Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Chassis CAN Logging To ASCII Text Plus Graphing

This site may earn commission on affiliate links.
Below are some additional comparisons between the existing P85DL and Bill D's new Run1 . I've created a new sub-directory for these comparisons between Model S P85DL and P90DL in the Mikebur directory under "SP85DL to SP90DL compare...", or just click on this link: https://drive.google.com/open?id=0BzwKZAn3p2LTZXRIeVNaQ3R6aW8

I've also added Bill D's virtual drag strip calculations (thanks Bill :)), though there appears to be something funky in calculations so hoping this is just because these are not full 1/4mile runs(?)

First, here's the original graph from mine to put all in one place for comparisons. Nothing new, though updated formatting to match Bill D...
MikeBur P85DL Dry Run 55F Ludicrous Mode Launch 0-100.PNG


Also, virtual drag timestrip numbers for 0-60 and rollout times now. Ignore 1000', etc.
MikeBur P85DL Dry Run 55F Ludicrous Mode Launch 0-100 virtual drag timestrip.PNG


Ok, so back to comparing this run with the new run from Bill D, we can see significant similarity though the increased battery voltage, due to higher SoC, does makes more power available. It's too wet here currently for me to do a 95% SoC MaxBat run, so although a likely theory is not proven fact:
v2 MikeBur P85DL vs Bill D P90DL launch.PNG


Zooming in on launch again appears to confirm that all available power is not being made available and this is not a traction limitation, otherwise we should see a steeper ramp on Bill D's higher SoC P90DL:
v2 MikeBur P85DL vs Bill D P90DL launch (zoomed).PNG


This becomes more apparent when looking at the Power, vs. doing the I*V by eye ;)
Power (kW) MikeBur P85DL vs Bill D P90DL launch.PNG



Bill D, when looking at the launch mode profiles, I agree that these are extremely similar, though I would rather posit that this is loading the tire sidewalls, rather than traction control - the shape of the RPM curve is very closely related to the blip we see in torque, ie. from the same run on my p85dl, the shape of these initial bumps are too coincidental for two different sets of wheels and tires at two different ambient temperatures to be coincidence. For launch mode, these are identical and for fast foot it also appears to be extremely similar (though smoother indicating this is the normal behavior vs. programmed behavior on launch):
MikeBur P85DL mph RPM relationship.PNG


Comparing this to when traction control is engaged significantly (albeit this example from P85+ before R_RPM was captured) shows a very different characteristics of torque variation. Don't know exactly, though interesting nonetheless.
P85plus Traction Control.PNG


Just because I know someone will ask, how do our cars compare on the mph/rpm representation. To ruin the surprise almost the same again :)
mph to R_RPM MikeBur P85DL vs Bill D P90DL launch (zoomed).PNG


Ok, this almost brings me back up to date, yes? :)

Hope this helps, Mike
 
  • Like
Reactions: benjiejr
... I've also added Bill D's virtual drag strip calculations (thanks Bill :)), though there appears to be something funky in calculations so hoping this is just because these are not full 1/4mile runs(?)...
Welcome to Cyberspace Raceway Park. You and I just drag raced and I was 2 car lengths ahead at the 1/8 mile (you were at 620 ft at my 7.07 ET). I'll meet you at the virtual concession stand. Virtual drinks are on me! (The obvious next step in comparing log data is to compile it into an animated drag race). :smile:

On the funky numbers for runs that are not full 1/4-mile runs - "lifted" automatically appears next to each item on the virtual Timeslip if the throttle is not still at 100% at that point. This is to avoid any confusion over the funky numbers.

Also, I should point out the virtual Timeslip is derived from the MPH data and I don't know the actual source of that MPH data. If the source is slipping wheels, the Timeslip may be off. However, so far, they seem very close to my actual Dragstrip results.
 
