4 ms·

As much as I love the futuristic and sophisticated sounds that electric traction drives make as compared to their fossil-fueled counterparts, I do think that th

As much as I love the futuristic and sophisticated sounds that electric traction drives make as compared to their fossil-fueled counterparts, I do think that the driving and control systems that actually generate the PWM waveforms that are sent to the motors are even more interesting. Huge, massive IGBTs weighing a kilo+ each chopping up hundreds of thousands of joules/s of energy without destroying themselves, current transformers, contactors, it really tickles parts of my brain. He only spent a handful of seconds on those, would be interesting to hear a deeper dive.

itishappy 2y ago
Same! That stuff is so freakin' interesting! The other deep dive I really want (pun intended) is submarine sonar transducer drivers, which as I understand do basically the same thing but bidirectional and with a LOT more waveform customization.
rwmj 2y ago
Photo on Wikipedia: https://en.wikipedia.org/wiki/Insulated-gate_bipolar_transistor https://en.wikipedia.org/wiki/Insulated-gate_bipolar_transis... Are they manufactured using regular silicon lithography? You'd think when making something so large you could just mix up the chemicals instead.
- theideaofcoffee 2y ago
  From what I understand, mostly yes. They are still built up layer-by-layer with standard litho techniques, with pattern masks, ion implantation for the p- and n-type layers, etching, CVD to deposit insulation material (for the 'insulated gate' part of the device). Though they don't have to be nearly as precise, as I understand it as something more sophisticated, like a cpu, they still need to lean on those to make it chooch.
- bgnn 2y ago
  I don't thik they're produced with the same process as CMOS, but I bet some photolithography is involved.
- Animats 2y ago
  It's amazing how small those things are. That's a transistor which can switch four megawatts. A manual switch with the same rating is bigger. Here's a data sheet for a somewhat larger one.[1] The control signal is only 6 volts at 5 mA. It's incredible that this is physically possible. [1] https://www.infineon.com/dgdl/Infineon-FZ1500R45KL3_B5-DataSheet-v01_00-EN.pdf?fileId=8ac78c8c821f38900182aac869dd3b27 https://www.infineon.com/dgdl/Infineon-FZ1500R45KL3_B5-DataS...
  - leoedin 2y ago
    As an embedded engineer sometimes working on PWM, it's interesting to think that the origin of the gate drive signals is still just some sort of microcontroller timer running at 3.3V and a few mA. I'm sure there's a few layers of level translation and isolation between the micro and the 4 MW power transistor - but it's still indistinguishable at the digital level from a circuit driving a tiny motor on my desk.
    - 0_____0 2y ago
      An ant driving an M1 Abrams main battle tank.
- rcxdude 2y ago
  It's definitely lithography on silicon wafers, but a lot of the details are very different. For example, it's pretty common for power devices to be manufactured to be formed from the full thickness of the wafer, as opposed to just on the surface. And a single device tends to be formed of many repeating blocks, since you do need fine details still to get the performance (And a lot of the challenge is making sure each of those blocks is as similar as possible, so you don't get hot-spots). (silicon lithography is more of a broad range of processes: digital vs analog vs power vs MEMS all have very different needs, and so a process designed for one will look very different to another, and one designed to allow multiple will be making compromises)
  - namibj 2y ago
    The "full thickness" aspect is commonly called "vertical" when talking about the devices.
1970-01-01 2y ago
Very early EVs (GM's EV1 and then a few other GMs) used banks and banks of water cooled IGBTs to spin the motor.
NikkiA 2y ago
The Mercury Arc rectifiers that powered the earliest electric locomotives were prettier to observe though: https://www.youtube.com/watch?v=whF80m6oQZ8 https://www.youtube.com/watch?v=whF80m6oQZ8

itishappy 2y ago

Same! That stuff is so freakin' interesting! The other deep dive I really want (pun intended) is submarine sonar transducer drivers, which as I understand do basically the same thing but bidirectional and with a LOT more waveform customization.

rwmj 2y ago

Photo on Wikipedia: https://en.wikipedia.org/wiki/Insulated-gate_bipolar_transistor https://en.wikipedia.org/wiki/Insulated-gate_bipolar_transis... Are they manufactured using regular silicon lithography? You'd think when making something so large you could just mix up the chemicals instead.

theideaofcoffee 2y ago
From what I understand, mostly yes. They are still built up layer-by-layer with standard litho techniques, with pattern masks, ion implantation for the p- and n-type layers, etching, CVD to deposit insulation material (for the 'insulated gate' part of the device). Though they don't have to be nearly as precise, as I understand it as something more sophisticated, like a cpu, they still need to lean on those to make it chooch.
bgnn 2y ago
I don't thik they're produced with the same process as CMOS, but I bet some photolithography is involved.
Animats 2y ago
It's amazing how small those things are. That's a transistor which can switch four megawatts. A manual switch with the same rating is bigger. Here's a data sheet for a somewhat larger one.[1] The control signal is only 6 volts at 5 mA. It's incredible that this is physically possible. [1] https://www.infineon.com/dgdl/Infineon-FZ1500R45KL3_B5-DataSheet-v01_00-EN.pdf?fileId=8ac78c8c821f38900182aac869dd3b27 https://www.infineon.com/dgdl/Infineon-FZ1500R45KL3_B5-DataS...
- leoedin 2y ago
  As an embedded engineer sometimes working on PWM, it's interesting to think that the origin of the gate drive signals is still just some sort of microcontroller timer running at 3.3V and a few mA. I'm sure there's a few layers of level translation and isolation between the micro and the 4 MW power transistor - but it's still indistinguishable at the digital level from a circuit driving a tiny motor on my desk.
  - 0_____0 2y ago
    An ant driving an M1 Abrams main battle tank.
rcxdude 2y ago
It's definitely lithography on silicon wafers, but a lot of the details are very different. For example, it's pretty common for power devices to be manufactured to be formed from the full thickness of the wafer, as opposed to just on the surface. And a single device tends to be formed of many repeating blocks, since you do need fine details still to get the performance (And a lot of the challenge is making sure each of those blocks is as similar as possible, so you don't get hot-spots). (silicon lithography is more of a broad range of processes: digital vs analog vs power vs MEMS all have very different needs, and so a process designed for one will look very different to another, and one designed to allow multiple will be making compromises)
- namibj 2y ago
  The "full thickness" aspect is commonly called "vertical" when talking about the devices.

1970-01-01 2y ago

Very early EVs (GM's EV1 and then a few other GMs) used banks and banks of water cooled IGBTs to spin the motor.

NikkiA 2y ago

The Mercury Arc rectifiers that powered the earliest electric locomotives were prettier to observe though: https://www.youtube.com/watch?v=whF80m6oQZ8 https://www.youtube.com/watch?v=whF80m6oQZ8