Do rear brakes activate always, despite regenerative braking?

Oh, absolutely, all of that is correct. During normal driving (power flow engine to wheels), the power flows partly on a mechanical path (engine to planet carrier to ring gear to output) and partly on an electrical path involving the planet carrier, sun gear, MG1, and MG2. That is nothing more or less than the way our transmission happens to work, as given a great explanation in Niels Blaauw's video.

During engine braking, power flows through the transmission the other way, and also over the same two paths. Neither path is optional: the torque imposed on the ring gear by the planet carrier is always 72% of the engine's torque at that instant, and the torque imposed on the sun gear by the planet carrier is always 28% of the engine's torque. (The amount of power following each path is, of course, that torque times the rpm, and there are even instants where the sun gear's RPM is zero so the power flow on that path happens to be zero despite the torque). If you break the electrical path between MG1 and MG2 (which is exactly the only thing the "neutral" shifter selection does), the mechanical path won't carry power either.

My point is, all of that is inside detail of how our transmission works, and what that accomplishes is no different from what any other transmission accomplishes, however it may work inside. Power goes from the shaft connected to the engine to the shafts connected to the wheels, or, for engine braking, it goes the other way.

Now, there is one thing our hybrid transmission can do that other transmissions can't: the power on its input and output shafts don't always have to be equal. There's allowed to be a net difference, positive or negative, which can (for short periods) come from or go into a battery. That's a neat trick, and when it's coming from the battery to help accelerate you, that's battery assist, and when it's going into the battery to help slow you, that's regen braking.

But when you're doing nothing but ordinary engine braking, with no net electrical output from the transmission being recovered into a battery, and only the normal self-contained electrical flow between components of the transmission itself because that's just how our transmission works, there's no need to call that anything exotic. That's just engine braking. Three transmissions all built three different ways inside would be doing the exact same thing in that case, no matter the details inside.

Again, revving the engine higher for the same road speed, to maximize the braking effect, is no more or less than choosing a lower effective gear ratio from the transmission—where 'lower' is as we usually mean it in normal driving, reducing the engine rpm to deliver more torque to the wheels. The same 'low' gearing during engine braking spins the engine faster and with lower torque relative to the wheels, which both maximizes the braking torque available from the engine, and presents a multiple of that to the wheels and road.

A driver with a manual transmission will grab the stick and downshift to do that. Typical automatics can be downshifted by the driver, and some will downshift on their own for the same purpose. A belt-and-cone CVT will move the belt to a wider spot on the output cone and a narrower spot on the input cone. Our CVT will do it by changing the timing of gate pulses to the MG1 and MG2 IPM transistors. They are all doing it different ways, but what they are doing has not changed.

We may well smile at the clever way ours does it, while keeping in mind that what we're smiling at is just the cleverness of how a CVT has been built out of two controllable motors and a planetary gear in the first place. Once we've acknowledged that, what it is doing during engine braking isn't anything extra clever beyond that, and isn't anything non-hybrid transmissions don't also do in their own ways.

And the cleverness we are smiling at is largely John Godfrey Parry Thomas's from 1908.

These days, it's probably safer to say "not much above 40% in combustion mode".

But of course that 40-ish percent for the combustion-engine efficiency doesn't generalize to what happens in engine braking. Looking just at the compression and expansion in the cylinders, at high rpm, engineers do model them as adiabatic, and in fact isentropic. (You can't defeat entropy, but there are isentropic ideal processes.) What's happening in our real engines falls short, of course, of the ideal models, but not by nearly as much as the combustion efficiency figure.

(Wait, I should have put that another way: even the ideal models put a limit on how efficient the combustion engine can be—and then our real engines fall short even of that, though recent ones get mighty close. But that's not the case for the expansion and compression; the ideal model really is isentropic, and the only inefficiency is how far short we fall of that.)

The remainder of that long summary looks pretty good to me, with a note that the conditions maximizing the vacuum/expansion effects are the ones that minimize the compression and exhaust-pumping effects (because there is so little of anything in there to compress or to pump), and vice versa. While either set of conditions can be used to achieve an engine-braking effect (and even more exotic combinations as the Ford patent proposes), our car still uses the closed-throttle approach where the vacuum/expansion effects predominate, and one reason at least for that choice appears to be the need to preserve heat and OSC in the catalytic converter.

 

Sorry vvillovv for confusion over ‘which engine braking’. I was thinking (out loud) about braking in a modern Toyota hybrid. e-CVT, so no belts, just the planetary gearset. It does become confusing, as the car will do it without any input from the driver on the brake pedal or gear selector. No doubt it happens sooner, more often and/or more strongly when the driver uses cruise control and/or selects ‘B’ (Prius) or ‘S (Rav4). In the rav4, there are different degrees (-/+ under S). When the driver selects these, the car computers take the request on board, then do whatever they like. I have not logged relevant PIDs under all these scenarios.

