Regen reduced when friction brakes are engaged

[TickTock]

Made a little progress. Found a speed signal *and* what appears to be a battery (or motor) current signal similar to those on the EV bus. The graphs got really messy and hard to follow once I added those but I did confirm that, yes, there can still be some regen even though the "Braking Regen" signal goes to zero. For example, when coasting to a stop in ECO mode, there is no "Target Braking Force" Or "Regen Braking" signals however, there is charge going into the battery. Kinda interesting - appears to be a constant current even though speed is reducing. The amount of regen did however, always drop when the "Braking Regen" signal gave way to friction braking. I was never able to keep strong regen once friction was engaged. I'll post some pics for those interested once I find a good way to plot them that doesn't make your eyes hurt.

 

[johnr]

Do the signals every indicate regen shuts off while the car's display shows strong regen? If so, I would wonder about the discrepancy. In my experience, in most every case, assuming we're going at a reasonable fast speed (say, greater than 45mph) and no panic-stopping, regen will continue to be maxed-out (based on the car's display) even when slowing rapidly enough that I'm sure friction brakes are being applied. The regen is greatly diminished at slow speeds, I guess because you simply can't generate much energy when the motor is turning too slowly - so I wouldn't consider any data to be valid at speeds less than say 25mph. Just my two cents, for what it's worth

 

[TickTock]

Some more plots to ponder:
First is heavy braking from 40 MPH with speed and battery/motor current. I tried to smoothly increase the brake pressure and you can see regen braking comes and goes but there is still current being pushed into the battery even after the Regen Brake Signal has stopped. Maybe if I was on a steep hill the amount of regen would stay higher?

Here is a plot where I allowed tha car to coast to a stop in eco mode (not really coasting, but foot off both brake and accel). Steady regen until about 5 mph.

I think I need to find a long hill. Have not been able to see any significant regen once friction is applied when coming to a stop on flat ground. Unfortunately, there aren't too many in the Phoenix valley. Maybe I'll visit South Mountain this week...

 

[DaveEV]

Weatherman says there is just such a transfer point in Prii (around 8 MPH). There's a similar transfer point in the Volt (somewhere between 4-7, depending on deceleration rate). So if the Leaf blends regen and friction at all, there'll almost certainly be a speed related transfer point as well.

There is definitely a similar one in the LEAF which you can feel if you're braking at the right level - you get a small reduction in braking force as the car switches from regen to friction around 5-8 mph like the other cars.

I have also definitely seen regen reduce under hard braking once a significant amount of friction brakes start getting blended in - by letting off a bit you can get regen to go back up, so there's definitely a threshold where adding more friction brakes reduces regen and vice versa.

 

[johnr]

That's strange. What the graphs indicate seems to misalign with my experiences. Possibly it's the lack of a long steep hill to better test this without losing speed too rapidly. Or possibly the gradual application of braking force changes the characteristics of the regen blending. In my own experience I found I could get the maximum regen by quickly stepping on the brake, firmly but not all the way. Gradually applying the brake tends to limit the regen at a certain level and does not allow it to increase.

 

[Ingineer]

TickTock, are you willing to share which CAN frames you found this data in? It would be nice to experiment further and add some more data.

Thanks!

-Phil

 

[TonyWilliams]

In my own experience I found I could get the maximum regen by quickly stepping on the brake, firmly but not all the way. Gradually applying the brake tends to limit the regen at a certain level and does not allow it to increase.

My experience is exactly opposite. I try to gently apply the brake pedal with just enough effort to maximize the regen. Quickly applying the brakes goes straight to friction brakes.

 

[Rusty]

TickTock can you associate a time baseline with your plots? That would help others (especially Ingeneer) in reproducing your experiments (whether qualitatively or quantitatively).

 

[TickTock]

Yeah, I was going to caution people not to read time into the X axis - it is linear with message recieved which isn't at a constant rate. It will take a little more effort to put on a time-scale. Will have to post process the logs (previous graphs were done in the CanDo program). Been meaning to do this anyway since I also want to plot things like ratio Ffriction/Ftarget versus ground speed, too. But my day job keeps getting in the way of the important stuff.

 

[TickTock]

OK. Finally plotted the data versus time. Interestingly, the data got more choppy. It appears the can data is bursty and block of messages are sent together (with close timestamps) even though it seems likely that the actual message content was accumulated over a longer period of time. Anyway, below is more of the same data except now the X axis is time in seconds. The Y axis is only valid for the line lebeled rpm. This is really mph. You will notice groups of spikes on the brake signals. That was my way of getting my bearings in the log. While at a stop, I pressed the brake twice before accellerating to 20mph, three times before the 30mph test, four for 40, and five for 50. At the very end I went to 20mph and allowed eco mode to bring it to a stop. There you can tell that total regen is the sum of braking regen and something else. It appears the regen turn off is less a function of speed then it is amount of brake applied *and how fast you apply it*. That latter part is interesting - along with the evidence that more regen is supported on longer braking down a hill suggests to me that Nissan is using a PID controller to apply the regen.

 

[TickTock]

TickTock, are you willing to share which CAN frames you found this data in? It would be nice to experiment further and add some more data.

Thanks!

