Why regenerative braking belongs on the brake pedal

[Rocky_H]

I really think pedal position is mapped to a target speed, not to a target torque. Of course, this is just my impression. I could be full of beans.

I agree with that. It does seem to be correlated to speed requested.

 

[scaesare]

I agree with that. It does seem to be correlated to speed requested.

If I hold my foot steady and road starts to incline upwards the car slows. It does not vary power/torque to maintain a given speed at a specific pedal position.

 

[Rocky_H]

If I hold my foot steady and road starts to incline upwards the car slows. It does not vary power/torque to maintain a given speed at a specific pedal position.

You're not finishing your thought. It slows....a little bit, but then does increase the power to keep it from slowing too much. It won't let the car slow down to a stop on the uphill. There is a little +/- margin range there. Same with a downhill. It will speed up a little bit, but then it will start applying regen to hold the car back, so it doesn't speed up too much.

 

[insaneoctane]

This discussion feels a bit circular. The last comment about how the hill shows the flaw in the target speed theory. The accelerator just selects motor load, but with regen, it is a bit more complicated than an ICE analogy!

 

[scaesare]

You're not finishing your thought. It slows....a little bit, but then does increase the power to keep it from slowing too much. It won't let the car slow down to a stop on the uphill. There is a little +/- margin range there. Same with a downhill. It will speed up a little bit, but then it will start applying regen to hold the car back, so it doesn't speed up too much.

I tested this on the way home yesterday. This isn't just a little error margin. A target speed is not maintained with a static pedal position... a constant power (and thus torque) is.

For reference, on my commute home, there's a section where I can set cruise control at 70 and leave it that way for 10+ miles while not having to disengage often if at all. It has some flat, and some gradual hilly sections.

For cruise control to maintain a set speed, it has to vary the power from near 0 on downhill sections to close to 50KW on some uphill sections. The flat sections are right at about 20KW draw.

If a static target speed was targeted by a pedal position, we'd see similar power draw fluctuation as the car attempted to maintain those speeds. I can wedge my foot between accelerator and console side kick-panel, and keep it right at about 20KW for as long as I want (or until my foot falls asleep, lol).

If I do this on the uphill sides I'll slow to less than 40, all the while the power meter stays at a constant 20KW. On the downhill sides I'll accelerate past 85 before I have to abort, as it's continuing to gain speed.

So I'm pretty sure a given pedal position targets a given power/torque, not a given speed.

 

[daniel]

Well, I did say I was as likely to be wrong as right.

 

[Ulmo]

On a thread about the Leaf, there was a complaint that Nissan got it wrong then they had the default regen setting when coming off the throttle to be very light regen. Actually that is the correct way to do things for several reasons. Reasons which non-automotive engineers and non-racing drivers generally would not be aware of. Most of the regen is done on the brake pedal on the Leaf, Volt, EV1, Toyota RAV4 EV (the original one made by Toyota, not Tesla), Prius, VW e-Golf and many other EVs and hybrids. Read some of the history in my link below.

Tesla got it wrong when they put heavy regen on the throttle. It's one of several things they got wrong, but probably the most serious one. It's not stopping me for buying a Tesla or two, but worth pointing out. When I test drove a Model S, I liked it much better with Low Regen. I realize I'm probably in the minority on that, but I also realize most people are very likely ignorant about the reasons why.

Please read and learn: Why Regenerative Braking Belongs... On The Brake Pedal

Also be aware that Tesla itself is not monolithic on the issue. There are engineers inside Tesla who agree that regen belongs on the brake pedal, not the throttle.

Please educate yourself first before commenting.

Flames to /dev/null

Excellent post! I agree. I had many problems with the Tesla Model S regeneration on accelerator:

• My foot would get tired when I wanted to let the car coast but I had to keep my foot on the pedal to do so, nursing its speed in a herky jerky way to do so, further exacerbating my pains. My factory computer chip controlled Mercedes never ever did that to me, and was sublimely wonderful.
• Sometimes I would have to react quickly to road events, and massive slowing was not the reaction I was looking for, and it became dangerous, almost causing accidents.
• I did not know using brake pedal didn’t use regen on Model S; that explains why I lost energy so fast whenever I drove anywhere through normal driving. If regen was on the brakes where it’s supposed to be, this wouldn’t be an issue at all. Come to think of it, I believe brakes did apply regen on the Model S. You need to retest.

