Kicad 6: new feature review for open source EDA tool (opens in new tab)

We use kicad for the design of a 28 layer PCB that includes a number of high speed (>10Gbps) impedance controlled lanes, including 25Gbps signal lines, and implementing three DDR4 banks, with a varying number of discrete chips, and breaking out from a large pin BGA, using a number of HDI features (blind/buried vias, etc).

We've been closely tracking the releases, and, at this point, our layout tech prefers using Kicad over another (proprietary) tool for the majority of layouts - so much so that's she's pushing for us to port all our designs to it.

I am amateur EE. My boards aren't very complicated but with exception of RF stuff (which I just don't understand) contain most of the interesting stuff: typically one or more STM32s below 200MHz, some flash, sensors, external interfaces like USB, display, some high power (>20A) stuff, some very sensitive analog stuff, etc. I have even recently started including my own SMPS (for lower power things).

Kicad is all I need and probably will ever need.

I get that better tools could help do some stuff but, realistically, great majority of work is outside of kicad (like learning, searching for parts, debugging, etc.) Even within kicad I spent most of the time thinking and tinkering with the schematics. So, according to Amdahl's law, there is very little I can gain upgrading Kicad to something else assuming it actually could make me more productive.

If you are pro and you can do that other stuff quickly and efficiently and EDA is majority of your work then, maybe the calculation is different. But I just can't imagine an amateur could benefit a lot.

> KiCAD is very good for an open-source tool that can produce basic PCBs. However, modern paid CAD tools are on a different level entirely.

Not all of them. I personally chose Kicad 5 over Eagle because Eagle's UX was a pain to use. I just never found any advantages over Kicad despite a few clunky parts of the Kicad 5 the overall experience in Kicad was just more productive. Of course, most of my designs have been sub-10MHz with a few 25-50 MHz Ethernet parts.

Statements like this just leave me wondering what useful features the high end software provides that are really missing? Do they automatically re-tune differential pairs after you move things around? Many companies still seem to have separate people or software for doing BOM management so it'd seem they're not superior in that aspect either. Just feels like I'm missing something.

> The differences may not be obvious for simple boards with low speed connections, but it’s a world of difference to use one of the high end CAD tools on a complex board with high speed traces.

That covers a significant amount of the board practical PCB's that are needed for a lot of projects. Still I've been told Altium and such are much better for high speed signals, but also been told that Altium _doesn't_ do 2.5D or signal simulation and that you need to goto the next tier of commercial vendors.

> The differences may not be obvious for simple boards with low speed connections, but it’s a world of difference to use one of the high end CAD tools on a complex board with high speed traces.

No, not really. KiCAD is more than adequate for DC to GHz range circuitry. Add to that being able to develop custom tools in Python and we are at a point where there's almost no comparison --in favor of KiCAD.

For context, we were doing GHz-speed boards with PCAD back in the dark ages (20 years ago? Can't remember). As someone else mentioned, you work with your board house for controlled impedance manufacturing. Not that big of a deal. Also, you have to understand the subject or it won't go well, no matter what magic and pixie dust the EDA tool might offer. If the designer does not understand transmission lines, software isn't going to save you.

The same is true for power distribution systems. Just 'cause chips are connected to the various voltages on the board it does not mean anything is going to work. PDS design is a subject in and of itself. You can't just throw a bunch of 0.1 µF capacitors at the problem and expect things to work. Much as is the case with transmission lines, when it comes to power distribution for high speed designs, you have to know what you are doing.

We are very seriously considering migrating away from Altium Designer to KiCAD. I spent a good portion of the last couple of weeks taking a look at this in detail.

I am sick and tired of paying thousands of dollars per year to maintain our licenses only to see them pile-on the bugs. Altium has been focusing so much on trying to find a buyer for the company that they have actually done damage to the product and the user base.

For example, nobody I know has any interest in anything involving the cloud when it comes to EDA. Nobody. It is interesting to note that, if you work in an ITAR environment, today, given software realities, KiCAD is far more secure than Altium. I cannot, with a straight face, approach our clients and tell them that Altium Designer isn't reaching outside our network without air-gapping the system. With KiCAD, I can issue such a statement with absolute certainty.

Here's another one: Nobody who knows what they are doing would ever use schematic symbols and PCB footprints from a library. I don't care where they came from. This means libraries have nearly zero value (or negative value) for serious design work. You have to make all of your own symbols and PCB footprints in-house and qualify them for your designs.

