EDA is dead. What comes next is exciting.
EDA is ‘Electronic Design Automation’, a term born in the 1970s to describe ‘automated’ tools that help electronics engineers design products — circuit boards, chips, simulation, verification, etc. ‘EDA’ isn’t dead because it is old, but because it now embodies everything that we despise about our tools.
Phases
I’ll talk about four EDA periods: pre, dawn, contemporary, post. In the pre-EDA days, 1970s and earlier, there were no design constraints imposed on engineers. Designs were largely free-form, easily and naturally capturing the individual creativity. Circuits and ICs were simple in comparison to today’s, so hand-drawing and eye-verifying them was manageable. Circuits from that period are beautifully laid out and organic-looking.
With circuits becoming increasingly complex, and computing increasingly accessible, some tasks started to be automated with software. The dawn of EDA period was highly innovative; developers needed to create software that can automate complex computing tasks with limited computing power. This scarcity of computing resources practically meant imposing strict constraints, and so engineers were no longer able to be as free with their designs.
There was no UX back then, and it really didn’t matter as long as the software worked (sometimes satisfied by simply not constantly crashing). The software was inherently single-user (collaborative design was not a priority, but also impractical at the nascent days of connected ‘personal computers’). A high-switching-cost lock-in mentality was cultivated, as companies identified it as a way to land-grab, capture and control customers, and also as a way to influence standards with their interests. This is ‘EDA culture’.
The legacy from the creative ‘dawn’ years — awful UX, single-user, lock-in — proceeded through the contemporary-EDA period. Unlike our software-development community counterparts’ tools, ours were not re-imagined to benefit from new standards and development methodologies (such as revision control, continuous integration, etc.) Our tools are, at the core, still the same as they were twenty years ago, receiving a facelift with new Windows icon sets to look more modern, like Botox. When your interface looks like a 747 cockpit one needs to consider whether the tool is fit for purpose.
Innovative-company tributaries eventually formed the ‘Big 3’ — a collection of vary large EDA companies such as Synopsys, Cadence, Mentor — through consolidation and acquisition. The ‘Big 3’ set the tone of the industry. ‘EDA culture’ is so entrenched there, that it is hard to imagine that they are capable to usher a new age of design tools. Would they make their entire EDA-dev team redundant? Will they rewrite their entire suite of tools using modern development and UI concepts? Unlikely.
The temptation to do something about it is strong with anyone who has ever used these traditional tools.
A seat at the table
I constantly see people trying to improve our tools situation. But the reality is that most innovators in EDA still see the ‘Big 3’ as the only way to ‘succeed’ (being bought) so they are driven to imitating the existing tools instead of radically innovating.
In 2011 I implored the founders of Upverter (browser circuit design tools for engineers) and circuits.io (browser circuit and simulation tool for makers) to innovate rather than imitate bad desktop tools. They didn’t. I also told them that simply adding ‘social’ to something isn’t innovation. Circuits.io was gobbled up by Autodesk a couple of years later and Upverter is still going with a tool that’s roughly a browser-based functional version of a traditional desktop tool.
Zak Homuth from Upverter recently wrote why disruption as a startup that wants to be bought by the ‘Big 3’ is hard
But as I was writing that post, I noticed that it was quickly becoming a laundry list of all the ways traditional EDA makes it hard to disrupt them. All the walls they’ve built that keep startups out, and lock customers in. All the things preventing real innovation. All the little battles and wars we had to fight at Upverter just to get a seat at the table.
The ‘seat at the table’ is the key problem. When you accept a seat at that table you are required to abide by ‘EDA culture’. It follows that it’s much harder for you to innovate and ‘disrupt’. Why would the bears at the table allow a bee to disrupt them?
I understand Upverter’s position, though. They need to maximise share-holder profits, and have therefore become risk-averse. I’m less understanding of open-source tools such as KiCAD and gEDA, that have the capacity to take risks and disrupt, as they are not shackled by VC money. The benefit of having a tool ‘open-source’ only goes so far; it actually has to be better somehow! (This is why Eagle is still popular with the hobbyist/maker/startup crowd despite being outmoded crippleware.)
