMagnet]

Not sure why Joe Average is so important to this thread. Maybe Joe Electronics Hobbyist.

It took me 2 clicks from this thread, not searches, to find that it was the uni website.

a similar example is the url http://www.uts.edu.au/ all acronyms but not that hard to click.

Come on, without any knowledge you can't expect people to think that's not a dodgy looking address.

Well maybe, but it seems like someone is squatting Kicad.org and www.kicad-pcb.org links to the uni anyway. I personally don't see the url as an issue.

I reckon Kudos to the developers (including c4757p) for their great software, soon may you take over the world. 

[EEVblog]

I reckon Kudos to the developers (including c4757p) for their great software, soon may you take over the world. 

Of course. I'm just trying to help them make the product a bit more polished.
If I look at it and see these little things that don't instil confidence in me, I'm sure there are others out there too.

[Tabs]

I installed KiCAD by adding a download source to some that runs an automated build script which compiles it for Ubuntu which is the linux distribution I use.
The process wasn't difficult and now the daily builds come via Ubuntus version of windows update. I can choose not to install if I wish.

That being said, what's the definition of stable?
I install the daily builds and I've never seen the effects of whatever was updated. Apart from the libraries (which was my fault for not changing to local copies).
None of the updates has broken my KiCAD installation and I've always been able to work on the one project I have in KiCAD.
I think that meets many peoples definition of stable. Including my own.

The problem with KiCAD (in my opinion) is its disjointed nature. Not sure if its a result of KiCAD being made up of by a rag-tag fleet of disparate programs and programmers.
With Altium everything is very cohesive. There's a pattern or similarity in driving the schematic capture and layout and there's a number of options when it comes to libraries.
There's nothing wrong with the workflow sch->footprint assignment -> pcb. Its just different to Altium where you assign the footprint upfront.

It has its advantages and disadvantages.

I'm told KiCAD has come along massively since CERN got involved (even if it is just 1 or 2 CERN programmers).
It just needs a better and more modern GUI and better support for library creation, management.

All of these things are on the roadmap and will eventually get fixed. I hear eagle is rarely updated so in this respect you would assume KiCAD would eventually surpass eagle.
Eagle has its market share because it was the first viable tool that supported open source development by having a free version. I'm not sure that new people should even consider
eagle now that CM is here. For private projects I think KiCAD still beats eagle (i've never tried eagle so my opinion doesn't really count - i'm just being lazy).

I think the open source community is going to drift towards a split between KiCAD vs Altiums offerings.
KiCAD needs a slick interface, Altium needs to re-assess limitations on CM or pricing of CS.

I think you will always get the die-hards who will say its not open if the tool is closed. There's no arguing with that mindset. Its probably why eagle never conquered.
You also get places like CERN who (correct me if im wrong) must make everything they do public domain because they are publicly funded (by governments).

For my professional (work) stuff - I stick with Altium
For my professional/hobby (home) stuff - I'll stick with KiCAD. it can only get better.

[EEVblog]

That being said, what's the definition of stable?

A version that is formally released that almost everyone uses.
It doesn't actually matter how "stable" it actually is, that word shouldn't even be used actually.

Why is this important? It's because when people jump on the forum with support questions then you don't have to ask what "build" they are using "whether they compiled it themselves or did something else weird etc. If you release say one major version every 3 months then support becomes much easier for everyone.
Until KiCAD does this, it will never catch onto mainstream IMO.
They have to get totally out of this mindset that it's an open source tool for tech people who care about the source, builds, and know about GIT and talk penguin etc.

[EEVblog]

Eagle has its market share because it was the first viable tool that supported open source development by having a free version. I'm not sure that new people should even consider eagle now that CM is here.

Eagle will remain popular whilst a few major OSHW players support it. The likes of Arduino, Sparkfun, Adafruit etc.
If Altium were smart they would approach a handful of these companies and give them a hand to switch. But Altium isn't that smart.

[EEVblog]

So the biggest problem with Kicad seems to be the URL of the download server, something that can be trivially fixed.

No. I said it's a small issue that does not instill confidence in the product.

The biggest problem with Eagle seems to be that people are spreading FUD about it being abandoned, although personally I think it's the stupid board size limits in the free/cheap versions.

No. The biggest problems are the stupid board size limits as you suggest, and the ancient UI.

The biggest problem with CM is that it's tied to Altium's cloud that no-one likes or trusts and which historically hasn't worked well for other software.

Correct.

Also, I find it amusing that people are talking about Kicad not being able to become an "industry standard", as if they care

I care. I would love to see KiCAD become a defacto industry standard for OSHW.
But being a 25+ year Altium user, I'd also like to see CircuitMaker succeed.

, and as if CM could ever become popular while it requires the use of their stupid cloud infrastructure.

If Arduino, Sparkfun and Adafruit started using it, you watch CM become the defacto industry standard within 6 months.

Eagle, DIPtrace, DEX, Kicad... None of them as as powerful as CM perhaps, but for open source stuff freedom and confidence are far more important.

