eevBLAB #62 - PCB Wars - The Rise Of KiCAD

[EEVblog]

Dave gives two predictions for the future of the open source KiCAD PCB Design package

[rs20]

Great video! Is it just me, or has kicad.com (the digikey thing) gone offline in the ~6 hours since you posted this?

[Magnificent Bastard]

Great video! Is it just me, or has kicad.com (the digikey thing) gone offline in the ~6 hours since you posted this?

Works for me!  However, I use OpenDNS for my DNS servers (I manually put their DNS IP addresses into my router, so all computers on my network use them).  If you are using the DNS servers provided by your ISP, it is possible that this domain is still listed as malicious, and they are not connecting you.  They are also (probably) censoring other websites that they politically disagree with.  Do yourself a favor, and switch to OpenDNS.  You can do it on your router like I did-- so all computers that attach to your router will use the new DNS servers, or you can do it machine by machine.  If you have a laptop that you take outside your house, I would type it in manually in that laptop.

[Magnificent Bastard]

I switched to KiCAD when Eagle became cloud/subscription based.  I absolutely HATE subscription based software.

KiCAD is (shockingly!) very good!  It has copper pour, push and shove, differential traces, trace length measurement and adjustment (by meandering), 3D rendering, etc.-- many of the features that the big commercial tools have.  The libraries are getting better, and they are working on upgrading the user interface-- so, it just keeps getting better and better!

 

[james_s]

I hope kicad never goes to the dark side, it would be a shitty day

How can it? It's open source, if it ever takes an unpopular path it will be forked.

[chickenHeadKnob]

What you (Dave) are predicting is not a remote possibility or off the wall projection. This  is essentially what happened in the embedded C compiler + IDE market from the 90's onward. Multiple for profit offerings based on Gcc and Eclipse.

[exe]

Interesting. I tried using kicad many times and was never able to do anything useful because its ui simply doesn't work for me. Like, I struggled to move/arrange components, a very basic thing to me. In diptrace if I move components, all wires are arranged automatically (not always in an optimal way, but still). So, no broken connections, no weird angles, loops are removed, wires are shortened/extended if needed, etc. Can kicad do the same for me? I have to admin I didn't do much research, but I did spend some time learning how to use it and my conclusion was that user-friendliness was not even on a roadmap.

[FrankBuss]

I don't think paid support would be working, because usually you get good answers on the KiCad forum for free.

[Rerouter]

At present Kicad does not have a drag mode for chips in the circuit, It does for the schematic, with the push and shove router core its something that they likely can add, but cannot say when it may happen.

[jancumps]

I don't think paid support would be working, because usually you get good answers on the KiCad forum for free.

companies will favor licensed and supported software - even if binaries and options are exact the same as the ones under the open source license.

See Apache and Tomcat servers - rejected in their free form in many companies but the engine behind several of the applications they pay good money for.

[Bratster]

Interesting. I tried using kicad many times and was never able to do anything useful because its ui simply doesn't work for me. Like, I struggled to move/arrange components, a very basic thing to me. In diptrace if I move components, all wires are arranged automatically (not always in an optimal way, but still). So, no broken connections, no weird angles, loops are removed, wires are shortened/extended if needed, etc. Can kicad do the same for me? I have to admin I didn't do much research, but I did spend some time learning how to use it and my conclusion was that user-friendliness was not even on a roadmap.

Are you talking about moving parts around in the schematic or moving parts around on the PCB?

in the schematic it's easy afaik, there's two different commands, one that just moves the component around but breaks the connections and another one that will drag it around keeping the connections.

Sent from my Pixel 2 XL using Tapatalk

[profdc9]

As much as I hate to criticize Dave here, I think he missed a much bigger point about Kicad.   I am guessing it is because he worked for Altium and has the perspective of a former Altium employee, seeing an open source product as a viable alternative to Altium.

The question isn't really if Altium's sales with be hurt by Kicad, or if Altium or some other PCB software vendor, or other CAD vendor will embrace Kicad as a low-end market solution. 

The real question is this: what happens when projects and ideas that currently can not be realized because the availability and cost of PCB design software is a major stumbling block can perhaps be realized with because the barriers to designing a product are further lowered?

