Torque sleep not functioning properly on some cars?

[redox]

I think that you are trying to "prove" something that is buried inside a complex system and that there are far too many variables in your real world to reveal the effect and you have no way to turn this particular parameter on and off to test the effect. I think you will drive yourself crazy trying to control all of the variables (temperature, road surface, altitude, speed, wind, tires, throttle pressure, etc.). I'm sure that Tesla has complex analysis systems that they use to measure and fine tune the motor controls but I don't think they would give you one of these systems so that you could prove that it is working to your satisfaction on your car. I'm perfectly content to let Tesla do the engineering and not make them prove to me that it is working as they say.

I think it's very challenging to do this at a microscopic level. However, at a macroscopic, we should all see, since the availability of the new firmware supporting Torque Sleep, a "general improvement" in power usage (Wh/mi) or range (miles) - of about 5-10% (I think that's the number communicated).
I can't definitely say that I'm seeing such improvement (everything else being roughly equal in terms of driving, temperature, etc), but I do see people who are reporting it, so there is a legitimate question as to whether there is something wrong with some cars.
I wouldn't try to control all variables: just look at a long term average. There should be some visible improvement (5-10% is visible/noticeable).

-- Greg

 

[mspohr]

I think it's very challenging to do this at a microscopic level. However, at a macroscopic, we should all see, since the availability of the new firmware supporting Torque Sleep, a "general improvement" in power usage (Wh/mi) or range (miles) - of about 5-10% (I think that's the number communicated).
I can't definitely say that I'm seeing such improvement (everything else being roughly equal in terms of driving, temperature, etc), but I do see people who are reporting it, so there is a legitimate question as to whether there is something wrong with some cars.
I wouldn't try to control all variables: just look at a long term average. There should be some visible improvement (5-10% is visible/noticeable).

-- Greg

I'm finding it hard to imagine why some people think there is something wrong with their special car. Torque sleep is entirely implemented in software and if you have the software, it will work the same in all cars. If you have the software, it is working. What would be special about some cars where the software would not work?

 

[ArtInCT]

I already have too many indicators nattering at me. I really don't want to have another blinking light distracting me and tempting me to "game" it. I'd much rather just drive the car. I trust Tesla to optimize the performance of the car and leave me to just drive.

I can appreciate your point. Perhaps the indicator could have an on / off setting so therefore, you will not be aware. I for one, a engineer by trade, like information and appreciate it. The cool part of the software implementation is that it can be tailored.

 

[billarnett]

I can appreciate your point. Perhaps the indicator could have an on / off setting so therefore, you will not be aware. I for one, a engineer by trade, like information and appreciate it. The cool part of the software implementation is that it can be tailored.

+1. It's really frustrating that there's all that info available but we can't see it. How about another "app" that looks like a bunch of meters showing battery & motor temps, motor rpm/torque, various currents and voltages, etc?

 

[ArtInCT]

+1. It's really frustrating that there's all that info available but we can't see it. How about another "app" that looks like a bunch of meters showing battery & motor temps, motor rpm/torque, various currents and voltages, etc?

Perhaps any indicator should be a form of a "training mode for torque sleep" that can be turned on or off. It should be in the central display and most logically in the discharge/ReGen area of the right hand sweep indicator. Something as subtle as a texture change or color intensity change would do. (Got to keep color blind folks in mind.)

As an example, our lowly Prius has an optional display that shows EV mode and charge discharge in a horizontal bar and as as heads up display on the windshield (did I say lowly?)(I did say lowly). A wonderful tool for learning how to drive the Prius correctly if the intent is efficiency. Once one learns how to drive the Prius that way, we' can turn that display off. So therefore it is not in your face all the time. Very customizable.

I have seen the service menus of the model S and when I get mine, I would love to have a read only view of those menus. Man there is data there... Tons of it from what I have seen. Not too often talked about here on TMC.... that future app sounds kewl to me.

