Electronics

Next-Gen Software-Defined PCB Design For Hardware Designers

Guests Duncan Haldane | Uploaded : 12/12/2023


The EEcosystem Podcast

Next-Gen Software-Defined PCB Design For Hardware Designers

Duncan Haldane, CEO and Co-Founder of Silicon Valley startup, JITX discusses his vision for increasing success and productivity for hardware engineers through automated software-defined PCB Design. Early release successes show promising indicators that JITX will help usher in a new and exciting era for design engineers.

Links & Resources

🤓Join the EEcosystem Engineering community 📬 and subscribe to our newsletter to receive a free copy of Samtec’s Signal Integrity Handbook. http://theeecosystemp.wpenginepowered.com

Show links: 💻 JITX Website and free demo https://www.jitx.com/

🗞 📰For SI/PI/EMI News and Technical Resources and to register for a free subscription visit the Signal Integrity Journal today. https://www.signalintegrityjournal.com/

💾 For free Technical Resources and to Learn more about Keysight Pathwave EDA Software Solutions visit the homepage now. https://www.keysight.com/us/en/products/software/pathwave-design-software/pathwave-advanced-design-system.html

👀 Visit Summit Interconnect for all your complex PCB manufacturing needs. https://www.summit-pcb.com/

🎯For all of your high-speed and RF connectors visit the Samtec website and access excellent engineering resources while you there. https://www.samtec.com/

🤓Visit the SIGLENT website to learn more about their test and measurement solutions For Every Bench. Every Engineer. Every Day. https://siglentna.com/

📡For Custom RF and MW PCBs visit the Transline Technology Website to learn more. https://translinetech.com

💾 For high-complexity EDA solutions visit SIEMENS EDA Website https://eda.sw.siemens.com/en-US/

🔌 Picotest specializes in high-fidelity testing and measurement tools, primarily for power-related applications. Visit their website for more product information and excellent training materials from expert Steve Sandler. https://www.picotest.com

Transcript

								 Judy Warner:
Hi Duncan, thanks so much for joining me today. I'm really looking forward to our conversation and learning more about Jetix.

Duncan Haldane:
Hi Judy, thank you so much for having me on.

Judy Warner:
Well, before we get off and going deep in the weeds, can you tell our audience a little bit about your educational and professional background?

Duncan Haldane:
Sure, so my most recent education was as a robotics PhD student at the University of California Berkeley. And then as professionally there I was doing paid research, so building robotic platforms to test new ideas about how we can take inspiration from nature. So how does this, how do, well, why does this monkey jump so well was a question I asked and what can we learn about that for to make better jumping robots.

Judy Warner:
Right.

Duncan Haldane:
From there, I transitioned right into JDEX. So I met my co-founders at Berkeley and we started to work right away.

Judy Warner:
So tell us about the relationships with your founders and, and what led you to go from jumping monkeys to, to JetX. Well, go ahead and answer that question, but then we have to fill in our listeners, what JetX actually does, but go ahead.

Duncan Haldane:
Sure, yeah, so to set the context, JITX is a way to design circuit boards by writing code instead of taking the usual graphical approach where you're drawing a schematic and Googling for parts and building spreadsheets. So it's just, how do we take hardware engineering and put it more into a code discipline similar to designing a chip or programming an FPGA? So that's the high level. And I got there based on my experience as a roboticist. So robots are complicated. to design and they're really hard to design if you don't want to be like broken all the time. And so I found that like that was one of the hardest parts of my job. Right? Like I could do the science, I could study nature and write down this like little sketch of an idea on a napkin. And then I spent all my time as a hardware engineer. to build this robot so it's robust enough so I can do like repeatable scientific experiments, which is a pretty high bar. So we invented a new way to design things like new optimization methods to search for robotic mechanisms. And I found that really rewarding. And then I took a class from someone who's now my co-founder, actually in the architecture department, so I was so desperate to automate. my hardware engineering work that I was like, I gotta find people automating hardware. And this guy was teaching a class in, he's computer science, but he was teaching in the architecture department. And architecture is in buildings, not as in computer architecture.

Judy Warner:
Right, right.

Duncan Haldane:
And it was about automated design. And how do you take... hardware engineering

Judy Warner:
Mm.

Duncan Haldane:
process no matter what really like and turn it into software so that new designs you basically just change some parameters and hit run and you get everything like all of your manufacturing files out the other side and he'd been doing this for like 20 years at that point with at MIT at other lab so that's Jonathan Bachrock who was a assistant professor at the time but is now one of the co-founders and then Patrick or other co-founder was his student at the time. We all just

Judy Warner:
Okay.

Duncan Haldane:
liked working together. We said like, automating hardware engineering is probably the most impact we could ever have on the world.

Judy Warner:
Yeah.

Duncan Haldane:
And we really liked working with each other. So we decided to spin out Gidex.

Judy Warner:
Okay. So I've been around hardware engineers and board designers for all of my adult life. And they always say, you know, I've never met an auto router that doesn't suck.

Duncan Haldane:
Ha

Judy Warner:
Or

Duncan Haldane:
ha.

Judy Warner:
why do they, why, how can we do such great, um, rendering say in, uh, building buildings or building semiconductors. And then we're still dragging traces. Around.

Duncan Haldane:
Yeah.

Judy Warner:
Now 3D, when 3D came, that was a massive innovation. So I think I want to start with that. Like,

Duncan Haldane:
Yeah,

Judy Warner:
why

Duncan Haldane:
all right.

Judy Warner:
is it, why has it stayed the way it is in your mind? And, and how do you imagine solving that problem?

Duncan Haldane:
That's a really great question. So I think, why is it the way it is? I think you can look at the history of the industry. So 1984, chips got pretty complicated. And they found that, hey, you could design more complicated circuits if you put down the schematic drawing tools and write code instead. And Synopsys is like, you know, we're going to start designing your chip for you. We're going to have we have this thing called synthesis and synthesis gives you correctness and optimization. And that's just like when you get those things, you can design better hardware. But that didn't happen for circuit boards at the time, even though I think the company started at the time, you had like auto tracks back then with an X. Right. And I think

Judy Warner:
Right.

Duncan Haldane:
that. through a series of acquisitions is now in Altium. It was like Protel

Judy Warner:
Right.

Duncan Haldane:
and then it was Altium.

Judy Warner:
Yeah.

Duncan Haldane:
But boards in the 80s were pretty simple, relatively simple. Like

Judy Warner:
they

Duncan Haldane:
people

Judy Warner:
were.

Duncan Haldane:
did heroics in the 80s to design circuit boards and Mac computers that actually worked. But it doesn't look like what we have today, right? So 2008, things got weird. You have like the iPhone, right? So now it's got, it's not this kind of big square thing. It's super compact and it has to last a really long time on the battery and it's got three radios. And that, it doesn't look like the same design, but you're still using the same tools because all

Judy Warner:
Right.

Duncan Haldane:
the tool companies grew up back then. You know, they figured out the product that was useful. It worked for the technology of the time. And then it got pretty well locked in.

Judy Warner:
Mm-hmm.

Duncan Haldane:
But today things are much more complicated. and the labor is different and the way we design stuff is different. But that's why it is the way it is today. One of the big reasons.

Judy Warner:
I see.

Duncan Haldane:
But that's, I think that's a little less interesting than talking about why otter-otters suck so much.

Judy Warner:
Yeah, so let's

Duncan Haldane:
Yeah.

Judy Warner:
talk about that.

Duncan Haldane:
So... I think so people definitely use auto routers and I'll tell you today some very large hardware designers are making their tool choices based on how good the auto router is. So because there's a massive labor shortage, the people that laid out their boards are retiring and they're not being replaced. So there's

Judy Warner:
That's

Duncan Haldane:
a

Judy Warner:
right.

Duncan Haldane:
need for automation. But I think the one reason auto routers are bad is like, there's two, I think, primary ones. One, they don't have enough information to do a good job. Right? They don't. Unless you spend a lot of time typing in constraints, they don't know what your design is, or what you're trying to do, or about the new material you're using, or what a good design even looks like. What do they track? Is it connected? Right?

Judy Warner:
Right.

Duncan Haldane:
How many views did I use? I was like, that's a start, but it's not like there's a reason people route their sensitive traces by hand and artistically, instead of throwing it to the auto-router.

Judy Warner:
Right.

Duncan Haldane:
So there's this like huge bottleneck in just entering data that makes auto routers work that's so high

Judy Warner:
Hmm.

Duncan Haldane:
that it's not productive anymore. Right.

Judy Warner:
That makes perfect

Duncan Haldane:
So

Judy Warner:
sense to me.

Duncan Haldane:
that's a big one. And another one is I think the way routers work does not fit with a realistic place-in-route flow. Because imagine that an auto router worked perfectly, what do you do? It's like, oh well, I will drag every component I have to its perfect position on the board. And I don't need to swap any pins. and I don't need to move anything, I'm just gonna exactly position everything and then hit route, and then it's going to do the traces. And nobody designs that way.

Judy Warner:
Yeah.

Duncan Haldane:
Right? It's like, 90% of routing is placement. Right? So I think to do a good job with an autorouter, you have to solve the information problem, so you actually know what you're doing. And then you have to, like, really embrace the human as part of the autorouter loop, rather than, like, the autorouter is a piece of software that you run.

Judy Warner:
Hmm. And so that's really like, again, with looking at a model, like with a synopsis or whatever, and you saw this can be solved with code. So

Duncan Haldane:
Mm-hmm.

Judy Warner:
we're going to talk about a few different things around that. But my first question is, you know, on the back of what you just said about auto routers, how, once, once you. finish say, finish Jetix, all of the development, all the vision that you all have as co-owners. How, what is that experience going to be like? What is, what do you imagine will be? the design or

Duncan Haldane:
Mm.

Judy Warner:
how will it perform for a person that is taking that on.

Duncan Haldane:
Right, right. So when we accomplish what we set out to build, what is the impact for the person using that software? What are we

Judy Warner:
Yeah.

Duncan Haldane:
delivering for the designer? Right?

Judy Warner:
Right.

Duncan Haldane:
The biggest thing, I think I'll know success when like the vast majority of designers can design better hardware with Gidex than without it. Like that's the bar for success

Judy Warner:
Mmm.

Duncan Haldane:
in my mind, right? It should be a tool for helping people be more ambitious. Right? To have like, cause hardware is so risky right now. Cause there's no time anymore to make mistakes. So one thing we do is like scrub designs automatically for mistakes. So you take a schematic review and you automate it and you run it on every single time you run your code. So it catches the issues that you'd only catch later by having an expert read your schematics. You try to take

Judy Warner:
Mm.

Duncan Haldane:
some of the risk. out of that design process, users

Judy Warner:
Hmm.

Duncan Haldane:
can focus on exploring. So people can focus on

Judy Warner:
Yeah.

Duncan Haldane:
optimization, so they can try out new ideas, so that evaluating that next generation radar chip is something you could do today instead of like the eighth thing on your backlog once you clear all of this work that's on your plate because the rocket needs to go or the car needs to drive. Like there's so much pressure put on designers today. Um, what I

Judy Warner:
So

Duncan Haldane:
really want,

Judy Warner:
yeah,

Duncan Haldane:
yeah.

Judy Warner:
I heard you, I heard a bit of that earlier in your conversation when you said robotics is so much fun. And then I had to make hardware.

Duncan Haldane:
Yeah.

Judy Warner:
Right. It's like you go in, I, a lot of engineers I've known are passionate about engineering and making engineering, doing those things. And then they get caught up with schematics and component shortages and things going out of stock and somebody waiting to review the design and then it and you get you spend a Vast majority of your time doing that Do you think jet X will be able to in some way free up the engineer to do actual engineering?

Duncan Haldane:
Yeah, I mean, and I don't want to knock that work. That is engineering. That's a human experience of engineering.

Judy Warner:
It is.

Duncan Haldane:
But yeah, we're attacking that today right off the bat. Like we want, the first goal is like, just make people really productive, right? And when you look at like, what does it take to move fast? He's like, well, you got to create designs quickly that are free from errors. And that's our North Star is just like the fastest way to make a board. And to make a board, there is a human engineering organization around it and interacting with that is also engineering. It's not maybe their favorite part of their job, but it's hard, and it's not something people can just kind of walk into.

Judy Warner:
Mm-mm.

Duncan Haldane:
And yeah, so we're, you know, we do things like... Alright, here's your power regulator circuit, now you can make it parametric. So, you just say what you want the startup time to be and the output voltage to be, and then the calculations that are in the spreadsheet, that's just code. And you run it, and it's gonna find your passive values, and then, oh, it's gonna go source from your preferred inventories, like, which exact parts to use, and then it's gonna put those in the design and run a schematic review. And that's what it takes to, like, productively automate something, because

Judy Warner:
Mm-hmm.

Duncan Haldane:
if you just do part of it, it's just like, well, here's your circuit design, but you can't buy the parts, and I don't know if it's right. Like, that's not productive.

Judy Warner:
Yeah, I was just talking to a, I just did a podcast with someone from Keysight. And he said, hope cannot be part of the design process. Right.

Duncan Haldane:
but it's always there, right?

Judy Warner:
I know

Duncan Haldane:
Like

Judy Warner:
it's like,

Duncan Haldane:
there's

Judy Warner:
I

Duncan Haldane:
no...

Judy Warner:
think I got it mostly right.

Duncan Haldane:
Yeah,

Judy Warner:
And.

Duncan Haldane:
it's like, yeah, I gave my expert engineers this 200 item checklist and asked them to read these 80 pages of schematic sheets and find all the issues. And they said they did that thing, and it's definitely better than having them not check it. But that's not a thing that humans enjoy doing or are especially good at. Like, as a super senior expert, I don't want to spend my time reading all the schematics or working on one project at a time. Like, you find this industry is driven by those people, by the gurus, and sometimes they have a capital G when they introduce themselves in

Judy Warner:
Yes.

Duncan Haldane:
guru, but they still work on one project at a time.

Judy Warner:
Right.

Duncan Haldane:
Right? And some of them are kind of sick of that.

Judy Warner:
Most that I talked to are definitely sick of it. So how far along are you with JetX right now and realizing some of those North Star and things that you're aspiring to as a company and a tool?

Duncan Haldane:
Yeah, so we launched general availability September of last year. And we have now just, and when we launched at the time, it was a sort of partially automated tool. You know, you do your high level requirements in code all the way through component selection and design checks. So you actually know that your schematic is going to work. And then... just recently we've launched our routing tool. So now this is like another way to optimize your design. So it has this new capability to help optimize, which is you can do your pin swaps interactively. So you don't have to like read your PDF and then look at your layout and then like figure out where in your schematic you can change the pins to make things more routable.

Judy Warner:
Mm.

Duncan Haldane:
And so yeah, we've just shipped that as a, oh, it just feels so good. So like the... the part where it's like it does your schematic and then you know you export over to Altium and it finished your layout. And that was fine. It was productive. But it didn't have like that magic feeling. But now when we bring in the routing algorithms that are helping optimize the entire design, I'm just I'm really happy with how it feels, honestly.

Judy Warner:
It's

Duncan Haldane:
It's

Judy Warner:
like,

Duncan Haldane:
hard to characterize.

Judy Warner:
it's like your, your first born child or maybe your second. I don't know. Something like that. It's very satisfying.
 

Duncan Haldane:
Yeah.

Judy Warner:
go crazy. It's wonderful.

Duncan Haldane:
Yeah, I hope we can build a better bridge.

Judy Warner:
I think so. Well Duncan, I know we're about out of time, but thank you so much. I hope you will come back as you keep improving this tool and share with our audience all the progress you're making. Is there anywhere you'd like our listeners to go besides your website to learn more about JetX or to get a demo or what would you recommend for people wanting to learn more about JetX?

Duncan Haldane:
So our website, jitx.com, is probably the best way to see the product, maybe get a demo. They're also free to email me directly. I'm always very interested in hearing all sorts of feedback. So they can reach me at d.haldane at jitx.com. And maybe we can put that in the episode description.

Judy Warner:
We definitely will. Well again Duncan, wish you all the luck. It sounds like you're really making some good progress in the area that's long overdue. So thank you so much for coming and sharing your journey with us and best of luck to you in Jetix and please come back and give us updates.

Duncan Haldane:
Yeah, thank you for having me, Judy. This was a great conversation.

Judy Warner:
Thank you so much to our listeners. I will make sure and put those links for you below in the show notes or in the description. So go on over to JetX.com and take a look or reach out to Duncan if you're one of those people that wants to help them go along the journey with them and help sort of co-develop this tool. I'm sure he'd be thrilled to hear from you. So thank you so much for joining us today on the podcast. I hope you've enjoyed this conversation. We'll see you next week. Until then, remember to always stay connected too. the ecosystem.