Of necessity, or perhaps in pursuit of maximum profit, companies like Altium must charge for the software, now with term-limited license agreements, which embodies the expertise of their engineers reduced into Altium code.  This expertise is hard-won and is therefore sold and treated like a scarce resource.

However, as our technology becomes more complex, it requires more complex tools that in turn embody the expertise and therefore labor of those who created the tools.  This adds further barriers to designing competitive products and requires more initial capital outlay, so that fewer are able to realize projects that use more complex technology.

Vendors who sell more commodity items than highly specialized PCB design software often have an incentive to break this impasse.  For example, as Dave showed kicad.com is now registered by Digikey.  If more people can design PCBs, well, this is going to require parts for these PCBs, hence Digikey's incentive for supporting Kicad.

The further you get into the commodity supply chain the stronger this pressure becomes.  Most Chinese vendors sell commodity, almost generically sourced electronics, and as I am not familiar with any major Chinese PCB vendors of PCB CAD software, so it would seem that these Chinese vendors would have the maximum incentive to drive demand of their electronic parts by making design tools widely available.

I think a manifestation of this is the commodity PCB design companies like JLCPCB, SeeedStudio, and many others.  Why charge such an absurdly low rate for PCB manufacturing?  Because I think they're playing the long game of encouraging self-contained Shenzhen-based electronics design, assembly, and product delivery.  They're leveraging their strength of commodity manufacture and putting it into the hands of inventors.  It's likely they can partially afford to do this because of the coordination and financial support of the Chinese government.

But I think we should ask ourselves: what products and ideas are currently are we never going to see because of the high price of tools like Altium?  The existence and excitement over Kicad, I think, shows at least indirectly the enthusiasm of those who might not otherwise be able to realize their ideas, even if Kicad is not even close to the level of a sophisticated package like Altium that probably embodies thousands of man-years of effort.  Of course the engineers Altium need to make a living, but a economy with too much scarcity limits its growth potential.  I am not a politician or economist and can't really speak how to strike a balance.

[Bud]

I don't think paid support would be working, because usually you get good answers on the KiCad forum for free.

Big companies do not get answers from forums. They want support contracts with service level agreements.

[rs20]

The real question is this: what happens when projects and ideas that currently can not be realized because the availability and cost of PCB design software is a major stumbling block can perhaps be realized with because the barriers to designing a product are further lowered?

PCB design software hasn't been a relevantly significant stumbling block for many years. Anyone who says their project failed because they couldn't handle the PCB software was clearly not cut out to succeed at all the rest of the process.

[jeremy]

The real question is this: what happens when projects and ideas that currently can not be realized because the availability and cost of PCB design software is a major stumbling block can perhaps be realized with because the barriers to designing a product are further lowered?

PCB design software hasn't been a relevantly significant stumbling block for many years. Anyone who says their project failed because they couldn't handle the PCB software was clearly not cut out to succeed at all the rest of the process.

I think you missed the point made by profdc9; the problem is not “being able to handle” PCB software, it’s the cost. And while ymmv, I have worked with a number of startups, and exactly zero of them have used Altium because the cost of even a single license is exorbitant (IMO), only one person can use it at once, and it now expires every year. When you have $20k to get a prototype rolling, why would you spend half of it on PCB software when there are options like KiCAD, Diptrace, etc ?

[Rerouter]

Kicad is not registered under digikey, digikey built up some libraries and integration to make it easier to build around there parts, however at no point was it locked to just them, equally you don't have to use what they supply. its not much different than CERN building the push and shove router, It was to there benifit, but it doesnt lock KiCad to them in any way.

If you get stuck on a task that the other software packages can do, but kicad doesn't make easy, at a basic you can submit feature requests, they don't happen instantly due to the nature of open source, but if its a popular feature its more likely to be picked up, Second alternative is some of the development team can be bought for dedicated development, e.g. pay them for there time, If you ask on the bug tracker you can probably arrange what ever madness you need built (up to limits of man hours), Or if you want it in house, you can write plugins to accomplish most things,

[FrankBuss]

Interesting. I tried using kicad many times and was never able to do anything useful because its ui simply doesn't work for me. Like, I struggled to move/arrange components, a very basic thing to me. In diptrace if I move components, all wires are arranged automatically (not always in an optimal way, but still). So, no broken connections, no weird angles, loops are removed, wires are shortened/extended if needed, etc. Can kicad do the same for me? I have to admin I didn't do much research, but I did spend some time learning how to use it and my conclusion was that user-friendliness was not even on a roadmap.