 

[gpetti]

I'm finding it hard to imagine why some people think there is something wrong with their special car. Torque sleep is entirely implemented in software and if you have the software, it will work the same in all cars. If you have the software, it is working. What would be special about some cars where the software would not work?

Would that it were that simple. The car is a delicate balance of hardware and software. As you pointed out there are many variables. For software that expects simple user input, this can sometimes be controlled to a relatively finite set of events and expected inputs. With a car that has a myriad of variable sensors, the software has an onslaught of data to deal with and decisions to make. Ideally the engineers have predicted different errors or hardware occurrences and written software to react accordingly but it is unlikely to catch every possibly eventuality. Andy could have a hardware problem or some subtly different parameter - a dodgy sensor or whatever. Ideally he shouldn't have to do all of this but I don't feel that he has had a conclusive, well thought out response that would personally give me confidence that they have examined his car carefully looking for anything untoward. Ultimately despite whatever test systems Tesla have, it is very difficult for them to re-create all the real world scenarios (which is why bugs slip through) so ideally they should be very glad of Andy's thorough analysis and research. To illustrate this, when I had the warning message on my car (btw, many cars didn't see this error even though we were all running the same software) I initially called roadside assistance and they told me there was nothing untoward in the logs. When my SC looked at the logs, they saw some potentially higher than normal temperatures and thus are replacing my front motor. Not sure yet if that was the right call or not but either way - different interpretations of the information. Reading logs is a bit of a black art and even for people in engineering who wrote the software in question, it is possible to overlook clues.

 

[redox]

I'm finding it hard to imagine why some people think there is something wrong with their special car. Torque sleep is entirely implemented in software and if you have the software, it will work the same in all cars. If you have the software, it is working. What would be special about some cars where the software would not work?

The thing is it's not just software - it's a combination of software, hardware, and other external factors (conditions). And not all hardware is the same btw.
Second, it's quite common to have bugs. They're everywhere. Have you never had an appliance that didn't work the way it should? (or a piece of software for that matter) Yet that same product is sold to many people so obviously it is working, but somehow not for you?
The point is that these systems are complex, especially in terms of interactions.

If you can't "sense" an advertised feature (but others can), it seems legitimate to have a doubt and explore, IMHO.

-- Greg

 

[AmpedRealtor]

I don't have a P85D - just a lowly P85 - but I recently made two identical trips on two different days, and look at the results...

Two days ago I drove 70 miles or so. It was in the low 60s where I live. I drove 70-75 MPH on the freeway. My average energy consumption for that trip, according to the "Since Last Charge" screen, was around 310 Wh/mi.

Today I made the same trip. It was in the mid 70s. I drove 65 MPH on the freeway. My average energy consumption for this last trip - an identical route - was 266 Wh/mi.

That's a difference of about 15% within two days, on the same exact route.

 

[Andyw2100]

I think that you are trying to "prove" something that is buried inside a complex system and that there are far too many variables in your real world to reveal the effect and you have no way to turn this particular parameter on and off to test the effect. I think you will drive yourself crazy trying to control all of the variables (temperature, road surface, altitude, speed, wind, tires, throttle pressure, etc.). I'm sure that Tesla has complex analysis systems that they use to measure and fine tune the motor controls but I don't think they would give you one of these systems so that you could prove that it is working to your satisfaction on your car. I'm perfectly content to let Tesla do the engineering and not make them prove to me that it is working as they say.

I'm not trying to control the variables like temperature and speed. I just try to record them accurately, so I can compare my actual efficiency to what EV Trip Planner predicts. It hasn't been all that difficult, mainly because for most of the trips it means making my wife do it.

As for Tesla's complex analysis systems, as this point I would settle for any indication that they know that torque sleep is actually working properly on my car. I really think that it is possible that all the Tesla engineer did when he pulled the logs was confirm that I had a firmware with torque sleep, and then extract some numbers. I have no evidence of engineering checking to see that torque sleep is functioning at all, much less functioning properly. I'm not trying to get them to give me any system to let me prove anything. I just want confirmation that my car is functioning properly, because the evidence that I have thus far leans strongly in the direction that it is not.

