Circuit Break Podcast #421: Tracing a Path for PCB Design Automation with Sergiy Nesterenko

Podcast Title: Tracing a Path for PCB Design Automation with Sergiy Nesterenko

Release Date: 2024-03-12

Episode: #421

Sergiy Nestorenko, founder of Quilter and former SpaceX engineer, discusses revolutionizing PCB design automation. He shares his journey from aerospace to starting Quilter, aiming to transform PCB design into a streamlined, AI-driven process. We delve into the technical hurdles, the fusion of engineering and advanced software, and the vision behind making circuit board design more efficient and intuitive. Sergiy also addresses the potential educational impacts and the future of hardware engineering in an AI-augmented world. Join us for a dive into the evolving landscape of PCB design and engineering innovation.

Podcast Audio:

Podcast Notes:

:rotating_light:**Contest Announcement** :rotating_light:: Introducing a new Circuit Break contest! This contest is themed around building food-related electronic projects. We’re offering over $5,000 in cash prizes, themed trophies, and free prototyping from MacroFab. The deadline to submit is March 31st, 2024. Thanks to Mouser Electronics for sponsoring the contest prizes!

Discussion Highlights

  • Transition to Quilter: Sergi discusses his first experience with PCB layout at SpaceX and the realization that led him to question and eventually automate the process.
  • Autorouters and their Limitations: Exploration of the limitations of current autorouting solutions in PCB design software and why they fail.
  • Quilter’s Vision: Sergi outlines the long-term goal for Quilter to become the compiler for hardware, enabling engineers to focus on design creativity rather than manual layout tasks.
  • Organic and Unconventional PCB Designs: Discussion on how AI and automation could lead to more optimal yet unconventional PCB designs, moving beyond traditional shapes and layouts.
  • Aesthetics of Earliest PCBs: Questioning the assumptions of traces leads to a discussion about the designs of the earliest PCBs in history.
  • Simulation Integration and FCC Compliance: Sergi emphasizes the importance of incorporating comprehensive simulations, including electromagnetic and thermal, into the PCB design process for ensuring compliance and performance.
  • Feedback and Learning from AI: The discussion reveals how AI, like Quilter, can introduce designers to considerations they hadn’t encountered, fostering a learning environment while automating tedious tasks.
  • Exploring Design Variations: Sergi envisions Quilter enabling engineers to explore thousands of design variations, including different stack-ups and materials, to optimize board designs beyond traditional constraints, enhancing innovation and performance.
  • Neural Network Integration: The conversation delves into how Quilter utilizes neural networks not for direct layout generation but for guiding classical algorithms in decision-making.
  • Feedback Mechanisms: Quilter encourages user feedback through various channels to refine its algorithms and user experience.
  • Future of Hardware Engineering: Reflecting on the future, Sergi and the hosts discuss the potential shift in hardware engineering roles with increased AI integration.

Relevant Links

This was a really interesting talk. Motivated by the claims and having just finished a fairly simple board, I gave Quilter a try. I am using a un-routed board and only placed all connectors on the board already to get a mechanically identical solution. My board is slightly above the sweet-spot with 606 pads and I am not going to spend a lot of time adding constraints for traces, because then there would be no difference to a ‘conventional’ autorouter.

Here are snapshots of my work as reference:

Quilter’s solution for fewest layers (2):

Quilter’s solution for shortest tracks (4 layers):

And then there were more candidates with up to 6 layers which I am not going to show here, because in the end they are all very similar.

I understand that this is a very early beta and a lot of decisions were made to give the algorithm a best possible chance of finding a complete solution. 3 mil trace width and clearance is possible, but for manufacturing reliability alone, values of 5 mil / 5 mil would make every board house (and the customer credit card) very happy. Some easy wins for making the customer believe Quilter understands the design would be to reinforce its understanding of mounting holes, clock crystal circuits and input/output capacitors for power supplies. One major issue I found was shorted tracks and unrouted connections in both of recommended and candidate 100% routed, no DRC error boards.
It did however place most of the components into a reasonable area in regards of functional blocks of the board. But then everything was spread out as much as possible to give the traces a best possible area to find a solution. And it did some ground via stitching, sometimes even close to signal changes from top to bottom layer.

Would I use one of those boards without further review? Absolutely not!

Out of curiosity I am going to do another run with the same board, but this time I just remove all traces and leave all components placed. I will report back with this result.

My takeaway so far: I’ll revisit this in about 5 years and hope that all the venture capital money did not get spent in vain.

And if somebody is interested, this board is a RP2040 controller with USB-C connectivity, up to 12V input power, 8 hi-voltage robust input signals, switched power out and a 4-channel composite video output (PAL/NTSC) with separate SPI Flash for graphics data. I am happy to provide board and schematic files.

1 Like

The component placing is what really gets me and probably where the software needs more improvement on.

Placement is really hard. But the autoplacement of those mounting holes? Yeesh! Admittedly, that’s difficult because there are probably no nets attached to them to give any guidance as to where they belong. There probably needs to be extra guidance other than component sizes and net connections.

I agree, there is a lot of truth in the saying, that a good layout is >60% the component placement. Yup, the mounting holes were a good one. But I have to admit, I did this on purpose to see how it deals with them. Mounting holes are a prime example for a constraint before the machine takes over.

And as promised here are the results, when I took my board and removed all trances but left all components placed already. Showing results for the ‘shortest traces solution’. It made a 4-layer board for this board.

Does not too bad at first, but not really.
Some details:
Massive short between VBUS, GND:

Shortest V_IO?:

SPI signals under crystal; 2 signals between pads of 0603 load capacitor

With a few more constraints and review / rework afterwards, this board might even have a chance of being operational. My job is still safe. :wink:

And the best part of the 2nd attempt was, when I entered the data, Quilter send me a passive aggressive note that the components are already placed and it would like to re-design the board. Sounded like it was miffed that I did the work instead. :rofl:

Is it “Yay, my job is safe!” or “Boo, I’m still doing manual place-and-route!” ?

It is one of the tasks I really really enjoy, so Yay!
BI, biological intelligence for the win.

1 Like

Thanks EE_Artisan! That was really informative.

With all respect to Sergiy, many of the algorithmic problems he mentioned in the podcast have known solutions. I’d recommend the following books:

  1. VLSI Physical Design: From Graph Partitioning to Timing Closure by Andrew Kahng
  2. Handbook of Algorithms for Physical Design Automation by Charles Alpert

I was thinking that I can technically have my design placed and routed by a contractor now so if I wanted to pay to free up my own man hours I can already do so. I don’t do it for many designs because it doesn’t seem worth the expense which I think brings up an interesting question. How much would you pay for such a service?

For a hobby project I’m not sure I would pay anything because the whole point of the hobby, for me, is the journey. I’m wondering how many others feel that way.

1 Like