Are you talking about moving parts around in the schematic or moving parts around on the PCB?

in the schematic it's easy afaik, there's two different commands, one that just moves the component around but breaks the connections and another one that will drag it around keeping the connections.

Moving is by hovering the mouse over the part and then press "m" and dragging is with "g" in the schematic (or press tab, if you have selected multiple parts and wires). On the PCB it works the same, but looks like dragging works only when in legacy mode (F9) not in OpenGL mode (F11), and it doesn't push-and-shove the traces like in the interactive routing (in OpenGL mode), so probably not that useful anyway.

[dave j]

I don't think paid support would be working, because usually you get good answers on the KiCad forum for free.

Big companies do not get answers from forums. They want support contracts with service level agreements.

This. I used to work for a bank that used OS/2 for its OS and the OS/2 version of Mozilla as its web browser. They paid IBM GBP 50K per year to support the browser and all IBM did was ask questions on the Mozilla forums. 

[EEVblog]

As much as I hate to criticize Dave here, I think he missed a much bigger point about Kicad.   I am guessing it is because he worked for Altium and has the perspective of a former Altium employee, seeing an open source product as a viable alternative to Altium.

That's a different question entirely.
KiCAD has a long long way to go to match Altium.

The question isn't really if Altium's sales with be hurt by Kicad, or if Altium or some other PCB software vendor, or other CAD vendor will embrace Kicad as a low-end market solution. 
The real question is this: what happens when projects and ideas that currently can not be realized because the availability and cost of PCB design software is a major stumbling block can perhaps be realized with because the barriers to designing a product are further lowered?

That problem already exists today, this is why low cost solutions exists, and it's why and it's why KiCAD has become popular.

Of necessity, or perhaps in pursuit of maximum profit, companies like Altium must charge for the software, now with term-limited license agreements

It's to maximise and gain sustained predictable profits, this is obvious to anyone who knows anything about how publicly listed companies operate.

However, as our technology becomes more complex, it requires more complex tools that in turn embody the expertise and therefore labor of those who created the tools.  This adds further barriers to designing competitive products and requires more initial capital outlay, so that fewer are able to realize projects that use more complex technology.

The thing about PCB design software is that you don't need those high end tools, they are just easier and faster to use.

But I think we should ask ourselves: what products and ideas are currently are we never going to see because of the high price of tools like Altium?

Practically none. Not having Altium or another high end tool is not a barrier to producing advanced products.

[EEVblog]

The real question is this: what happens when projects and ideas that currently can not be realized because the availability and cost of PCB design software is a major stumbling block can perhaps be realized with because the barriers to designing a product are further lowered?

PCB design software hasn't been a relevantly significant stumbling block for many years. Anyone who says their project failed because they couldn't handle the PCB software was clearly not cut out to succeed at all the rest of the process.

I think you missed the point made by profdc9; the problem is not “being able to handle” PCB software, it’s the cost. And while ymmv, I have worked with a number of startups, and exactly zero of them have used Altium because the cost of even a single license is exorbitant (IMO), only one person can use it at once, and it now expires every year. When you have $20k to get a prototype rolling, why would you spend half of it on PCB software when there are options like KiCAD, Diptrace, etc ?

Nope, RS20 nailed it.
Who cares if a startup uses Altium or not? it's not going to magically make their project better or more viable, or faster, or easier or whatever.
There is practically nothing that Altium can do that you can't ultimately do with free or low cost tools.
Sure Altium might automated things better, maybe making it a bit faster, or maybe making it a bit easier, but not having Altium is not a showstopper.
Any skilled PCB/product designer can use any tool to produce the same end result. Heck, I could still use AutoTrax for DOS if I really had too.
And if you are a startup that has to learn PCB design from scratch then you have more problems than just what PCB design tool you can afford.

[jeremy]

Nope, RS20 nailed it.
Who cares if a startup uses Altium or not? it's not going to magically make their project better or more viable, or faster, or easier or whatever.
There is practically nothing that Altium can do that you can't ultimately do with free or low cost tools.
Sure Altium might automated things better, maybe making it a bit faster, or maybe making it a bit easier, but not having Altium is not a showstopper.
Any skilled PCB/product designer can use any tool to produce the same end result. Heck, I could still use AutoTrax for DOS if I really had too.
And if you are a startup that has to learn PCB design from scratch then you have more problems than just what PCB design tool you can afford.