You'll likely get little argument here.

[tggzzz]

I've said it before, KiCAD will never become a standard until it has a stable release and looks like a real product that someone has responsibility for. None of this daily build rubbish.

This isn't a vote for KiCAD, but you might care to realise that there is a very widespread movement towards incremental releases. Some example phrases are "release early, release often", and "devops". Naturally it can (and will) be taken too far by some people

I'll note that I've never my linux system borked by any of its daily releases. My other system, which is a "real product that someone (theoretically) has responsibility for" has been borked - and it seems that some people get crippled every couple of months! Conclusion: having a single person with responsibility is neither necessary nor sufficient.

As for anything that relies on a corporate cloud - never. Just look at Microsoft's "Plays For Sure(TM)" music purchase system. Now they've turned it off it is "Silence For Sure".

[Thor-Arne]

I'm sure the CERN KiCad development roadmap has been posted before.

Several important improvements is listed here, and for what I can see from the daily builds the future is promising.

But I agree that the web-page needs more improvements, and "stable" releases should be releases at more or less frequent intervals, say every 3-6 months.

I've had very little trouble using KiCad, and it didn't take too much effort to learn the UI.

[ludzinc]

What I don't understand is why CERN is investing in KiCAD.

Why not just pay for a known working commercial product, and redirect the software engineers wages back to physicists?

[c4757p]

Because open.

http://press.web.cern.ch/press-releases/2011/07/cern-launches-open-hardware-initiative

[sync]

What I don't understand is why CERN is investing in KiCAD.

Why not just pay for a known working commercial product, and redirect the software engineers wages back to physicists?

Maybe for the same reasons CERN developed the world wide web instead of just using a commercial product.

[Tabs]

That being said, what's the definition of stable?

A version that is formally released that almost everyone uses.
It doesn't actually matter how "stable" it actually is, that word shouldn't even be used actually.

Why is this important? It's because when people jump on the forum with support questions then you don't have to ask what "build" they are using "whether they compiled it themselves or did something else weird etc. If you release say one major version every 3 months then support becomes much easier for everyone.
Until KiCAD does this, it will never catch onto mainstream IMO.
They have to get totally out of this mindset that it's an open source tool for tech people who care about the source, builds, and know about GIT and talk penguin etc.

This is an interesting point. Whilst on the surface its very valid. A known baseline for comparison is always the best place to start when fault finding or asking for help. You just log on to the forum, tell them the problem, tell them the version and if its a known issue your more likely to get a solution. The same argument applies to daily builds. The only difference is if you assume the majority of the installed user base is on the official release version then you increase the chances of someone having faced the same problem. If your'e on the daily builds then at most you have a 1 day window where others on daily builds can see if the they can replicate the problem. If the problem is still there the next day or more then you can assume its a major issue, in which case it will be sorted faster. This bugfix cycle wont change if you assume the opposite from my starting position. ie if the installed base was majority daily build users.

The fact is, if you have a problem which is in the software (as opposed to a duff setting you made) then your'e only going to find the fix in a later version.
This is why, when your'e on the formal version, the first response you get to a problem is "Have you installed the latest update?".
Whenever I have an issue with Altium, this is the first thing that gets said. I'm always advised to install the update.
If your problem is a setting somewhere or a "How do you do xyz" then you can resolve that regardless of version (I would assume that the S/W doesn't play musical chairs with setting locations or how to work a feature).

On Altium, I fear the dreaded "Access Violation" error that I always seem to get sooner or later on a new release. Thats why I have v14 for official work & v15 to do the parts of my workflow that v14 cant do properly eg Variant BOM generation using database parameters.

Point is all S/W (even pro level commercial) have their issues. Many of them have moved towards the daily release system for updates because it improves the cycle time for bug fixes and reduces the scope to introduce new bugs to only the bits that was changed. Thats a smaller amount of code delta than between major releases. I have an Altium subscription and i seem to be averaging 2-3 days between notifications of an update on v14 or v15. That may as well be daily builds. As always, I can choose to update or not.

Since the users of KiCAD and Altium are free to choose their own update frequency, I don't think the daily builds should be criticised as harshly as they seem to be.

Other issues like the website are insignificant. I agree with Dave that it could be better, but its not a deal breaker.

The more fundamental issues should be considered.
CM makes you dependent on a commercial entity, with its own goals and interests. it also has silly restrictions that seem to have been self enforced on Altium by Altium. Its going to be hard for them to loosen the strings on CM without takings the legs out from under circuit studio (which may violate commercial agreements with element14) or Altium Designer itself. 

KiCAD wont ever have these restrictions or these commercial issues. It will only ever have technical issues like GUI design and library management. These will be solved (eventually ... I hope).

The only thing that makes the difference then (as in the future) is marketing and market adoption.
Will the the Altium marketing hype be able to convince the open source community enough for them to give up the freedom (as in liberty) that true open source tools allow their users?