- - - Updated - - -

Perhaps it's not the cold, but rather snow/slush/water on the street. If the AWD system is kicking in to maintain traction, torque sleep isn't going to be able to engage.

We're not driving on snow/slush/sleet/water a large enough percentage of the time for that to account for seeing almost no efficiency gains from torque sleep. The highway clears pretty quickly after it snows. I appreciate the thought, but I really don't think that could be it.

- - - Updated - - -

I've seen the same behavior that LolaCC reports. It's subtle, though: I can't feel it but I can see it happen on the power meter.

I'll definitely try to watch for it on the power meter. Thanks!

- - - Updated - - -

Andy:
I know you desire to make sure that your P85D is functioning correctly vv Torque Sleep. However, since you share seat time with your wife and the weather and conditions are sub optimal in your local, I suggest you wait a month or so to when the weather is better and then see what happens. Till then, just enjoy the car and yourself. Seems to me that you and your wife are able to get back and forth OK even in the cold and snow as it stands now. Were I you, that is what I would do... Drive it like you stole it and smile.

But even if I were to do that, what would that prove? Let's say I suddenly see efficiency improvement from torque sleep when the weather is warmer. Tesla never said "torque sleep doesn't work in the cold." If that is why I'm not seeing an efficiency improvement now, don't you think all Tesla owners and potential owners should know this about torque sleep? The winter is when we most need efficiency gains and every little bit of range, because when it is cold is when range suffers. If it turns out that torque sleep just doesn't work at all when it is cold, I really would like to find that out.

- - - Updated - - -

I already have too many indicators nattering at me. I really don't want to have another blinking light distracting me and tempting me to "game" it. I'd much rather just drive the car. I trust Tesla to optimize the performance of the car and leave me to just drive.

Information is power. You can choose to ignore the information if you want to, but if you don't have access to the information you are operating in the dark. I'd love to have some sort of indication as to which, if either, engine is torque sleeping.

- - - Updated - - -

I think it's very challenging to do this at a microscopic level. However, at a macroscopic, we should all see, since the availability of the new firmware supporting Torque Sleep, a "general improvement" in power usage (Wh/mi) or range (miles) - of about 5-10% (I think that's the number communicated).
I can't definitely say that I'm seeing such improvement (everything else being roughly equal in terms of driving, temperature, etc), but I do see people who are reporting it, so there is a legitimate question as to whether there is something wrong with some cars.
I wouldn't try to control all variables: just look at a long term average. There should be some visible improvement (5-10% is visible/noticeable)

-- Greg

Well, in addition to the trip data I'm collecting, I did include some long-term data. It doesn't show anything close to a 5-10% improvement. As I explained in my initial post, it shows an improvement of 14 wh/mi. 14/416 = 3%. But as I explained, that is certainly an inflated figure, because of the reasons I stated earlier, those being:

1) New tires for about half of the original period being measured
2) Range mode never used in original period and then used extensively in subsequent period
3) Far more 0-60 launches done in initial period

I actually think almost all of the 3% can be attributed to the things above.

- - - Updated - - -

I'm finding it hard to imagine why some people think there is something wrong with their special car. Torque sleep is entirely implemented in software and if you have the software, it will work the same in all cars. If you have the software, it is working. What would be special about some cars where the software would not work?

I just edited myself before hitting "post", which was probably wise, as I might have had my first ever post moderated away to snippiness or someplace. But I really can't believe you posted that.

Yes, every Tesla ever made and every part in every Model S ever made has always worked flawlessly forever. There was never a part anywhere, ever, that malfunctioned, or had been manufactured with a small defect. There was never a bug in a firmware version that could have an affect on some cars but not others. None of those things are remotely possible. I'm not sure what I was thinking.