If I have followed correctly, ChapmanF prefers to restrict ‘regenerative braking’ (which probably includes regenerative coasting when the driver does not touch the brake pedal) to the circumstances under which electricity generated by MG2 (and/or MGR in an AWD) flows (at least in part) to the battery for recharging. I think he also prefers to restrict ‘engine braking’ (whatever the elements that contribute to it) to negative torque transmitted from ICE via the transmission to the wheels. If that is right, those who agree will need another term for the negative torque applied to wheels when MG2 (and/or MGR in an AWD) generate electricity while the battery is deemed “full”. This electricity is dissipated by powering MG1 as a motor to rev the engine and increase the amount of mechanical engine braking. To me this is the clever part, a double return on investment (see the nomogram for more fun).

Anyhow, this may cease to be interesting long before I understand it fully. I want one of those new-fangled non-entropic cars. Preferably one confirmed by the EPA, not just advertised by a car maker.

 

No, that doesn't represent what I was saying. Leaving aside AWD and MGR for simplicity, there are still two MGs that are integral parts of the transmission, MG1 and MG2, and electrical power flows between them at all times when the transmission is carrying any power at all. If the electrical path between them is ever interrupted, the mechanical transfer of power through the gears also ceases. (Edit: I have overstated this case mildly. *)

Whenever the power being generated in MG1 and consumed in MG2 (or vice versa) are equal—that is, there is no net difference—the transmission is operating as a self-contained, simple transmission, and doing nothing that any other conventional transmission doesn't do.

That leaves two other cases of interest: a net inflow of electrical power to the transmission (either MG generates less than the other consumes, with the difference flowing in from the battery), or a net outflow of electrical power from the transmission (either MG generates more than the other consumes, with the difference flowing out to charge the battery). You'll notice I'm not specifying which MG is generating and which is consuming at any given time, because the transmission can operate in modes where the answer is mildly surprising. But there are definitely three important cases: net flow from battery to transmission, net flow from transmission to battery, and no net battery flow (simple transmission behavior).

If you have not watched the excellent video prepared by Niels Blaauw, he has done a very thoughtful job of developing all the transmission's modes of operation up from basic principles,

Also, Ichabod's online interactive nomogram was unavailable for a long time, as support for Flash faded away, but was at last reimplemented for modern browsers.

Again, I expressly do not make the distinction you think I am making here, and so there is no need for any more than one term for engine braking. MG1 and MG2 are both integral parts of the transmission, and whenever negative torque appears at the wheels because the transmission has delivered the corresponding power to the engine to be dissipated there, that is engine braking. Because all flow of power through the transmission requires both MG2 and MG1 to participate, no further division of this case is meaningful.

Whenever negative torque appears at the wheels with the corresponding power being delivered to the battery, that is regenerative braking. The power delivered to the battery is simply the net excess that either MG is producing over what the other is consuming at that moment, and to recognize regenerative braking it's not necessary to pry any further into their internal affairs.

I will note that not every time net electrical power flows to the battery is a case of regenerative braking. If the torque at the wheels is not negative, then the car is driving or coasting, and the power management ECU has simply decided the battery needs some charging, and has upped the engine torque slightly above the driving need, so as to use the net excess to charge.

It is also entirely possible for regenerative braking and engine braking to be going on at the same time. There is simply a certain negative torque at the wheels, with some of the corresponding power going electrically to the battery as a net outflow from the transmission, and the balance being used to spin the engine. The car does seem to smoothly crossfade from one to the other as the battery crosses a roughly 70% actual state of charge threshold.

* the part I overstated: what's indispensable for the transmission to carry power from the input to the output shaft or vice versa is for MG1 to be electrically active. That can include cases where no power's being generated or consumed by MG2, in which case the electrical power at MG1 is all a net inflow/outflow with the battery.

 

Sorry if I have misunderstood your ideas ChapmanF. Wow those ideas are complex. My logs of Toyota PiDs in real life seem much simpler; but the way I drive does not cover all possible scenarios, by design.

Is there any experimental evidence for regenerative braking by MG1?

 

We seem to have different ideas of what complexity is. We've got a simple one-input-shaft, one-output-to-final-drive transmission like any other, its inner details being its own business, but with one feature other such transmissions don't have, which is the ability for its input-shaft and output-shaft power to be sometimes unequal, with the net difference going from/to a battery.

