Chassis CAN Logging To ASCII Text Plus Graphing

[wk057]

Clarification:
My impression is that currently the front never turns off. That's not in conflict with what you've said here.

The front is definitely not in use with range mode off while traveling at < ~55 MPH and not accelerating or decelerating, based on what I've seen so far.

 

[MikeBur]

The front is definitely not in use with range mode off while traveling at < ~55 MPH and not accelerating or decelerating, based on what I've seen so far.

Ok, I snuck out and got some more data...

Note the flip-flopping between active motors. The smoking gun is here!

First section of road:

33mph Range Mode OFF, uphill then flat

33mph Range Mode ON, uphill then flat

Other section of road:

40mph Range Mode OFF, downhill then uphill

40mph Range Mode ON, downhill then uphill

More data, graphs and xls on the google drive

- - - Updated - - -

Also note the strong oscillations on rear motor when range mode is off

 

[brianman]

@wk - "not in use": does that show up as zero torque and zero power in the charts? I'm not seeing that in Mike's charts.

 

[MikeBur]

@wk - "not in use": does that show up as zero torque and zero power in the charts? I'm not seeing that in Mike's charts.

I thought this did Brian, ie. zero torque for front motor (the *light* green line)when range mode OFF, and active when range mode is ON. Whereas rear motor is the opposite. Only mph is left axis, Tq is right axis.

Am I missing something?

 

[brianman]

I thought this did Brian, ie. zero torque for front motor (the *light* green line)when range mode OFF, and active when range mode is ON. Whereas rear motor is the opposite. Only mph is left axis, Tq is right axis.

Am I missing something?

Nope, you're not. I totally misread the graphs. Axes, colors, everything. Mea culpa.

Sidenote: Thanks for capturing and sharing all this interesting stuff.

Also, what does negative torque mean (F_Tq in the last graph in your latest post)? Downhill/regen braking (like it does for negative power)? For some reason, I was thinking torque was an absolute value thing (i.e. like speed is an integer while velocity also has direction).

 

[wk057]

Mike, is the label in your second to last chart incorrect? Looks like a range mode OFF chart, but says ON in the text on the image.

Negative torque is regen.

 

[lolachampcar]

I did some slip mode testing and found little difference in performance. The first launch is normal while the second and third were slip mode on. Repeatability is scary.
Dropbox - SlipStartLaunches2_26_16.TXT

1792.96 100.0 +060.05 +176.8 +156.3 295.02 -1309.7 +07228
1789.57 100.0 +000.30 +386.4 +231.4 367.96 -0219.7 +00352
3.39 sec
SoC BatT BatOdo
080.7 30.52 016335




1825.90 100.0 +060.05 +179.0 +158.3 295.51 -1309.0 +07230
1822.59 100.0 +000.25 +372.2 +226.6 367.55 -0206.4 +00331
3.31 sec
SoC BatT BatOdo
080.3 31.13 016335




1917.34 100.0 +060.05 +181.0 +154.8 298.14 -1271.9 +07233
1914.03 100.0 +000.20 +323.9 +208.1 368.43 -0188.4 +00345
3.31
SoC BatT BatOdo
079.5 32.92 016336

 

[smac]

i want to say, you're welcome... Though did you really mean Mark? ;-)

Definitely worth also looking at Chassis CAN Logging To ASCII Text Plus Graphing - Page 24, as it shows even under minimal acceleration conditions the front motor is preferred. I won't have ability to do any more until Monday now.

lol... oops, sorry Mike

- - - Updated - - -

Interesting plots of the range mode differences at 40!

I guess this gives us a ready way to determine how long Torque Sleep takes to awaken! Steady state cruise with range mode on or off, then floor the accelerator. Easy way for us to determine how long it takes for the other motor to come on stream.

I am still assuming the non P dual motor cars have different diff ratios front to rear, so range mode might make a slight difference if the motor with the most available wheel torque isn't the one in operation at that time.


Fascinating thread this one!

 

[MikeBur]

Mike, is the label in your second to last chart incorrect? Looks like a range mode OFF chart, but says ON in the text on the image.

Negative torque is regen.

Yep, graph label is wrong. Just checked. Can't update currently, though yes 2nd to last is Range mode OFF, as per the text. Damn cutnpaste;-)

- - - Updated - - -

lol... oops, sorry Mike

- - - Updated - - -

Interesting plots of the range mode differences at 40!

I guess this gives us a ready way to determine how long Torque Sleep takes to awaken! Steady state cruise with range mode on or off, then floor the accelerator. Easy way for us to determine how long it takes for the other motor to come on stream.

I am still assuming the non P dual motor cars have different diff ratios front to rear, so range mode might make a slight difference if the motor with the most available wheel torque isn't the one in operation at that time.


Fascinating thread this one!

np, on Marc/Mark/Mike - I've been called a lot worse ;-)

The other range spreadsheet has me playing with throttle more. Likely we already have this answer in there...?

as for the p vs non-p motor gearing, I completely align on your belief. Having two identical hearings would be a squandered opportunity considering the algorithmic complexities are already considered for BIG.little dual motors. Oh, and we need a volunteer then. I'll reach out to Breser to see if he's interested. He lives reasonably close

 

[Andyw2100]

I have a theory concerning torque sleep that you guys with loggers and dual motor cars may be able to test.