- - - Updated - - -

I don't have a P85D - just a lowly P85 - but I recently made two identical trips on two different days, and look at the results...

Two days ago I drove 70 miles or so. It was in the low 60s where I live. I drove 70-75 MPH on the freeway. My average energy consumption for that trip, according to the "Since Last Charge" screen, was around 310 Wh/mi.

Today I made the same trip. It was in the mid 70s. I drove 65 MPH on the freeway. My average energy consumption for this last trip - an identical route - was 266 Wh/mi.

That's a difference of about 15% within two days, on the same exact route.

I'm not sure what your point is, or how that applies to this discussion. I'm fully aware that efficiency varies greatly with temperature. If you take a close look at the numbers I included, you'll see huge variations in some of the same trips (taken on different days.) But I'm plugging the temperatures in to EV Trip planner, and then comparing my numbers to what EV Trip Planner predicts. EV Trip Planner also accounts for the temperature.

Taking temperature into account, I am consistently seeing efficiency worse than EV Trip planner predicts. Taking temperature into account, and IN THE COLD other P85D owners are seeing efficiency 8-12 % BETTER than what EV Trip Planner would predict.

That is one of the main reasons I think something may be amiss with the way torque sleep is functioning in my car.

 

[redox]

I don't have a P85D - just a lowly P85 - but I recently made two identical trips on two different days, and look at the results...

Two days ago I drove 70 miles or so. It was in the low 60s where I live. I drove 70-75 MPH on the freeway. My average energy consumption for that trip, according to the "Since Last Charge" screen, was around 310 Wh/mi.

Today I made the same trip. It was in the mid 70s. I drove 65 MPH on the freeway. My average energy consumption for this last trip - an identical route - was 266 Wh/mi.

That's a difference of about 15% within two days, on the same exact route.

I think you'd expect that drop simply from the slower speed (65 vs 70-75 MPH) & temp difference. If you go to Tesla's own site and play with the "range estimator" (you can tweak things like speed, city/highway, temperature, etc...) you can see the range (and therefore Wh/mi) varies by that amount when you change speed by 5-10 MPH in the 60 MPH "zone". It's under the "top 5 questions" section.

Those higher MPHs are quite costly energy-wise.

-- Greg

 

[yoyo]

I am wondering the same thing about whether torque sleep works "differently" on different P85Ds...
Before and after the torque sleep, my average WPH did not seem to change.
with .179, just took a trip yesterday, heated the cabin and ended full charge right before leaving
outside between 20 to 15 F
Range mode on all the way, mostly highway. ~ 50 miles each way. highest speed was 75MPH, 25% segment about 50MPH. cabin temp set to 67.

I ended the trip with 79 rated miles remaining (started with 251 rated miles). I used about 10 rated miles for pre-heating before coming back. So that's 162 rated miles used for 100 miles driving.

tesla web shows at 70MPH, heat on, 32F outside, 217 mile range.
For me, it would be only 156 miles with only 15F colder.... what did I do wrong???

I am still not getting the range

 

[Andyw2100]

I am wondering the same thing about whether torque sleep works "differently" on different P85Ds...
Before and after the torque sleep, my average WPH did not seem to change.
with .179, just took a trip yesterday, heated the cabin and ended full charge right before leaving
outside between 20 to 15 F
Range mode on all the way, mostly highway. ~ 50 miles each way. highest speed was 75MPH, 25% segment about 50MPH. cabin temp set to 67.

I ended the trip with 79 rated miles remaining (started with 251 rated miles). I used about 10 rated miles for pre-heating before coming back. So that's 162 rated miles used for 100 miles driving.

tesla web shows at 70MPH, heat on, 32F outside, 217 mile range.
For me, it would be only 156 miles with only 15F colder.... what did I do wrong???

I am still not getting the range

The 75 MPH makes a big difference. The 15 degrees colder does too.

