Hacker News new | past | comments | ask | show | jobs | submit login
LibrePCB 0.1.3 released, now with DRC (librepcb.org)
138 points by dbrgn 4 days ago | hide | past | web | favorite | 33 comments

Before people start asking:

Comparison with KiCAD and Eagle: https://librepcb.org/compare/

Talk by Urban that explains why he started LibrePCB, and how the library concepts are different from other similar tools like KiCAD: https://www.youtube.com/watch?v=vu-h5y6tK34 (Slides: https://archive.fosdem.org/2018/schedule/event/cad_librepcb/...)

If you want to try it: https://docs.librepcb.org/

Previous HN discussion: https://news.ycombinator.com/item?id=18569969

I was at FOSDEM 2018, when LibrePCB and Horizon EDA were presented, and they both looked very interesting. I will try LibrePCB for some small boards I want to make, the library concept sounds like something we sorely need.

There will be another LibrePCB talk at FOSDEM next year!

I started using librepcb for some smaller projects last year after having some issues with kicad. My thoughts after manufacturing ~10 boards:

* It's simple, but what is there works reliably.

* Library management works really well.

* There are some issues with grids and alignment

* DRC was available in the ui but could not find any errors (have not tested 0.1.3 yet).

* No auto place/route. Horrible default placement.

* Minimal manufacturing support but what is there works

I moved from kicad when my old designs started looking different/broken due to updated components (!!). How can kicad team allow this?

Thanks for your feedback! Great to hear from people that are using LibrePCB :) I agree with most of these things.

Note: I think there was no DRC available in the board editor before 0.1.3, only an ERC which reported things like unplaced components. In 0.1.3 you have an actual DRC that checks things like copper clearance, minimum drill diameter, etc.

PS: If you have a nice project to share, feel free to post it at https://librepcb.discourse.group/t/projects-madewithlibrepcb... !

You are right. But IIRC unplaced components was the only thing the ERC could detect :(

See also: HorizonEDA [1], which I've used for a handful of projects and are really happy with - mostly due to how much its library/parts UX is better than anything I've every used before.

[1] - https://github.com/horizon-eda/horizon

I agree. Used it for multiple projects already. As one of the main contributers to the parts library I especially like Horizon's parts pool concept for these reasons:

- the modularity (+inheritance) allows you to reuse a lot of existing stuff and change things in one place while affecting all

- the simple json format allows you to create parts/packages/etc from scripts very fast

- the package footprint editor is great and has generation scripts for a ton of existing packages (SOIC, QFN, etc.)

I heard it was planned to release a 1.0 on the next FOSDEM in january.

I discovered Horizon just now from the LibrePCB site and will certainly give it a try. The part management looks so much more sensible than anything else.

It's fantastic that LibrePCB exists to solve the key pain points around KiCad, but I wonder what the end state is. Network effects come into play pretty strongly here with libraries, reference designs, and community knowledge. It took years of improvements and Eagle getting heavily monetized by Autodesk for KiCad to become the de facto EDA tool for hobbyists and open source hardware developers. This space isn't that large. Is there a scenario where LibrePCB gets enough critical mass behind it to be competitive with KiCad, or will KiCad gradually improve and continue to grow to prevent that from ever happening?

I'm not so sure that we would want one community around one package for electrical engineering. The industry has always been relatively stagnant, such that even a mediocre UX like Altium's is a breath of fresh air compared to Zuken, Mentor, Cadence, etc. and we could do with a lot more experimentation in open source ECAD to make up for the missed opportunities. That means a lot of orthogonal experimentation in file formats and version control, DRC, standardization of schematic and PCB modules, and so on that should not be coupled to one project.

If there can be multiple graphics editing programs (GIMP, Inkscape, Krita, ...) and multiple 3D modelling tools (FreeCAD, OpenSCAD, Blender, Wings3D, ...) why shouldn't the space be big enough for multiple EDA tools (KiCAD, LibrePCB, Fritzing, Horizon, ...)?