-Phil

Sorry for missing this before. I am logging everything I learn here. There are four tabs. Flow is the table showing what messages go where (from the SM). EVcan and Carcan contain what I was able to deduct for those can busses. The final is a copy of Turbo2Ltr's very detailed notes on the carcan. I plan to eventually encorporate it into the Carcan tab.

 

[TickTock]

in case you missed the hidden url before:
https://docs.google.com/spreadsheet/ccc?key=0An7gtcYL2Oy0dGRaSWl6VTV2eXBQMy1ON2xZSzlMUXc

 

[Ingineer]

OK. Finally plotted the data versus time. Interestingly, the data got more choppy. It appears the can data is bursty and block of messages are sent together (with close timestamps) even though it seems likely that the actual message content was accumulated over a longer period of time. Anyway, below is more of the same data except now the X axis is time in seconds. The Y axis is only valid for the line lebeled rpm. This is really mph. You will notice groups of spikes on the brake signals. That was my way of getting my bearings in the log. While at a stop, I pressed the brake twice before accellerating to 20mph, three times before the 30mph test, four for 40, and five for 50. At the very end I went to 20mph and allowed eco mode to bring it to a stop. There you can tell that total regen is the sum of braking regen and something else. It appears the regen turn off is less a function of speed then it is amount of brake applied *and how fast you apply it*. That latter part is interesting - along with the evidence that more regen is supported on longer braking down a hill suggests to me that Nissan is using a PID controller to apply the regen.

Thanks for this.

FYI, the CAN frames are occurring at regular intervals, so there must be something wrong with your DAQ setup.

-Phil

 

[Ingineer]

My own observations are in agreement with the brake pressure delta/time being used to determine when to cancel (fade out) regen. I would assume that a high velocity brake application suggests a possible panic stop, so the hydraulic brakes assume most of the braking that way EBFD (Electronic Brake Force Distribution) and ABS/TC/SC can be ready to do their thing. Since regen can only operate on the front axle, they probably just took the easy way out, and eschew regen to make the control system easier. The software used in ABS has been developed over a long time and is well proven using stringent development practices. Uprooting all of it to add regen would take a lot of effort and time, maybe we'll see it in the future when EV's are more pervasive.

-Phil

 

[Ingineer]

Sorry for missing this before. I am logging everything I learn here. There are four tabs. Flow is the table showing what messages go where (from the SM). EVcan and Carcan contain what I was able to deduct for those can busses. The final is a copy of Turbo2Ltr's very detailed notes on the carcan. I plan to eventually encorporate it into the Carcan tab.

Thanks for compiling this! (And thanks to Turbo2Ltr as well) I will try to help add in things I discover. I have access to Consult III+ which for the most part uses active queries to return all the parameters listed in the SM. All the frames we see right now on the EV/Main CAN buses are "passive" frames sent at regular intervals for inter-module communication. Passive frames will only contain messages needed to make the car operate and not necessarily any additional diagnostic information.

-Phil

 

[suresnoi]

So under certain condition, regen indeed stops when friction brakes are engaged.

My question would be : how often does that happen when driving "normally" on the highway, on a smaller road, in town etc.

If this only occurs a few times in a month, I guess it's nothing much to be concerned about. But looking at these graphs, it seems the car gives-up the regen quite "easily".

Can someone relate this to real life driving ?

To be honest it's a little disappointing. A prius will keep the regen at it's maximum even under pretty sudden/hard deceleration. Only at the very last moment when the ABS is about to engage or is engage does the regen stops.

 

[LEAFfan]

So under certain condition, regen indeed stops when friction brakes are engaged.
My question would be : how often does that happen when driving "normally" on the highway, on a smaller road, in town etc.

Just glance at your energy regen screen when you are braking.

 

[aqn]

Weatherman says there is just such a transfer point in Prii (around 8 MPH). There's a similar transfer point in the Volt (somewhere between 4-7, depending on deceleration rate). So if the Leaf blends regen and friction at all, there'll almost certainly be a speed related transfer point as well.

My observation is that in ECO, the regen meter (center display) shows regen at speeds all the way down to 6-7 mph. The regen meter steadily decreases to a tiny thin line which would finally disappear at around 6-7 mph.

 

[aqn]

It appears the regen turn off is less a function of speed then it is amount of brake applied *and how fast you apply it*.

My commute includes a downhill of about 288 feet drop over 2500 feet (12% grade). My observation is that (in ECO with less than 8-9 bars of battery) full regen will only occur if I keep a light pressure on the brake. Regen will decrease to less than max regen if I brake harder. That's in line with the current theory here of the Intelligent Brake Unit (IBU) increasing hydraulic assist and decreasing regen braking because it thinks the driver wants to slow more quickly.

 

[aqn]

So under certain condition, regen indeed stops when friction brakes are engaged.

I wouldn't say that regen "stops when friction brakes are engaged". If the regen meter is to be believed, regen is reduced, not stopped, when driver applies friction brakes hard-ish.

To be honest it's a little disappointing. A prius will keep the regen at it's maximum even under pretty sudden/hard deceleration. Only at the very last moment when the ABS is about to engage or is engage does the regen stops.

Just curious: is the Prius' "max regen" behavior verified by instrumented measurements or deciphering of CAN bus data etc.? Or is it what its display indicates?