Here's an interesting KiCAD advantage: I can actually ship the entire EDA tool with the design. Clients can archive all of it, software and design. That is powerful.

If we make the transition (very likely) we will our annual Altium maintenance fees to the KiCAD organization. For me this isn't about getting free software at all. I prefer to pay for software because it generally means it will be well supported. However, that changes rapidly once it become obvious that the company isn't using the money to actually work on what matters most. By supporting KiCAD financially I would help ensure it keeps moving in the right direction.

Oh, yeah, and code. We can definitely help with plugins and maybe even some main application development or bug-stomping. It would be fun.

Autodesk has a long history of buying up competitors that make cheaper products, going back to the MS-DOS days.

>You’re seeing too much conspiracies

Nah fam, I'm seeing my favorite tools like Netfabb disappear and become subscription-only parts of the Autodesk ecosystem.

CAD is not just EDA, or the niche that KiCAD occupies.

I think kicad 5 was meant to fix the library mess while kicad 6 is supposed to focus on UX.

It was sold to me as "what blender did last time", which is music to my ears.

> it’s very hard to displace the incumbents

I don't know how this fits into the narrative, but Protel, which started as an affordable, very good option continued to improve and is now Altium.

Nah, GPL is still shitty, it's enforcement is a joke and I have numerous examples of large companies abusing it even now. And it doesn't stop the developer buyout and forks scenario, just makes the fork inconvenient.

We poisoned pilled KiCad even further by getting it trademarked under the Linux Foundation. That's a whole extra layer of protection.

GPL is not a magic spell, the other open source alternatives didn’t disappear, they were rendered obsolete.

When I was writing a column for JavaWorld I had an editor who was great, she would mark up my column with things like "You can strike this whole paragraph, we get it already." She was really awesome to work with.

That guy does some amazing work. One thing Horizon needs IMHO is a larger development team to ensure it continues on its path.

But it uses json for parts. So scraping and just writing it to your library is a real option, especially when you fall back on existing parts. It can also import parts from kicad, although I didn't try that.

Yes, initial placement is a pain. Especially on a brand new layouts!

However, I've been contemplating the idea of creating a smarter auto-placement for new boards where it'd take the relative positions on the schematic into account. As in taking the relative distances of parts on the schematic, probably using a weighted heuristic / k-means to determine part clusters.

That'd be about 80% of what I'd want from a full auto-router. The new S-Expression eschema format should make it almost trivial to do as an external program. I tried on the old format and it was way to difficult to parse. Though if the Python API's really have been updated then it could be made into a plugin.

1. Learn to carefully structure schematics as hierarchical sheets. 2. Use the replicate-layout plugin to duplicate schematic sheet layout across multiple instances. 3. Get a graphics tablet.

The expensive as hell "default" alternative, Altium Designer, has autoplacer to comlement the autorouter, and it usually works really nicely.

Are you aware of what TopoR can do? And yes, for 1/2 layer routing it seems to work fine when fed from Kicad.

For relatively insensitive signals, or when you can constrain the layout with rough guidance to something that should easily pass signal integrity, this technology makes it so you can go get lunch while the computer gets a decent layout that you'd typically check before sending it off to the prototype fab or get started on quickly running one in-house (for that <2h turn-around from layout to power-on, including printing and soldering).

They are useful if you really know what you're doing.

Someone who tends to only use the autorouter in a bad way probably wouldn't do much better doing it by hand anyway as they wouldn't be aware of the pitfalls.

They'd still end up with 500 vias, traces all over the place, bad placement and not doing any pinswapping anyway

For me, the most useful thing was learning the hotkeys. I struggled adding vias for longer than I'd care to admit.

Nice, I didn't realize. All the tutorials and stuff out there all show a hierarchical schematic as the only way to get multi-sheet, instead of just a flat schematic with some stuff on one page and some stuff on another. I know the UI has changed a lot over the years.

It's less useful for individual components - I can already represent those with pads without affecting readability of the schematic too much. But I sometimes would like to carve out a chunk of stuff onto a separate board, and connect it with, say, a ribbon cable, yet keep it on the same schematic sheet, or at least within the same project. That is not doable in KiCAD 5 at least.

Maybe with the new s-expressions it will be more feasible for someone to create an importer for Renode. Assuming enough annotations or whatever.

Electronic Design Automation