  • Funny
Reactions: benjiejr
Welcome to Cyberspace Raceway Park. You and I just drag raced and I was 2 car lengths ahead at the 1/8 mile (you were at 620 ft at my 7.07 ET). I'll meet you at the virtual concession stand. Virtual drinks are on me! (The obvious next step in comparing log data is to compile it into an animated drag race). :smile:

On the funky numbers for runs that are not full 1/4-mile runs - "lifted" automatically appears next to each item on the virtual Timeslip if the throttle is not still at 100% at that point. This is to avoid any confusion over the funky numbers.

Also, I should point out the virtual Timeslip is derived from the MPH data and I don't know the actual source of that MPH data. If the source is slipping wheels, the Timeslip may be off. However, so far, they seem very close to my actual Dragstrip results.

LOL - Awesome! :)

Here's another graph for fun - CHAdeMO charging from 50% to 90% for free (thank you City of Portland! :))
MikeBur P85DL CHAdeMO charging 50-90 02-29-2016.PNG


I have generated gifs for creating animation, though it looks kinda boring... anybody want (teasing? :p)
 
Looks like you wouldn't want to be racing for virtual pink-slips unless charged up--about 8% SOC is good for one car length to 1/8 mile.

With equally charged packs i'd bet the data will be identical due to the common launch control logic. The main difference in a drag race will be due to SOC of the packs and maybe a secondary effect from tire grip.

Bill D and Mike B, excellent data and graphing as usual--thanks for sharing.
 
Looks like you wouldn't want to be racing for virtual pink-slips unless charged up--about 8% SOC is good for one car length to 1/8 mile.

With equally charged packs i'd bet the data will be identical due to the common launch control logic. The main difference in a drag race will be due to SOC of the packs and maybe a secondary effect from tire grip.

Bill D and Mike B, excellent data and graphing as usual--thanks for sharing.

/agree. The SoC is key for BatV, though this is likely still a little less for p85dl over p90dl as the first set of graphs showed ~10kW power advantage to p90dl. Still, you'd still be within the car length at 1/8th, though imagine this will grow over 1/4 to a car length still. Need to revisit this when temps here are ~70 and then I'll go 95% SoC and MaxBatt and see what we see :)

You're welcome on data. Have supercharger session logged and graphed from 50-90% SoC from same trip. Very different characteristics in terms of charge ramp/drop off. I'm going to see if I can overlay with the CHAdeMO one for grins...

- - - Updated - - -

Btw Did anyone have easy way to check battery model? Might have got lost in the graph deluge
 
Few more graphs on the charging side of things:


First, supercharging session. From 58 to 90% SoC - normally wouldn't charge this much on SC though a) it was completely empty and b) I was logging :)

Initial ramp is as expected, with significant falloff/lower power due to reasonably high SoC as start.
MikeBur P85DL Supercharger charging 58-90 02-29-2016.PNG


Next comparison between this Supercharging session and CHAdeMO. CHAdeMO is longer and although started at lower SoC, still shows benefit of SpC over CHAdeMO. (Dotted is CHAdeMO, translucent lines is SpC)
MikeBur P85DL CHAdeMO vs Supercharger charging 02-29-2016.PNG


- - - Updated - - -

Few animated gifs for charging. I thought these were pretty boring, though some wanted them. Let me know if you want these to be shared in future. Kudos, and paypal donation, to ezgif.com - if you use remember to use optimize for transparency (reduces from many MB to much few KB) and speed up animation (about 6ms between frames seems good).

Animations and source data on Google drive. btw - If you select 1 (instead of 0) for GenImages in the Battery*.xlsm files and click "animate" on the excel tab then per frame gifs will be output into c:\temp\ directory

Supercharging:

Individual Cell Voltage whilst charging
MikeBur Supercharger CellV.gif



Module Temp
MikeBur Supercharger ModuleTemp.gif



- - - Updated - - -

CHAdeMO charging session animated Gifs:


Cell Voltage
MikeBur CHAdeMO Cell Voltage 2-29-16.gif



Module Temperature
MikeBur CHAdeMO Module Temp 2-29-16.gif



Again, if these are not useful let me know. It's a lot easier to *not* generate these ;-)
 
Last edited:
  • Like
Reactions: benjiejr
@MikeB, i sent you email about battery model.

