This is starting to sound like an MU2 (maybe all TPE) engine start.

Two 24v batteries with the option to start with batteries in parallel or series. Never knew there were starter gens on a pt6 that could support a 48v+ start.

The doubled voltage is only on the dedicated circuit to the starter, controlled by the SGC. The avionics, lights, etc. are always on the usual 24-28v.

De-ice for the prop, inlet and elevator horns also share the standard 28v from the starter/generator. For the wings and tail de-ice, the control circuitry is also on the 28v system, but the power for heating is totally separate, powered by a dedicated alternator that is unregulated, so voltage varies with RPM. There's no way to use that alternator as a backup for the regular 28v system, and with so much battery, no need to.

JMB got some breathless coverage at Sun 'n Fun
https://youtu.be/MIdWyJsSbO8
I passed on taking my plane to SnF, so the plane they had on display is older but a really nice example, now owned by the shop who helped build it and have maintained it since.

If the battery controlled the voltage regulator, you wouldn't have so much concern about the battery just switching offline. The alternator could be driven with enough field to deliver the desired current into the batteries, rather than just being a straight 28V regulated output.

But that would mean dropping the bus voltage below the nominal 28v while the batteries were charging. This would likely be better than causing the batteries to disconnect but I don't think it's a very good solution and it would involve a much more significant change to the aircraft's electrical system than just altering the battery type. That said, IIRC the Earth X requires the voltage regulators to be adjusted to something higher than 28v in order to charge the batteries. So maybe it would only be necessary to drop the voltage down to 28v in order to prevent excess charging current.

But it seems to me that Earth-X could/should incorporate charge limiting in their BMS in a way that doesn't require disconnecting completely. Then again, to provide the full benefit of a battery on the bus the charge limiter would need to maintain a fairly low impedance to the bus and that's somewhat opposite of the normal concept of charge limiting but I know it can done in a way that's effective for short duration transients.

But that would mean dropping the bus voltage below the nominal 28v while the batteries were charging. This would likely be better than causing the batteries to disconnect but I don't think it's a very good solution and it would involve a much more significant change to the aircraft's electrical system than just altering the battery type. That said, IIRC the Earth X requires the voltage regulators to be adjusted to something higher than 28v in order to charge the batteries. So maybe it would only be necessary to drop the voltage down to 28v in order to prevent excess charging current.

But it seems to me that Earth-X could/should incorporate charge limiting in their BMS in a way that doesn't require disconnecting completely. Then again, to provide the full benefit of a battery on the bus the charge limiter would need to maintain a fairly low impedance to the bus and that's somewhat opposite of the normal concept of charge limiting but I know it can done in a way that's effective for short duration transients.

My home LiPo batteries have BMSs that incorporate a charge limiting circuit. In the BMS configuration you set both a charge and discharge limit. They then have MOSFETs that will open the battery if those limits are exceeded. Since MOSFETs are inherently directional (they only conduct in one direction) turning off the charge circuit does not remove the battery from the circuit for discharging or vice versa. There is another circuit in parallel with the charge MOSFETs that is the “current limit” circuit. Once the charge MOSFETs turn off due to charge current limit (as opposed to other reasons such as cell or pack voltage limits or temperature limit etc) then it will turn on and it has a choke in it which allows the BMS to limit charging current to a low value (20A for the ones I have but it varies from model to model) until the next time the BMS sees a discharge at which point it will try the regular unregulated charging MOSFETs again.

But that would mean dropping the bus voltage below the nominal 28v while the batteries were charging. This would likely be better than causing the batteries to disconnect but I don't think it's a very good solution and it would involve a much more significant change to the aircraft's electrical system than just altering the battery type. That said, IIRC the Earth X requires the voltage regulators to be adjusted to something higher than 28v in order to charge the batteries. So maybe it would only be necessary to drop the voltage down to 28v in order to prevent excess charging current.

But it seems to me that Earth-X could/should incorporate charge limiting in their BMS in a way that doesn't require disconnecting completely. Then again, to provide the full benefit of a battery on the bus the charge limiter would need to maintain a fairly low impedance to the bus and that's somewhat opposite of the normal concept of charge limiting but I know it can done in a way that's effective for short duration transients.

The more I think about it the more I like having the charging current limit the bus voltage. That would require something to monitor the charge current (a capability already in the BMS) and a "special" VR that allows an external input to lower the setpoint down to something like 26v. None of that sounds technically complex but from the perspective of simplifying the installation I'd want to avoid extra wires coming from the batteries and either incorporate the current sensor inside the VR or use an external shunt wired to the VR.

Given the minimum lithium cell voltage I'm pretty sure there couldn't be much if any charging current if the bus voltage was at or below 26v. Note that there is a scenario where the charging current has to be considerably less than what's normally allowed. This is when the battery has been overdischarged, either by cutting the output off a little too late or when the battery was fully discharged and then self discharges for a while.

Could use a stock VR, but bring the battery voltage signal and the current signal together at the VR input. For example, suppose we put 1kOhm from the battery to the VR input, then draw a current from that node from the current sense amplifier where 1mA = 1V reduction in bus voltage.

This. They pair up in series for 52v starts, then it's back to all parallel.

Could use a stock VR, but bring the battery voltage signal and the current signal together at the VR input. For example, suppose we put 1kOhm from the battery to the VR input, then draw a current from that node from the current sense amplifier where 1mA = 1V reduction in bus voltage.

Or you could let the VRs work exactly the way they always have, put a standard BMS on the batteries as I described above, and stick a capacitor on the bus to absorb transients. That’s the simplest solution to me.

That's how the EarthX BMS works as well, if the charging current is exceeded the battery is cut off from charging but can still discharge. That's what they say, although exactly how is beyond my electrical pay grade.

The original plan was to just replace the Evo's pair of conventional batteries with a pair of lithium batteries. At the time, TrueBlue was the only 24v option, and they had worked in other Evos, but we discovered behavior that was incompatible with our more modern electronics. This behavior was both undocumented and denied by TB, which cost us many frustrating months. By the time TB admitted that yes, they'd known all along that was by-design behavior for their batteries, EarthX came out with their first 24v battery and we finally had an alternative. The only challenge with the EarthX battery was the charging current limitation, and after some time working on current limiting circuitry, we settled on doubling up to 4 x 24v batteries as the simplest solution.

As you can imagine, after our history with TB, we discussed what we were planning with EarthX in great detail. They were initially hesitant about the parallel/series setup, and I believe it was new territory for them. Now, their website has notes in red about usage in parallel, and they recently introduced a "for turbines" version of their 24v battery with the charging current limit raised to 300 amps. I suspect we had a hand in both. I would certainly have just used 2 of their turbine batteries, had they existed then, in lieu of our 4-batteries solution, and it would have saved a couple of pounds, but at this point I'm happy to stick with what we have.