But to compare the efficiency you are getting to the efficiency I am getting and the efficiency other P85D owners are getting please try to put that exact trip into EV Trip Planner as best you can, matching up the route exactly, the temperature estimate to what a good average was, etc., and then for speed, tweak the speed factor to make it work out to the length of time the trip took you. I say that because you said part of it was at 75 MPH, but some portion was at 50 MPH. EV Trip Planner will show you the speed it is using for each segment in the "details" tab. So if you had gone a constant speed, you could just use a speed factor to match up your speed with the correct speed there. But since you didn't, just try to get the total time EV Trip Planner estimates to match the length of time the trip took you by adjusting your speed factor until you get there. This is really important, as you will see, as you adjust that speed factor, how much it changes the numbers.

Once you have EV Trip Planner estimates for what the efficiency should have been on the trip, compare that to what it was. I know you have the range miles you used--172. Do you have any of the other numbers--the average energy (wh/mi) and the total energy used? If not, that's OK.

I'll be very interested to see how your numbers compare to EV Trip Planner numbers.

Thanks.

 

[yoyo]

ok, I put it into trip planner, selected P85D 19". It showed 68RM + 70RM with average 405Wh and 423Wh.
I used 1.15, 67F, and 18F outside, and 300lb.
so it would be 138RM vs my 156RM -- though in 7 mile segment (each way) it shows speed of 49mph, but really should be 60mph (speed limit 55mph)

so I guess it is not that far off...11% off

oh, I should mention, I am in Chicago region and it was windy (always windy here)... not sure which direction the wind was....

 

[Andyw2100]

ok, I put it into trip planner, selected P85D 19". It showed 68RM + 70RM with average 405Wh and 423Wh.
I used 1.15, 67F, and 18F outside, and 300lb.
so it would be 138RM vs my 156RM -- though in 7 mile segment (each way) it shows speed of 49mph, but really should be 60mph (speed limit 55mph)

so I guess it is not that far off...11% off

oh, I should mention, I am in Chicago region and it was windy (always windy here)... not sure which direction the wind was....

OK, so using a speed factor of 1.15 got you an estimate of about the amount of time it took you? That's probably close enough. That 7 mile segment you went faster than EV Trip Planner thinks you did, so it's estimate for that segment will be lower than it should be, but that should not be significant since it's a long trip, and the overall time is correct.

As for the overall, here's the thing: first of all, having EV Trip Planner beat you in range miles by 11% is actually a pretty significant amount. EV Trip Planner usually beats me in range miles, but usually by less than 11%. Much more importantly, EV Trip Planner isn't yet taking torque sleep into account at all. Until your post, I didn't know EV Trip Planner had a P85D option. (It must be very new.) I had been selecting just the regular Model S85, as had everyone else who posted comparisons. (I just checked my numbers from today, and there is almost no difference between the P85D and the S85 as far as EV Trip Planner is concerned--certainly not enough difference for it to be taking torque sleep into account.) Anyway, other P85D users, who think they are getting improved efficiency from torque sleep, typically beat EV Trip Planner estimates by 8-12%! So you're looking at an efficiency 19-23% worse than other P85D users, on that trip! Of course the wind could have been an issue, as you suggest. To get an idea of which way the wind is blowing, and how hard, for future trips, see this thread: Tool to estimate the head wind while driving !

Another issue could have been your speed. It has been suggested that at speeds of 75 MPH and up torque sleep won't kick in at all, or that its benefits will be reduced.

Check this thread for some other P85D users posting comparisons to EV Trip Planner: Comparing P85D Torque Sleep efficiency (versions .139 and .140) to EV Trip Planner

Although the wind could have been an issue, and your high speed could have been as well, the trip data you just entered, along with your overall data add support to my theory that some P85Ds are not seeing the torque sleep efficiency gains, and that the cold weather could be the reason.

 

[Andyw2100]

I just had a bit of an epiphany about this issue. Tell me if this makes sense. (I really think it does.)

I think torque sleep, the way it is currently set up, is not working in the cold. This is what I think is happening and why.