Last winter, when Tesla rolled out the first firmware that enabled torque sleep on our P85Ds I was struggling with the fact that the efficiency numbers I was seeing, as compared to EV Trip Planner expected numbers, were not as good as other members' numbers compared to EV Trip Planner. (By comparing to EV Trip Planner expectations I was adjusting for temperature, elevation change, etc.) I began to question if torque sleep was even working on my car. Spring and Summer arrived, and I started beating EV Trip Planner numbers by about the amount others were.

My theory now is the following: above some required amount of power needed, torque sleep just doesn't kick in at all, or works far less effectively. So our driving 65 to 70 in extremely cold, dense air was above the threshold at which the real benefits of torque sleep would have been noticeable. Once the temperature warmed up, we saw the benefits of torque sleep.

The way those of you with loggers could help test this theory, over time--I'm certainly not suggesting anyone set out specifically to test this--would be to occasionally note, record etc. a bit of a highway cruising run at some set speed, noting the temperature, wind speed if possible, and then of course later noting the torque from the front and rear motors. If my theory is correct, at some combination of speed and temperature we'll find torque sleep not kicking in. We may also find that torque sleep works when cruising at 80 MPH in 70 degree temperature, but drop the temperature to 40, and cruising at 80 no longer results in any torque sleep at all.

If we're able to get a handle on this, aside from the fact that it will completely explain why I was pulling my hair out last winter, it may help us maximize efficiency if that's what we want to do. If we learn that at x temperature we can travel y speed with maximum torque sleep, but when we kick it up to y+1 we get no torque sleep, that would be helpful information for everyone to have.

Does this make sense?

 

[brianman]

The way those of you with loggers could help test this theory, over time--...--would be to occasionally note, record etc. a bit of a highway cruising run at some set speed, noting the temperature, wind speed if possible, and then of course later noting the torque from the front and rear motors.

Partially in answer to Lola's offer to lend out some logging hardware...

If I was to get into the data game, I'd probably want to combine: Tesla telemetry logging, lola hardware logging, and dashcam video footage on an ongoing basis. Hence interest in acquiring hardware rather than borrowing it.

 

[llavalle]

as for the p vs non-p motor gearing, I completely align on your belief. Having two identical hearings would be a squandered opportunity considering the algorithmic complexities are already considered for BIG.little dual motors. Oh, and we need a volunteer then. I'll reach out to Breser to see if he's interested. He lives reasonably close

Happy to help with a 85D. I have a logger and parser but from what I gather, no one has found a front rpm message, correct?

If you have, poke me and I'll give it to you right away.

 

[MikeBur]

Happy to help with a 85D. I have a logger and parser but from what I gather, no one has found a front rpm message, correct?

If you have, poke me and I'll give it to you right away.

thanks, I don't believe anyone has discovered this yet. Reading energy and torque from front and rear motors will show if the same criteria are used for motor "idling" for range on/off. I found it useful to use TACC and combination of couple of road geographies. I did 35, 40. I was unable to do 50 on TACC due to single lane restrictions.

 

[MikeBur]

...
I guess this gives us a ready way to determine how long Torque Sleep takes to awaken! Steady state cruise with range mode on or off, then floor the accelerator. Easy way for us to determine how long it takes for the other motor to come on stream.
...

Here's one from the same series as above. Range mode being OFF and just increasing the accelerator gently. Admittedly with multiple variables - road geometry for this stretch is slight dip then up - so varying throttle and geometry.

I believe I may have caught the sequence for the front motor enabling. Note that while the accelerator is increasing there is a momentarily blip on speed (0.15mph) and -ve Tq on front motor, even though there isn't on rear motor. After a corresponding +ve Tq blip (of very similar magnitude), the front motor is now helping. As TPS reduces front motor goes back to idling (after a small torque blip). Interesting to note, even if I'm incorrect here.

Overview graph:

Zoomed-in graph:

- - - Updated - - -

I have a theory concerning torque sleep that you guys with loggers and dual motor cars may be able to test.

Last winter, when Tesla rolled out the first firmware that enabled torque sleep on our P85Ds I was struggling with the fact that the efficiency numbers I was seeing, as compared to EV Trip Planner expected numbers, were not as good as other members' numbers compared to EV Trip Planner. (By comparing to EV Trip Planner expectations I was adjusting for temperature, elevation change, etc.) I began to question if torque sleep was even working on my car. Spring and Summer arrived, and I started beating EV Trip Planner numbers by about the amount others were.

My theory now is the following: above some required amount of power needed, torque sleep just doesn't kick in at all, or works far less effectively. So our driving 65 to 70 in extremely cold, dense air was above the threshold at which the real benefits of torque sleep would have been noticeable. Once the temperature warmed up, we saw the benefits of torque sleep.