 

[Ulmo]

I tested this on the way home yesterday. This isn't just a little error margin. A target speed is not maintained with a static pedal position... a constant power (and thus torque) is.

For reference, on my commute home, there's a section where I can set cruise control at 70 and leave it that way for 10+ miles while not having to disengage often if at all. It has some flat, and some gradual hilly sections.

For cruise control to maintain a set speed, it has to vary the power from near 0 on downhill sections to close to 50KW on some uphill sections. The flat sections are right at about 20KW draw.

If a static target speed was targeted by a pedal position, we'd see similar power draw fluctuation as the car attempted to maintain those speeds. I can wedge my foot between accelerator and console side kick-panel, and keep it right at about 20KW for as long as I want (or until my foot falls asleep, lol).

If I do this on the uphill sides I'll slow to less than 40, all the while the power meter stays at a constant 20KW. On the downhill sides I'll accelerate past 85 before I have to abort, as it's continuing to gain speed.

So I'm pretty sure a given pedal position targets a given power/torque, not a given speed.

Yup. That’s what I remember from the Tesla, and I mostly hated it, but I started getting a little used to it. It was distracting and painful, though. I didn’t like it.

I think my Mercedes used a combo method, smoothing pedal input into constant speed and ramping out of that just a bit for hills, prompting me to input higher or lower pedal as the hills changed, but only slightly, and as soon as I did, the car would comply with my wishes. Absolutely wonderful to drive.

 

[Ulmo]

Same here.

Coasting is always more efficient because regen has losses both in taking the energy back into the battery pack and taking from the battery pack back into motion.

Bjorn did a video that showed how much energy he could save by going into neutral for hills as opposed to doing the regen-re-accelerate mess. It was a lot, and saved a lot of wear and tear, adding miles he could travel.

 

[Ulmo]

Unless you drove over an expansion joint or pavement irregularity while braking, then it felt as if the car lurched ahead--really poor brake feel. And the third generation was so bad they had to recall the cars to fix it. The only way in which regen on the brake pedal can be done smoothly is to have the friction brakes do a lot of the braking so that the transition isn't felt. It's a bad idea.

I disagree. I could do very-near-limousine stops in my Mercedes all day. The braking computer was that good. It was their SBC system, so they discontinued it due to cost.

 

[Ulmo]

GM did it very well 20 years ago in the EV1; Toyota a little later with the second generation Prius.

I still can’t fathom why they cancelled that car. I’ve read the official reasons and the conspiracy reasons, but none of them make sense!

 

[Ulmo]

One man's opinion, with the linked backup being...one man's opinion.

“One man’s opinion” is a mantra of the Marxists.

 

[Ulmo]

Regen does not belong on the brake pedal, it adds unneeded complexity and poor brake feel. Tesla got it right.

How is it right that when my right foot got tired, my car would come to a stop on a freeway when I gave it a rest? That’s downright stupid! Tesla got it wrong. They should open source their software, so we can fix it. It would require some validation before usage, of course.

 

[Ulmo]

This depends on road conditions. Coasting in an automatic uses gasoline as opposed to engine braking and coasting in an EV uses energy to counteract the counter electromotive force as opposed to regen. It's easier to just maintain speed if you wish to "coast".

Of course slowing down just to speed up again is wasteful so one must judge base of future road conditions (if a hill is coming up etc).

Coasting was managed by the computer in my Mercedes very well. It interpreted no pedal as coast, and throttled the engine and transmission gearing to a very good state for coasting, using very little energy, but just enough to stay sane, and be ready for the next input. It was almost impossible to trick the Mercedes computer into doing the wrong thing, and that was on some really weird situations, so I didn’t mind it at all.

 

[mongo]

How is it right that when my right foot got tired, my car would come to a stop on a freeway when I gave it a rest? That’s downright stupid! Tesla got it wrong. They should open source their software, so we can fix it. It would require some validation before usage, of course.

Huh? I think any car (with cruise control off) will come to a stop/crawl if you give your right foot a rest. (If you want to coast pop into neutral)

I grew on on manual transmissions and like accelerator regen.

