For inverter nerds: 400 to 800 boost?

[hobbit]

So I've been pondering how some of the 800V cars, notably Ioniq, do a boost
trick with the motor to raise 400V charging up to the native battery voltage.
I've known for a while that boost tricks with motor windings are a thing,
and had a little proof-of-concept hack and page to detail low-speed regen,
but in that case the voltage is coming from the spinning motor itself.

For this, though, the motor is stopped and we need to keep our boosted voltage
away from the incoming charge leads. So this landed on my kitchen whiteboard...

Isolate the battery and one leg of the inverter behind a switchable diode. For
"boost" mode charging, open the switch. Bring one of the non-isolated legs
to the charging voltage by turning on its high-side transistor [under where
it says "400"], and apply modulation to the low side of the isolated leg.
A field builds in two of the motor windings, and then when the low leg is
"let go", flyback pulses flow through the isolated leg's upper embedded body
diode, allowing current to flow to the battery along the dotted-line path.

For 800V charging and normal driving, close the diode switch and go back to
regular 3-phase inverter configuration. [Third leg not shown here]

Obviously the reality will be significantly more complex, but I think I'm
finally discerning the principle. It is a very cute hack, and I understand
why it might be a bit power-limited. Perhaps the third motor winding is
involved too, and there's probably some thought toward preventing the motor
from exerting any torque against the drivetrain which is presumably captured
by the parking pawl.

Anyone know more about this?

_H*

 

[bwilson4web]

It pretty well parallels my limited knowledge of switching regulator circuits. But one unknown, what happens to any permanent magnets? For induction motors, no problem.

I've been turning over in my mind to feed 240 VAC from my NEMA 14-50 into a voltage doubler. Then mastering the fast DC charging handshake. Both the Tesla and BMW are L2 limited to ~30 A, 7.2 kW. But if I can spoof the DC fast charging protocol(s), I could fast DC charge at 9.6 kW, ~33% faster:

• 1 hr charging -> 45 min.

Locally, we have an 80 A , L2 charger. Doubling and DC converting and I'd be looking at a maximum 16.6 kW on 208 VAC to 19.2 kW. I would probably have to handle up to 100 W waste heat, assuming 95% efficiency.

Bob Wilson

 

[hobbit]

Might want a look at my commentary on DCFC and just how stupidly complex the
network protocol actually is. There may come to be available modules on the market
where someone else has black-boxed all those details into an easily usable widget,
but other than trying to order comm boards from a DCFC manufacturer [and likely
being told a hard no] I know of no such at the moment.

I'll note that the apparent rejection of NACS in that is a little out of date, but that's orthogonal
to the networking. In fact Tesla was probably pretty pissed off that they had to implement
it at all, but here we are.

_H*

 

[bwilson4web]

My goal is to bypass the 31-32 A limits of the embedded AC-to-DC controllers in the Tesla and BMW. By providing a compatible DC source, I can run closer to grid side, wire limits. A little 'glue logic' and voila.

When I self-install the Tesla CCS-1 compatible, socket controller and wiring kit, I'll look at tapping the control line and 12 V bus. I appreciate the heads up on the RF carrier. I was hoping it was traditional CANbus and not just picking up EM leakage.

Speculation, I suspect the original (in my case, current) NACS protocol has one or more CRC/codes embedded required from the DC source to enable connectivity. The upgraded socket controller would necessarily have software/hardware to 'denature' these codes allowing CCS-1 protocols to work. That will be my approach, the CCS-1 protocol, since I have both NACS and CCS-1 sockets in my EVs.

Flogging with a wet noodle, I didn't see a timestamp or version in your article. Old man eyes.

Bob Wilson

 

[hobbit]

To pull this back on track, I found the relevant patent linked on some other forum. It's actually a lot simpler
than I thought, and possibly a little more elegant as it likely accomodates more boost power and obviates
the possible motor-torque problem as a side effect. The key is access to the motor's neutral point between
its windings, which I assumed was NOT available. Then, all three windings can be driven to boost in parallel.

So charging either goes directly through R1, or "does boost" through R2 and winding flyback.
The patent's text clarifies

The controller can control the second relay to be short-circuited, boost the
voltage of the DC power of the EVSE through duty control of the switching
elements S1 to S6 using inductance according to coils constituting the motor
and booster circuits realized by the switching elements S1 to S6 of the
inverter 13 and apply the the battery.​

and later

Various voltage and current control methods performed through the controller
are disclosed in KR-<various> applied by the Applicant and methods of controlling
switching elements in boost converter topology are known in the art, ...​

So there we more or less have it. Much more text is devoted to making sure relays don't close
across high voltage differentials or open under load.

_H*