I mean, you might not care, but Altium certainly cares if a startup uses their tools. Otherwise why do the sales people always talk about that? And why did they create their lower cost tools?

Autodesk even aggressively pursues startups, to the point where you can get an $10k subscription for free (I've done it personally).

It's the same problem with MATLAB; python is eating their userbase (at least from what I know, it's rooted in academia) from the bottom up, particularly due to this whole machine learning fad going on; I read an article today that Netflix is basically all python, including their machine learning, statistics, optimisation, etc which were traditionally MATLAB core functions. I once stupidly told a sales guy for MATLAB that I gave a recent talk inside a university on using python for mathematics/linear algebra (jupyter notebooks), and he was mortified. Students in the past would just pirate MATLAB, but now they are using python instead (and their employers love it, or at least they claim to). Even students and staff inside the university (who get Altium for "free"; it's already installed on their computers) are using KiCAD more and more in a way I never saw with Eagle. This is not withstanding that Altium (both the technical implementation and as an organisation) is a huge pain in the ass with licensing. It's also looking to me like ARM vs RISCV might become a similar battle in some market areas.

The point is not that people are too stupid to make PCBs, it's that they have made PCBs in a previous job, but can't justify the high price of the license fee for the exact reason you stated: "There is practically nothing that Altium can do that you can't ultimately do with free or low cost tools.". The opportunity cost is just simply too extreme.

[Rerouter]

Another in agreement with RS20, the expensive tools give you better optimized options, e.g. you want to run a 16 lane bus of impedance controlled, length matched traces through 300mm of snaking over the board and through Via's, they can probably automate away most of the fluff (I am guessing), the free options still let you do it, it just falls to you to calculate some of the values and spend some more time thinking it through, essentially trading your time vs the cost of the tool.

If the tool is free to begin with, you can train a lot of people from scratch how to use that tool in the same time you could pay off there seats in a more expensive tool, but it comes down to what your doing, and if the designer for your hardware is already an expert in the paid tool, they may have better value hitting the ground running with it (its always a trade off)

With kicad, I have done a few hybrid-flexible PCB's, curving traces is dead easy, and teardropping takes a few minutes if you use scripts from the forum, or a few minutes more if you cheat with using filled zones, there is very little that I have ever come stuck against,

The most memorable was when I was coercing it to show multiple PCB 3D PCB models with components on the same viewer, I was working with another person on the same project but a different board per person, each major change we made, we would re-export the 3D model and for both of us the 3D view would then update with the others changes, I had to write a little patch program to insert a tag into the models after export because it turns out for certain 3D formats it was never really decided on a good way to define metric or imperial measurements, and adding a way to define this went against the standard, but a later version of the file format had a tag for this, and Kicad would interpret that correctly even on the older version, This quirk has since been patched, but that was the main time i have been left in a position where something I was doing involved working around a limitation in the software.

[jeremy]

Rerouter, if I may ask, why didn't you use Altium instead of KiCAD for those boards? Is it perhaps because in your work, the cost of the Altium license outweighs the added usefulness?

[Rerouter]

Started with Kicad in high school, I was a broke kid, and at the Time I don't think I met the requirements for the student license, as such have never used Alitium, The ads for it scream good, but there support forum screams bugs from 5-6 versions ago still being a thing,

I was the earlier hire at my job, when I arrived I told them I was used to KiCad, and so the company just rolled with it going forward, later on 2 other people where added, they where fresh with no prior experience, so they where just trained up in what I had started.

[jeremy]

Started with Kicad in high school, I was a broke kid, and at the Time I don't think I met the requirements for the student license, as such have never used Alitium, The ads for it scream good, but there support forum screams bugs from 5-6 versions ago still being a thing,

I was the earlier hire at my job, when I arrived I told them I was used to KiCad, and so the company just rolled with it going forward, later on 2 other people where added, they where fresh with no prior experience, so they where just trained up in what I had started.

I think you might be the poster child for the point I was trying to make and the exact kind of person that Altium/MATLAB/Autodesk are worried about.