 

[Ulmo]

How much of this problem is caused by regen on a Tesla being so weak a stopping force? Could it work better if there was a small bank of supercapacitors that could soak up more energy faster? The super caps could be tuned to handle every acceleration in the following events:

• Coming off a red light (when it turns green)
• Going onto a freeway
• Merging into a street
• Coming out of a curve

The super caps would then allow much greater regen power from the motors for braking purposes, and the brakes could regen far deeper into the brake press, and no one would need to know how much brake is too much for max regen. In downhill runs, the car would notice the trend (even map the route ahead), and start moving energy from both the supercaps and the motors into the batteries for long term charging (moving more energy out of the caps out of curves and using the caps for regen more while coming into curves, without any knowledge by the car of curves since it’d naturally do it per braking needs). When the supercaps and lesser extent the batteries were full and offered no regen room, the friction brakes would apply the force. Force needed would be calculated by the computer by the pedal amount, and the computer would get it in priority from the supercaps, the battery, and the friction brakes. A reserve in the caps and battery would be there to soak the difference in time between no available regen and friction application for a smooth transition during braking in a topped-out-charged situation.

 

[scaesare]

Yup. That’s what I remember from the Tesla, and I mostly hated it, but I started getting a little used to it. It was distracting and painful, though. I didn’t like it.

I think my Mercedes used a combo method, smoothing pedal input into constant speed and ramping out of that just a bit for hills, prompting me to input higher or lower pedal as the hills changed, but only slightly, and as soon as I did, the car would comply with my wishes. Absolutely wonderful to drive.

I've never once been either distracted by it or found it painful.

If I want constant speed I engage cruise control. I leave it on always (that survives restarts), and simply engage it as necessary... which for me is probably 80-90% of the time. Ionly typically don't use it when I'm in stop-n-go traffic... which is when constant speed isn't practical anyway.

 

[Mr X]

No.

 

[arnis]

Regen on accelerator is appropriate according to situation.
When driving aggressively, it is welcome.
When driving efficiently/smoothly, it is not.
Driver can adjust pedal position to lessen regen, but it is more complicated and requires extra brainpower.
The most important factor is the choice.
At almost no circumstances is friction brake with not maximum regen appropriate - that happens on Teslas with regen on LOW.
Multiple EV's offer sufficient regen adjustment (from non to max) and will also blend it if brake is used.
This is the best way. Except like I mentioned, add pressure sensitive paddles behind the steering wheel to adjust regen only.

I believe many other comfortable accelerate-decelerate methods can be developed.
For example active accelerator pedal (Merc has that already AFAIK, but it is suggestive according to nav data, when to decelerate) that gives feedback WHEN acceleration ends and where regen starts (like camera shutter button, but much softer).
So relaxing the foot without removing it from the pedal will result in coast. But lifting it will add regen.. smoothly.
For me, most manufacturers have their accelerator pedal WAY too soft (too easy to push to the floor).
I'd rather have a real pressure (especially in CHILL mode) to accelerate vehicle significantly (driver has to actually push slightly to have heavy acceleration, not just touch the pedal with one toe). So similar like brake pedal feel.
It's easy for active pedal to adjust pushback force according to speed and energy required to coast at that speed. Ideally, car should be teached, how heavy driver's foot is (is it made out of lead or not) before first drive, and then, later on, keep exactly that pressure for constant speed. A gram more for accelerate, a gram less to decelerate. That would be PERFECT!
I can imagine this system can be used for added safety - before AEB event it can start pushing accelerator. Or when AP requires holding the steering wheel, same story. It will decelerate vehicle (no less than coast) way earlier.


It (blending) can definitely be done perfectly. Tesla just went the easy way of not dealing with that.

How much of this problem is caused by regen on a Tesla being so weak a stopping force? Could it work better if there was a small bank of supercapacitors that could soak up more energy faster?

Supercaps are ideal for that. Car needs just 100-300Wh of capacitance for regen. It will help with regen efficiency. And also overall efficiency.
But actually, Tesla's battery can regen at much higher rates, of at least 100-120kW. It's just software that limits it. And the limit is there as there is no blending on Tesla's. Tesla couldn't allow 120kW regen with all pedals released. Like I said, no adjustability and no blending.
Leaf, with 24kWh battery, can accept regen at 50kW (again, software limited, but at least it is reasonable limit). Tesla, with 70kW battery can definitely regen at 150kW (it can charge at 120kW rate for minutes, why can't it charge that fast for few seconds). It will not be as efficient as 70kW regen, but it would work. But it requires blending or separate button/paddle for reasonable control.

 

[SDRick]

Ideally, I suppose it would be best to have a selectable choice on which pedal to use for regeneration. That said, I love the one pedal driving not having to move my foot back and forth between the two peddals. Those complaining about their feet hurting could also try autopilot or TAAC.