The idea behind torque sleep is that when the car is just cruising along, and not requiring a lot of the motors, the bigger back motor will shut down.

(I'm only using the following numbers to illustrate what I'm saying--they may be way off.)

Let's say that in California a Model S on level ground with a couple of average size passengers and a temperature of 60 degrees Fahrenheit, cruising at 65 MPH uses 300 wh/mi. When the Model S computers sense the car settling in at that speed and that energy usage the back motor shuts down, and then the energy usage decreases. If it was colder or the car was going faster, maybe the car would be using 320 or 330 wh/mi at the time that torque sleep kicks in.

Now let's switch to my situation in New York State. It's a lot colder here. The air is denser. Cruising at the same 65 MPH is going to take a lot more than 300 or even 330 wh/mi. Let's say it takes 360 wh/mi to cruise at 65, on flat ground, in 15 degree temperature. But what if torque sleep is set up never to kick in if the Model S is drawing more than 355 wh/mi? In that case, torque sleep would basically never kick in when it was very cold.

Again, the numbers are just for illustrative purposes.

We have reports of other people saying they don't think torque sleep works at all above 70 or 75 MPH, depending on who the report is coming from. That difference could depend on their weather, and on how much weight the car was carrying. It is consistent with what I'm suggesting now. There is some energy draw threshold above which the car will not enter torque sleep, and that threshold may be exceeded at a highway cruising speed when it is cold.

This might also explain why I've seen some very slight efficiency benefits, as sometimes I may be cruising at a low enough speed that the energy requirements allow torque sleep to kick in once in a while, or in warm enough temperatures once in a while.

And if I'm right about any of this, it's really important that I get Tesla's attention on it. Because for one, they could conceivably tweak the torque sleep settings either for everyone, or differently for cars, by region. Or, if they decide that this is just the way that torque sleep is going to be, then they need to explain that it won't work when it's cold, and get that information out there.

Thoughts on this new theory?

 

[BerTX]

Makes sense to me.

There are probably other factors that could cause the same thing, but not on a long-term basis like cold. A headwind will cut your range because of the increased drag, but the range reduction will be magnified over the normal effect because the car will also not be able to use torque sleep.

Most of these things will not cause you to think there is a problem with the car, because they won't last for months.

 

[redox]

Thoughts on this new theory?

It's an interesting theory, but that would mean I should be able to notice improvements (it's been pretty warm here in NorCal), but I don't really see much of an improvement actually.

-- Greg

 

[BerTX]

It's an interesting theory, but that would mean I should be able to notice improvements (it's been pretty warm here in NorCal), but I don't really see much of an improvement actually.

-- Greg

But there can be any number of things that can cause increased load, and therefore might be canceling torque sleep. Driving in even slightly hilly terrain with the cruise control on. Lower tire pressure. Wind. High speed. Wet highway. Snow on the road. Snow tires. Traffic. Long slight uphills. Mechanical flaws (wheel alignment, brakes dragging, etc.). Driver bad habits (repeatedly accelerating and slowing unconsciously as they drive).

Some of these things might only happen occasionally, so they would not be noticed or the cause easily identified. Some of them might happen over a long period (like the cold) that would make it appear that torque sleep was not functional at all.

Or maybe this is all wrong and torque sleep isn't functioning on some cars.

 

[Andyw2100]

It's an interesting theory, but that would mean I should be able to notice improvements (it's been pretty warm here in NorCal), but I don't really see much of an improvement actually.

-- Greg

Have you compared some trips to EV Trip Planner estimates to see how your numbers look relative to some of us?

 

[donv]

I suspect it's simpler than this. When the car is at a steady state speed, it requires x horsepower to maintain that speed. Tesla sets a buffer over and above x for hills, passing, and general responsiveness. When the front motor alone can produce x+y horsepower, it sleeps the rear motor. When it can't, it runs both.

Thoughts on this new theory?