I don't think it's a hugely controversial statement to say that the hobbyist circuit/pcb design scene is a lot smaller then than the number of hobbyist graphic designers. So the question is more can the community actually support two/three of these programs while also offering enough features for both to be usable - or do we just get two/three effectively unusable programs with no library support.

I find it far more likely that ECAD industry will go the way of Blender where the high cost of per seat licensing will drive big companies to start contributing to open source alternatives to stop the bleeding. Given the diversity in needs of different industries, each could contribute the pieces they need into a whole that makes up a viable competitor.

The bigger problem than the number of hobbyists is that the vast majority can't even afford to prototype the high end of what PCB designers can do. For example, a 12-24 layer board with mixed RF and 1ghz+ digital signals (like the hundreds of smartphone main boards connected to a camera that are design each year) can cost thousands or tens of thousands to fab in small quantities, let alone assembly and parts to put it together.

KiCad already has some corporate sponsorship, including CERN. I think this will help keep KiCad ahead of its open source competitors.


Edit: And to your latter point (and it is a reasonable one), the HackRF was done in KiCad. This demonstrates that something at that level of complexity can be done with open source tools and still make enough profit to be open hardware. Promising!!!


That's a very interesting point, I do agree that could happen if enough companies are unhappy with the price, though that's not something I know much about. The price of switching to a new tool would be very high though, I would think, so any viable competitor needs to be very feature complete and probably needs a high degree of compatibility with whatever tools they're currently using.

And yeah, I was considering going into that but didn't want to talk too much out of my knowledge zone. There is definitely a huge divide between the professional PCB designers and what hobbyists (like me) tend to make. My most complicated board is a 2-layer with mostly through-hole and a few surface-mount. BGA, the really small surface mount parts, any anything crazier than maybe 4-layers is just completely out of my reach - I could theoretically design them with a lot of learning, but the expense of getting the design fabricated and the impractically of soldering them manually just makes it a complete non-starter. But yet, those parts are pretty much all that you'll see on modern boards. What I'm doing is so much simpler than what professionals are doing it's really not comparable. I mostly don't need the complexity of something like Eagle, but for companies doing modern PCB design those complex features are an absolute requirement.

You can solder 0402 scale stuff with a pair of tweezers, a $20 USB microscope and a regular soldering iron. 4-layer layout is actually easier than 2-layer, it means less track gymnastics required. This kind of stuff could qualify as 'professional level', a good choice for all kinds of general boards (sure, not smartwatches probably).

For me, the biggest limiting factor is not really skill, but cost. I can theoretically do all of these things, but the cost tends to be prohibitive and the risk too high. 2-layer boards are the best and 4-layer boards are my limit because going any higher just goes way outside the price I can justify putting into hobby projects that may not pan out. 2-layer boards you can sometimes get on sale dirt cheap (ignoring shipping...), and the mark-up when going from 2-layer to 4-layer tends to be at least 2x, and for most places I've looked at more like 4x or higher.

The same thing goes for small surface mount stuff. Ex. I was looking at using a TSSOP package for a project, but I just couldn't justify the cost of ~10$ a chip when I was unsure I could even solder it. Even for my last project, I messed up a few boards because I had trouble soldering the leads for the Micro-USB connector on it.

On the topic of 0402's, I agree I could probably solder them with enough patience - though that goes back to the risk issue, if I can't then I blew a bunch of money on boards I can't put together. For my projects, I thankfully haven't yet run into a situation where I needed to go that scale, so the lowest I've gone are 0805's, which are nice and chunky.

And to be fair, I don't really know what qualifies as "professional level", I'm mostly thinking of things like computer motherboards, phones, smart watches, etc. Tons of things have PCBs in them that don't require that level of detail, and I don't know the percentages of people doing each - perhaps there is a decently sized professional market for a tool that can do smaller boards with less layers/complexity much smoother.

> the biggest limiting factor is not really skill, but cost

Can we put a number on the cost?

Seems to me like if you want a "serious amateur" electronics lab for doing things like prototyping boards with BGA components then you need a budget of around $300 - $1500. Price depends on whether you want cheap-and-cheerful starter kits that will eventually be replaced verses professional stuff that ought to last for decades.