The SpC vs Chademo was interesting and shows the typical CC-CV expected with chademo, tesla has to be different of course...

Good work on the animations, i'll bet everyone will want to check their own car at least once to be sure all the cells are about the same and no laggards or outliers, then it won't be used again unless there are issues or problems develop somewhere along the way. Then it is a great tool to quickly check the state of health cell by cell.
 
Ok, final set of graphs from range mode asks previously. Range mode at 65, 70, 80, 90mph then an attempt at comparison.


65mph Range mode, with bump of 5mph to 70mph via. TACC at the end.
MikeBur 02-29-16 65mph Range Mode.PNG


70mph Range mode
MikeBur 02-29-16 70mph Range Mode.PNG


80mph
MikeBur 02-29-16 80mph Range Mode.PNG


90mph
MikeBur 02-29-16 90mph Range Mode.PNG


Note SoC close, though declining and predominantly flat.

Ok, to comparison. Had to try this a few ways, though settled on putting mph and RPM on same axis (secondary on right) and torque, etc. on left.
Few things to note: 1) Use the line pattern from mph to determine which line pattern for other values, 2) to get to real mph divide by 10, to get to real rear RPM multiply by 10 - tried to make it easy ;-)

Although reinforcing the faster you go the more energy you use, I don't see this as the square, ie.
Mean @65708090
Power (kW)23.8728.8031.4643.96
F_Tq19.4421.0120.7026.00
R_RPM77008298947910654
Deltas65-7070-8080-90
Power (kW)4.932.6512.50
F_Tq1.57-0.305.29
R_RPM59711811176
Temperature was approximately the same. General direction was south. Road was predominantly flat. Atmospherics similar (dry).s
This is a head puzzler for me as the delta from 70 to 80 was a lot less in terms of energy than I expected, anywho the graph is below and data is on the share (created a subfolder called rangedata).
MikeBur 02-29-16 65-70-80-90mph Range Mode v2.PNG


Time to turn in for the night. Hope this stimulates some interesting conversation - cheers, Mike

- - - Updated - - -

oh, sorry - forgot to add as it's implicit. Rear motor Torque was zero for the time I chose - bar the acceleration at 65mph it's mighty impressive how well torque sleep works...

- - - Updated - - -

@MikeB, i sent you email about battery model.

The SpC vs Chademo was interesting and shows the typical CC-CV expected with chademo, tesla has to be different of course...

Good work on the animations, i'll bet everyone will want to check their own car at least once to be sure all the cells are about the same and no laggards or outliers, then it won't be used again unless there are issues or problems develop somewhere along the way. Then it is a great tool to quickly check the state of health cell by cell.

Thanks Kenny. Saw and responded on email.

The battery .xlsm contains macros for animations, and anyone can check the macros for themselves - or let me know and I'll try to animate the data for them (time withstanding). The TMS-Spy might be a better long term solution for general charging info, though LolaChampCar's logger is the real star here - I just make pretty(?) pictures :)

- - - Updated - - -

Wow Mike!
Cross posted to TM.

Thanks Bill. Wouldn't be possible without your HW and FW logger :)

- - - Updated - - -

Apologies, it's late and I messed up on the table a little bit as I (poorly) assumed same timing.

Here's the correct chart:
Mean @65708090
Power (kW)23.4127.8830.1943.91
F_Tq19.1720.5420.0225.99
R_RPM77038300948410656
Deltas65-7070-8080-90
Power (kW)4.472.3113.72
F_Tq1.37-0.525.97
R_RPM59611841172

Note most numbers were only a few thousandth's out, though the Power delta between 80 and 90 mph was significantly off. Anywho, can't figure out to edit inline table in the original part of this post, sorry. Use this one.
 