To me, keeping that picture of the forest in mind isn't especially complex, and can help in avoiding some of the rabbit holes (like thinking two names for engine braking are needed) that can be fallen into when trying to pin its behavior on individual trees.

Have you watched the Blaauw video yet? Maybe you have, but the way you're still writing as if the above are my personal ideas suggests maybe you haven't.

It's possible you have enough in your logs anyway to try something that might be as good as a lot of words in this thread.

Do you have some instances of engine braking captured? If so, you can take one or more of those examples, start with the PIDs that you logged, and flesh out the complete description of what the transmission was doing just then: the rpm, torque, and power at each component, and the overall ratio the transmission is providing. As long as you have enough of those items in your logs, you can complete the rest from the nomogram (and its invariant that the torque at the engine is accounted for 28% by MG1 and 72% by the mechanical path) and from the basic facts in play like power being rpm ✕ torque. Then we can see if we agree on the results, and whether that implies a behavior different from a conventional transmission in the same circumstances, and how many names for engine braking might be needed.

 

Not sure the "ideas" (fundamentals and concepts) are all that complex but visualizing it for the first time can be difficult. I like the Weber State Prius transaxle videos starting with their overview.

In the end its not necessary to understand the complex explanations we are offered here. If the plane flies it must be possible. Which is evidence enough. Same goes for Toyota transaxles and regen.

 

So I take it there is no experimental evidence for regenerative braking by MG1? I have not seen such evidence either (though opinion seems to be divided here on Prius Chat).

My take on the matter:

"Negative torque from MG1 into the transmission seems only to deliver power along the path of lower resistance to reduce ICE rpm. The computers do not seem to allow the condition where (a) MG1 has +ve rpm and -ve torque (b) ICE is locked to deliver that torque from MG1 to the wheels, and (c) the wheels reduce rpm in response."

But this is just speculation. It is consistent with the observations, and I am not sure I would want to drive while computers allowed the condition mentioned, but it is still speculation.

 

Can you explain a little more your interest in that particular question? I don't know how divided opinion on PriusChat is, as you might so far be the first person I've seen express an opinion on it.

If I were asked for my opinion on it, I'd probably have to assume the question meant something specific like "can nonzero MG1 torque be observed during regenerative braking of the car?", and then I'd probably say something like this:

• Any time engine braking is not also involved, I would not expect to see MG1 torques during regenerative braking of the car very often, or for longer than a moment at a time. My reasoning: if engine braking is not also involved, then you'd like to keep engine torque at zero to maximize the power you can collect for regen, and as MG1 torque is always 30/108 of engine torque, if you have a zero engine torque target, you also have a zero MG1 torque target.
• In the high state-of-charge transition region where both some regenerative braking and some engine braking coincide, there will be an MG1 torque. This will represent an electrical power flow into MG1, so that the regenerative braking power is only the net of what MG2 generated after this is subtracted.
• At certain moments during regenerative braking, it is possible the ECU will attend to some other need, such as starting or stopping the engine because of crossing the engine-must-run road speed, or because of heating demand or catalyst temperature or its other multifarious reasons.
  
  The natural way for the car to start the engine, while the road speed is high enough, is via regenerative braking of MG1. Yes, I said 'of' there, and not 'by'. But if regenerative braking of the car is, at that time, happening, the power out of MG1 for that moment adds to it. That may either be a momentary increase in the regen power going to the battery, or the ECU may momentarily decrease the regeneration at MG2 so the net power stays the same and the driver does not feel a brief increase in braking.
  
  Likewise, if the ECU elects to bring the engine to a stop, it can do so with MG1 torque in the − direction, and that will take a brief bit of power. The net of the power generated by MG2, after this bit is subtracted, is what remains to go into the battery.
• For most practical purposes, poking into those details is not needed to see when the car is doing regen braking. The transmission might be doing more than one thing at a time, but as long as it is applying − torque at the wheels and producing a net outflow of electrical power, the car is being regeneratively braked. That's what I was getting at in post #43:

I should add that regenerative braking of either MG1 or MG2 happens lots and lots of the time, and not always anything to do with regenerative braking of the car. Regenerative braking of MG1 happens for long stretches of normal driving (except "heretical" overdrive), and to start the engine if the road speed is high enough, or to stop the engine if the road speed is low enough. Regenerative braking of MG2 happens for long stretches during overdrive cruise, and when engine braking, and (last but not least!) when actually regen braking the car.

Turning back to engine braking, I had the chance yesterday to gather some of that kind of data. I'm in Indiana (and not even southern Indiana where there's terrain), so it's a little hard to find adequate hills, but a power outage after lunch left me with little else to do and it was a beautiful day for a drive, so I went and found one I remembered not too far away. And by the time I returned, the power was back on.