The way those of you with loggers could help test this theory, over time--I'm certainly not suggesting anyone set out specifically to test this--would be to occasionally note, record etc. a bit of a highway cruising run at some set speed, noting the temperature, wind speed if possible, and then of course later noting the torque from the front and rear motors. If my theory is correct, at some combination of speed and temperature we'll find torque sleep not kicking in. We may also find that torque sleep works when cruising at 80 MPH in 70 degree temperature, but drop the temperature to 40, and cruising at 80 no longer results in any torque sleep at all.

If we're able to get a handle on this, aside from the fact that it will completely explain why I was pulling my hair out last winter, it may help us maximize efficiency if that's what we want to do. If we learn that at x temperature we can travel y speed with maximum torque sleep, but when we kick it up to y+1 we get no torque sleep, that would be helpful information for everyone to have.

Does this make sense?

Agree more data is necessary.

If could be torque sleep is disabled due to road temp, so as to minimize potential for loss of traction in potentially icy conditions moving to one motor across all speeds (eg. FWD better than RWD for slippy conditions

There's too many variables with too little data, at this point, to align on any one theory for your observation at the moment, eg. how does this relate to battery temp, outside temp, impediments to velocity (eg. wind, rain, wet roads, snow, air density, etc), across varying speeds.

Creating / using a logger that can store minutes of data on request, in combination with vbox, with something like TeslaLog (for environmental weather station info based on coarse GPS) would seem ideal? I believe Brian was heading down the same path?

 

[kennybobby]

Here are some charts from Bill's slip start testing at ~80%SOC of post #287. Run 1 is normal ff launch and runs2-3 are using slip start.

In run1 you can see where the rear rpm is modulated by traction control, the current has pulled back and the torques have a big bucket. There is still a little torque bucket with slip start, but it is much less of a pull-back and the resulting 0-60 times are quicker.

In the zoomed shot of run 2 you can see how the tiny fluctuation in current results in a large perturbation of the rear torque--this is seen in every run by all the cars. In addition i put some arrow lines to illustrate how the current ramp changes slope--obviously a steeper slope is more current in less time, and would result in more torque faster. The 450 ft-lb torque limit is reached very early on with only ~200 Amps.

Were there any tires smoking or slipping on these runs?

 

[llavalle]

thanks, I don't believe anyone has discovered this yet. Reading energy and torque from front and rear motors will show if the same criteria are used for motor "idling" for range on/off. I found it useful to use TACC and combination of couple of road geographies. I did 35, 40. I was unable to do 50 on TACC due to single lane restrictions.

I don't have enough data right now but from a short 30sec run with 15s in range mode and 15sec in normal mode, current is stable and total torque seems equal.

Range mode is 100% fwd. Normal mode is exactly 50%/50%.

On my 85D that is.

 

[MikeBur]

Here are some charts from Bill's slip start testing at ~80%SOC of post #287. Run 1 is normal ff launch and runs2-3 are using slip start.

View attachment 112959

.... snip others ...


In run1 you can see where the rear rpm is modulated by traction control, the current has pulled back and the torques have a big bucket. There is still a little torque bucket with slip start, but it is much less of a pull-back and the resulting 0-60 times are quicker.

Were there any tires smoking or slipping on these runs?

Interestingly I see both rear *and* front torque modulation during this time, indicating that this is likely *both* motors being modulated (though we obviously only have rear RPM currently).

I really could not get the tires to let loose. What Soc/BatTemp was this - oh, hang on, I can find that... hmm... 80%. Maybe new tires in order for someone? ;-)

- - - Updated - - -

I don't have enough data right now but from a short 30sec run with 15s in range mode and 15sec in normal mode, current is stable and total torque seems equal.

Range mode is 100% fwd. Normal mode is exactly 50%/50%.

On my 85D that is.

Thanks. Will be interesting to see 50/50 split data and how that changes at < ~50mph/80kph speeds

 

[llavalle]

Thanks. Will be interesting to see 50/50 split data and how that changes at < ~50mph/80kph speeds

My only dataset is at 100kph

 

[Andyw2100]

Agree more data is necessary.

If could be torque sleep is disabled due to road temp, so as to minimize potential for loss of traction in potentially icy conditions moving to one motor across all speeds (eg. FWD better than RWD for slippy conditions

There's too many variables with too little data, at this point, to align on any one theory for your observation at the moment, eg. how does this relate to battery temp, outside temp, impediments to velocity (eg. wind, rain, wet roads, snow, air density, etc), across varying speeds.

Creating / using a logger that can store minutes of data on request, in combination with vbox, with something like TeslaLog (for environmental weather station info based on coarse GPS) would seem ideal? I believe Brian was heading down the same path?

Glad to hear you are interested in this, Mike.

Last winter I was querying Tesla quite a bit, and couldn't get answers that actually made sense. They looked at logs from my car and came back with the answer that torque sleep was functioning properly, but the numbers they reported didn't actually support that conclusion. (I posted about all this at the time, and could dig up the thread and link to it if you are interested.)

Thanks!

 

[lolachampcar]

No real drama on any of the three runs but the data deffinitely shows traction issues on the first non-slip start run. Perhaps the first run warmed the tires a bit.