M.M., my bad; I should have said common denominators since there are obviously several scenarios where ICE is being called for. I was just hoping your poll and software/engineering knowledge could help sort out all these experiences into some kind of logical groupings.
Got it, I probably should have been able to figure it out from context.
So I’ve spent some time experimenting and have managed to reliably replicate one of the regen-braking ICE startup scenarios, although the exact conditions are surprisingly specific.
From my house, there is about a 4-block stretch of gradual downhill slope with two flat areas, then a couple of stop signs within 30 feet of each other (it’s an odd intersection), and another block or so of steep downhill with a stop sign at the bottom. So basically flat block, gradual downhill block, flat block, slightly steeper downhill block, stop sign, steep downhill block, stop sign.
If I drive normally through there on a full battery, nothing happens. I’ve done it dozens of times without an ICE start. If I drive
very carefully with efficiency in mind, still nothing happens. If I slam on the brakes at or between any of the stop signs in there, nothing happens (I was trying to replicate your hard-stop ICE start possibility).
But if I’m sort of hypermiling and put the car in neutral for the first flat stretch, then in gear and regen, then in neutral again, then regen again, the ICE will start. I haven’t done enough experiments to say definitively, but it looks like one shift into neutral isn’t sufficient--it seems to take two. It isn’t just a matter of the amount of regen going into the battery, otherwise it would start on the steep downhill nearly every time, since there’s definitely more energy available on that slope than is being consumed moving away from the stop sign.
I’m just guessing randomly at this point, but something about maybe the gear change causing the regen subroutine to put a spike of energy into the battery, which triggers some sort of over voltage protection, could be a culprit. That would actually align with why a hard brake might also cause it to happen, although I couldn’t replicate that myself.
One possible factor that I haven’t tested definitively is that when plugged in my car sometimes shows 99% on the app when full, and sometimes shows 100%. Which seems completely random and I had chalked it up to variables in the calculation, but it’s possible that the car is making decisions based on this slight difference in calculation, which could cause it to be more sensitive to regen braking at 100% than at 99% (although it’s weird that they didn’t leave more buffer than that before it starts getting worried about overcharge; I think the volt has maybe 1kWh of overage before it goes into its hill-braking-full-battery routine).
Other things I noticed:
Once started, the ICE will run for about 5 minutes (either until it’s warmed up, as expected, or possibly there’s an ICE braking timeout). I also confirmed the bug that in this particular state the car doesn’t turn off the white dashed line on the energy gauge even when it’s back in EV mode. I drove around for a while to make sure, the display is definitely stuck in the wrong state.
When the ICE is running like this, if you put it in 4 regen chevrons the indicator will flash and it will go down to 3 chevrons. This seems to confirm that it’s in a reduced-regen mode; perhaps (again, pure speculation) three chevrons worth is as much as it’s capable of dumping with the ICE. This happened even after climbing a couple steep hills, so is tied to the mode, not battery state of charge.
In this mode, you can hear the load on the ICE change between when you’re coasting or braking and when your foot is on the accelerator, although in either mode the RPM never goes above idle. I would guess this is because when you’re coasting, it's disconnecting the ICE’s generator entirely, so it isn’t trying to charge an already-charged battery, but once you put a load on the bus it reconnects it so that the idle energy goes to the wheels. That’s roughly what it sounds like although it’s possible I’m interpreting sounds incorrectly.
Also of interest in this mode, at exactly 14mph the car turns something either on or off related to regeneration. If I start regenerating at, say, 18mph (3 chevrons, the max in this state), just before the speedometer shows 13mph there is an audible click from the engine compartment and a slight but noticeable reduction in the regenerative braking force. I have no idea what this is; it sounds a bit like a relay clicking (maybe the ICE’s generator being mechanically disconnected at speeds where friction braking should be taking over?), but could be other things.
Oddly on that 14mph click, there is no hysteresis built into whatever is controlling it; if you’re on just the right downhill slope that the car will be hovering around 14.0mph depending on whether whatever the click is controlling is increasing regen or not, it will click on and off quite quickly.
Also, it’s hard to tell for certain, but it appears that reduced regenerative braking is visible on the energy indicator. While braking at some stop signs near my house, the amount the line will go below 0 seems to change depending on how full the battery is. Maybe somebody else who has more consistent hills near home can eyeball this and chime in.
Other unrelated thoughts:
It was noted elsewhere here that the mechanical transmission for highway driving is not a CVT, it’s a fixed-ratio gearbox. That being so, it seems quite unlikely that the car is capable of using the mechanical transmission for engine braking since the gearbox is designed for efficient, relatively-low-RPM operation at highway speeds. That would narrow the ICE regen down to the seemingly-nuts version of putting a load on the ICE by dumping power into the generator.
Summarizing, we appear to have one state of engine exercise that happens without any user indication after around a month without running the ICE (which is expected although should have had a user indication), a second ICE run state that can kick in shortly after starting driving on a gradual downhill on a full battery, but only in certain situations (probably related to a full-battery-long-hill-ICE braking state, but being started erratically by a bug), and some additional state related to sitting in park for several minutes.
Grrrr, I thought computers were supposed to be consistent!
They are, but with a complex state machine that has a lot of variables and unpredictable analog inputs it can be hard to suss out what exact condition or set of conditions caused something to happen.