Even this hill was so short I had to start with a force-charge to 70% at the top in order to observe the car's transition from regen to engine braking on the way down. Indiana....

I did not touch the brake pedal, but had the cruise control at its lowest setting, a dozen MPH below the road speed, so the cruise control was hard at work to get the speed down.

The work proceeded all by regeneration into the battery until the state of charge reached 77.2%. At that point, light engine braking began, and increased while regeneration decreased, until at state of charge 80.0% it was all engine braking and no regen.

At 77.2% SoC just before that transition began, road speed was 37 MPH and increasing; the power being regenerated was not quite equal to the input from gravity. Speed increased to 47 MPH during the transition and was back down to 36 MPH by the time it was all engine braking at 80.0% SoC. (The hill had already begun to flatten out by that point. Indiana....)

A couple incidental notes:

• The power management control ECU has PIDs for "Regenerative Brake Torq" and "Rqst Regenerative Brake Torq". Both remained zero throughout, both at first when regenerative braking was employed, and later when engine braking was employed. It seems these PIDs are used only when regen is requested by a message from the brake ECU, and not when employed by the cruise control (which is implemented within the power management control ECU itself).
• The power management control ECU also has a "Request Power" PID, giving the kilowatts the ECU has requested from the ECM when it wants (positive) engine power. But this PID remains zero during engine braking; it does not report the kilowatts the engine is being asked to soak up.* But we do have enough other PIDs to complete the picture anyway.

Just before the transition, the braking power is easy to calculate. MG2 was doing 4606 RPM at 27.88 Nm, so it was subtracting 13.45 kW from gravity's input. Of those 13.45 kW, what arrived at the battery was 45.06 amps at 262 volts, or 11.81 kW.

1.64 kilowatts have gone missing. The car uses around half a kW just being awake, and the remaining 1.1 kW or so, around 10% of the power being handled, would not be too astonishing as losses along the motor to inverter to battery path. Also, we have to expect a certain fuzz in the calculations involving different PIDs because Techstream is polling for them, and not getting answers that all come from the same instant.

After the transition, at the point of strongest engine braking I saw, we can directly compute the portion of the power that is taking the MG2 to MG1 path. MG2 is doing 4331 RPM at -20.50 Nm, so it is subtracting 9.30 kW from the gravity input. None of that is going to the battery, and what's arriving at MG1 is 9974 RPM at 7.37 Nm, or 7.70 kW. Again there's a ~ 1.6 kW gap to explain with a combination of the car's usage, electrical losses, and PID polling fuzz.

That 7.70 kW is not all of the power the engine is burning off, but only the portion that came over the MG2 to MG1 path. But we also have enough data to compute what followed the mechanical path.

The MG1 torque was 7.37 Nm and is always 30/108 of the torque on the engine, so that was was 26.5 Nm at 3936 engine RPM, or 10.94 kW being soaked up.

MG2's RPM was 4331 and a Gen 3 has a 58:22 reduction between MG2 and the PSD ring, so the PSD ring RPM was 1643. The PSD ring sees 78/30 of MG1's torque, so that was 19.16 Nm, and at 1643 RPM that's 3.30 kW, which is what was left of the gravity input power after MG2 had subtracted 9.30 kW from it.

So the gravity power input being handled was 12.60 kW, of which MG2 removed 9.30 kW to follow the electrical path, leaving 3.30 kW to follow the mechanical path.

The engine was soaking up 10.94 of those 12.60 total kW, leaving 1.66 kW to explain with the car's power usage, incidental losses, and PID polling fuzz.

The engine was turning 3936 RPM and Gen 3 will allow up to 5300, so there was surely more engine braking capacity available if I'd had the hill to show it. Indiana....

Because the PSD ring was doing 1643 RPM and the engine turning 3936, the CVT had an effective gear ratio of 2.396 at that moment.

In my last stick-shift passenger car, 1st gear was 3.416 and 2nd was 1.842. That car also had a higher final drive ratio, 4.105, than Gen 3 Prius 3.267. So that car's 2nd gear was 7.561 to the wheels, and the Prius in this example was effectively 7.828.

In that other car, I probably would have downshifted to 2nd on that hill at that speed, and the engine braking would have sounded and felt pretty much the same. While our transmission is built differently inside, it isn't behaving any differently in this example than a conventional transmission would.

* I think RGeB presented at least one graph showing an Engine Torque PID that did show negative values during engine braking. That may have been a PID available from the ECM. But I was logging only PIDs from the power management control ECU, and its Request Power PID doesn't reflect that, and just sits at zero instead.