With all that they’ve learned so far, I think that a brave decision would be for KiCAD and gEDA to join resources, with the help of CERN funding (see below for other potential resources), and create a new tool from scratch to overtake Altium, Eagle, and the rest, at least in some areas. I’d love for that to happen.
The only way ahead is to recognise EDA culture, and actively steer clear of being sucked into the ‘EDA vortex’. Call ‘EDA’ something else. Create a new table. Innovate long and hard. Eventually appreciating parties will join the feast. Who might those be?
Small hitters
We have a new generation of electronics enthusiasts that are coming from the software and design world. We need to actively welcome them, and stop snubbing them. They bring modernity to our craft, and are also experimenting with creating tools that provide us with a view into lingering problems. We need to learn from them and understand their needs (instead of telling them that ‘it has always been done that way’). These are our new consumers and contributors — not the engineers working for big companies who are prevented by bureaucracy, sunk costs, and internal empire-building from adopting new tools and ideas.
Take a look at Z-of-Z (highly visual Gerber viewer, Windows Only) and Copper (unsuckify Eagle design visualisation, Mac only). Neither are perfect, but show the tools people are creating to resolve the lack of real innovation in traditional tools.
Big hitters
Google, Facebook, Twitter, Microsoft, Amazon, and many others, famously innovate where there was a void, or tools/platforms unfit for purpose. Their public facing products may be web and software, but many of these big companies are also hardware companies. More crucially, they are hardware developers with a software development mentality. They will, eventually, take interest in innovating in post-EDA era tools. They will invest, buy, contribute, and acquire.
Here’s my advice to young companies in our industry: focus on newcomers and look for interest from large tech companies other than the ‘Big 3’. The ‘old-timers’ will eventually take interest.
Complaining is easy
That’s true. It’s much easier to complain than innovate. I’ve tried, failed, and am trying again, differently.
I founded Boldport when I created a browser-based automated build management tool for FPGA projects. It failed within a year and I learned a lot about the EDA business in the process. I took a break and then re-invented Boldport as an ‘electronics craftsmanship’ design company after creating an open-source PCB design tool called PCBmodE. I now consult and run the Boldport Club, where people can get a monthly delivery of unique electronics kits.
In the past five years I’ve written about the EDA business and my personal frustrations with innovating in it. When I re-started Boldport and authored PCBmodE I mentally rejected the notion that it is an EDA company and tool, respectively. I also purposefully created a business that does not rely on the adoption of an “EDA” tool for keeping the company afloat — fool me twice, shame on me — if it ever picks up interest that would be a wonderful. In the meantime I treat it as an internal tool that’s open-source.
Here are a few problems I’d like to see people tackle, some of which are on the roadmap for PCBmodE
- Completely and elegantly solve backwards and forwards annotation. This has caused a decades-long headache.
- Explore layout-led design. Layout encapsulates all the information, and it makes sense to me to start there, rather than with schematics.
- Improve schematics and layout presentation on screen to help with identifying problems.
- Smart DRC that takes functional information into account. This is hard, but we’re getting to the point where vendors supply machine-readable datasheets.
- Minimal UI with a highly intelligent back-end that does the ‘right thing’ most of the time. Let’s take a lesson from other fields to see how we can do better.
- Abandon auto-routing (no-one uses it anyway) and invest in guided routing that makes sense.
Finally, I regularly get contacted by people trying to innovate in the area, but are frustrated with the response that they get. Not by the feedback itself, but by the lack of any interest in giving feedback! I explain to them that I think that they are going after the wrong crowd, and this article frames the argument in more detail than I usually provide. I truly think that those who will reject ‘EDA culture’ and avoid the ‘EDA vortex’ of bad design, will eventually succeed, but they need to be committed to a long journey.