On the one hand that's real money but on the other hand that's also what it costs to buy a new laptop or travel to a conference.

Yes, this is what I was getting at. Somewhat selfishly, I asked this question because one slice of the company I just stated involves developing open source hardware. I want to pick the right EDA tool in terms of capabilities and usability, but especially community adoption in order to reduce friction for people to use and fork our projects.

Also, to other commenters’ points, I expect to use Altium or something “worse” for >6 layer boards and pick a free, open source tool for <=6 layer. This is due to the capabilities of the tools, familiarity from contract manufacturers, and the low likelihood that a hobbyist is going to get high layer count or HDI boards fabbed and stuffed in any case.

As a Horizon contributor (and avid Blender user) I see this in a similar way: I think trying different conceptual approaches is something everyone can profit from.

Every single tool in your list has it’s own pros and cons and some of them are targeting a different audience altogether (e.g. Fritzing).

I liked the ease of use with which LibrePCB allowed users to use existing Libraries, so I proposed a similar thing for Horizon.

Speaking of which, Fritzing (and it's older cousin Arduino) desperately need some competition.

As a tool targeting beginners I feel it deserves a better UI (some stability would not hurt either).

I think Horizon is aiming more at professionals or even small teams. carrotindustries (who started the project) said he was very much inspired by Altium, but wasn’t happy with existing open source EDA tools. So it is unlikely this scope will change unless someone new enters the game.

I try my best in helping to keep the entry barrier low and to smooth rough edges or paper cuts where needed. My goal is to make it not impossible for people who start out with electronics, while still maintaining a tool professionals can work with. Documentation is a big part of that: https://horizon-eda.readthedocs.io/en/latest/feature-overvie...

> Network effects come into play pretty strongly here with libraries

positive only when managed correctly. Kicad includes tons of low quality libraries. I have had non-working pcbs thanks to crappy kicad libraries.

Monoculture is not good. Positive changes happened after LLVM/clang rose up to challenge GCC. LibrePCB cannot hope to be a drop in replacement for kicad like clang was for GCC, but a file converter tool seems plausible.

Is it possible to import/convert KiCAD libraries? Library coverage of the gazillion of available components is something that takes a lot of work and time. Even KiCAD with its head start, and higher user count that comes with it, is still lacking a lot of components. So whatever one thinks of KiCAD's libraries or library management… it would still make a huge difference for the user to have an import function or conversion tools so as to not have to start pretty much from scratch as long as the library coverage is lacking.

Thanks for posting this. I'm planning to start work on my first ever PCB tomorrow and I'll try learning this instead of KiCAD and see if I need to switch or not.

If you are completely new to PCB design and manufacturing I would recommend kicad instead.

I like them both but kicad has excellent tutorials on all aspects of the process. See for example the Conceptual Electronics videos on YouTube.

Great. Now let's work on standardizing a footprint description language for components so we don't have to create a new footprint library for every new tool on the planet. (And maybe even so manufacturers will just release their footprints using that format, so we don't have to struggle there either).

Cute, but EasyEDA is where the hobbyists are at right now. They realised that to win hearts and minds, the winning combination is to have a PCB editor, a components shop, a PCB manufacturing service and a PCB SMD assembly service all integrated (up to a point) in a single product.

You're dismissing a free non-commercial open-source cross platform EDA software project by promoting a web-based closed source commercial EDA tool by saying "that's where hobbyists are" without even showing any evidence for that (and in a dismissive tone). I think you're on the wrong forum.

I've been using EasyEDA for the last two years and have started ordering some of my parts from as well. It's worked pretty well for me, just be sure you verify the pinouts and packages for any user generated parts.

Engineering $/w industry in general is full of proprietary, fleecing, outdated, and $mug, bloatware and crapware.

Any open-source effort is welcome.

If it were up to me, I'd dedicate the rest of my life to developing open-source engineering s/w alternatives, so I could play a small part in thoroughly crushing the commercial land$cape of engineering $$$$ware.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact