@Matt Venn If there's room left on TT02, could we fill the rest with as high power of a mosfet as we can fit? I ask because I've recently discovered this new amazing world of open source chips and would love to design the first open source BLDC motor driver so we can have more open source robots in our lives doing useful things. What I've been struggling to find docs on, is how high power of a Mosfet you can do with sky130. Am I correct in assuming if you just add more fets in parallel, you get more current handling, so it's just a matter of count? Which then means its a matter of how much space you use up on the chip? Any BLDC driver is going to need 6 high current mosfets.

Much better to make your own application. Tinytapeout is a very specific design and I don't want to add random other stuff to it that could cause it to break

Gotcha. Yeah, I was thinking your multiplexer would mean this doesn't make sense for tinytapeout.

it's not a power mosfet either

Ok, I'll ask over there. Thanks and loving your videos btw. They've gotten me further on this than anything else out there. You've got me totally jazzed.

@John Lauer: There was a power MOSFET design done by @Weston Braun and is currently being expanded and tested by @Jorge Marin for the IEEE-SSCS ChipIgnite run. The GF180MCU process appears to have been used for automotive chips originally, and has some very high-current digital drivers on it. The standard GPIO cell drives up to 16mA, and there is a high-drive pad that drives up to 24mA. Everything in the GF180MCU process runs at 5V (it has 3.3V capability but I don't have any infrastructure to support it yet).

Ok, that sounds very promising. So it sounds like I'd have to look at all the 180nm pdks, rather than the 130nm I've been focusing on.

The art of CMOS power transistor design is in doing the layout of the interconnects in such a way that the current is evenly distributed over the whole design. If you get current crowding this will impact reliability of your transistor. AFAIK the open source analog design tools don't provide the means to check for this; e.g. thermal analysis, electromigration checks, etc.

GF180 has 3.3V, 5V, 6V and 10V capabilities but only 5V standard cells are provided at the moment. Which voltages do you need for your MOSFETs? Do you also need ADC sensing?

@John Lauer Contact me at tspyrou@precisioninno.com. I am running the OpenROAD development team. We shold have a discussion. @Indira Iyer pointed me at this post.

Would love to chat with you Tom. The idea on the BLDC controller is to try to get the first open source controller for a broad range of 3-phase motors. Having it run drone motors would be a start, but lower current for gimbal motors would be an amazing start. We're still talking 500ma. Getting to 2A would be even more useful. It would be nice to eventually be able to power motors for actuators in walking robots like the MIT Cheetah/Boston Dynamics dog style robots. Those require upwards of 48v / 50a power. BLDC's will run on any voltage, it's just a matter of enough current. Higher voltages run better through motor coils. And, yes, we would have to monitor the current so would need ADC built-in.

There are openings in the new gf180 mpw shuttles funded by Google if you could be ready to tapeout dec 5th and work within the efabless caravel harness.

That would be amazing, but not sure I'll have time over Thanksgiving with guests in town. Maybe next week I can get some stuff done. I've never done this before, so have a big learning curve ahead of me. Would LOVE to get it into that tapeout though.

Oh got it. The gf shuttle would be a good way to make a simple chip and get samples back to play with. We dont have any high drive mosfets on the shelf but could find a contractor to make one.

Interesting. So the contractor would make the design and then it would become part of the open source repository? That would be even better.

Thanks @Tim Edwards for the tag. Hello @John Lauer. I'm a postdoc researcher at AC3E in Chile, we're starting an analog/mixed signal group within a research line with expertise in (discrete) power electronics and now we are exploring the integration of power electronic topologies. Your project sounds really interesting, and though we don't have design experience in GF180, we are definitely interested in exploring its limits as we are doing for SKY130. Let us know if there are any topics in which you think we can collaborate.

Hi Jorge, great to e-meet you. I'd love to collaborate. It seems to me that one of the first things we could do is model the MOSFETS in Spice just to see what's possible in the kind of area available on these chips inside these free programs. Have you guys actually been able to build any of this for SKY130? Or are you just starting out like I am?

@John Lauer Integrated Power FETs are mostly physical/mechanical design problems. 2A is certainly reasonable. GF is probably a better process though given the higher voltages. The challenge is in how the FETs are interconnected, how the terminals are brought out to package pins, how many pins are required, size of bond pads to accommodate the requisite bond wires, and managing the heat due to power loss. There are commercial tools (field solvers) that are typically used to address the on-chip challenges. For motor applications, one will need to consider inductive flyback voltages. Typically, a process with a buried layer is required, but judicious choices for external Schottky diodes can mitigate many of those issues.

Hello @John Lauer, we are currently trying to tape out a 3-level flying cap dcdc converter in sky130, and we would like to iterate a bit more on this design because we have unsolved challenges: https://platform.efabless.com/projects/881 Power device modelling is one of them, so any discussions regarding this are interesting to us.

Hi. So I was told by Philip about this conversation. I'm open for commissions of projects at the moment. If someone would be willing to donate/finance the hours (two weeks, 50$/h) I'd have to invest into it, I could write a set of scripts which would generate MOSFETs with proper heat distribution based on parameters like target I_max, as well as current limiting and reverse polarity protection and all the other stuff needed for power MOSFETs. Send me an email to leviathan@libresilicon.com if you're interested

Hi David, yes, I'm quite interested in talking to you further about commissioning you for help on this. I could fund you for two weeks on this.

If you make this open source we can host the work on the open road project if that is helpful.

Yes, this would be open source. Would love that.

I suggest, that I create a project on the LibreSilicon redmine (https://redmine.libresilicon.com/projects), where I track bug reports, milestones and hours, and just mirror everything to your open roads project (It's actually a simple hook in gitolite, which allows autopush to backup repos when pushing to gitolite)