Loving these animated graphs MikeBur! Your pack looks to be very well balanced. Be good to see something similar from older packs and to see what the impact of age vs mileage is on balancing.
 
? i'm not seeing the TPS on the range mode charts--just curious what the pedal was doing

Here's some power calculations from an online aero and rolling friction estimator:

5100 lbs, 0.24 Cd, 0.01 Crr, ideal conditions

mphkmhaero Froll Ftotal lbs%aero%rollPower Wtotal HPaero hproll hp
60 97 247.60 226.86 474.45 52.19% 47.81% 12,726.02 17.07 8.91 8.16
65 105 290.58 226.86 517.44 56.16% 43.84% 15,035.57 20.16 11.32 8.84
70 113 337.01 226.86 563.86 59.77%40.23% 17,644.90 23.66 14.14 9.52
75 121 386.87 226.86 613.73 63.04% 36.96% 20,577.06 27.59 17.39 10.20
80 129 440.17 226.86 667.03 65.99%34.01% 23,855.11 31.99 21.11 10.88
85 137 496.91 226.86 723.77 68.66% 31.34% 27,502.11 36.88 25.32 11.56
90 145 557.09 226.86 783.95 71.06% 28.94% 31,541.11 42.30 30.06 12.24
 
Last edited:
This is great data guys! Loving the animated graphs Mike.

So let's say I was interested in collecting such data from my car. Is there a post in the last 45 pages of this thread that details what hardware & software I would need to acquire in order to get started? I'm comfortable soldering PCBs and making my own cables, FYI.
 
Last edited:
Although reinforcing the faster you go the more energy you use, I don't see this as the square, ie.
Mean @65708090
Power (kW)23.8728.8031.4643.96
F_Tq19.4421.0120.7026.00
R_RPM77008298947910654





Deltas
65-7070-8080-90
Power (kW)
4.932.6512.50
F_Tq
1.57-0.305.29
R_RPM
59711811176

90 is close to a square from 65. I think 80 is just measurement error.

Could you do some light acceleration and regen let's say 0-30-0 with range mode on/off with f/r torque?
 
? i'm not seeing the TPS on the range mode charts--just curious what the pedal was doing

Here's some power calculations from an online aero and rolling friction estimator:

5100 lbs, 0.24 Cd, 0.01 Crr, ideal conditions

mphkmhaero Froll Ftotal lbs%aero%rollPower Wtotal HPaero hproll hp
6097 247.60 226.86 474.45 52.19% 47.81% 12,726.02 17.07 8.91 8.16
65 105 290.58 226.86 517.44 56.16% 43.84% 15,035.57 20.16 11.32 8.84
70 113 337.01 226.86 563.86 59.77%40.23% 17,644.90 23.66 14.14 9.52
75 121 386.87 226.86 613.73 63.04% 36.96% 20,577.06 27.59 17.39 10.20
80 129 440.17 226.86 667.03 65.99%34.01% 23,855.11 31.99 21.11 10.88
85 137 496.91 226.86 723.77 68.66% 31.34% 27,502.11 36.88 25.32 11.56
90 145 557.09 226.86 783.95 71.06% 28.94% 31,541.11 42.30 30.06 12.24

thanks Kenny. TPS appears as zero because I'm using TACC -lol.

love the table. So I should read this as 15, 17.5, 23,8, 31.5 kW as the estimated for 65,70,80,90 respectively? Maybe 90 was up more of an incline than I believed?

- - - Updated - - -

This is great data guys! Loving the animated graphs Mike.

So let's say I was interested in collecting such data from my car. Is there a post in the last 45 pages of this thread that details what hardware & software I would need to acquire in order to get started? I'm comfortable soldering PCBs and making my own cables, FYI.

thanks Marc. Definitely a team effort collecting data and visualizing between a bunch of us. LolaChampCar has a number of loggers he's generously loaned out, so likely a PM there will kick this off. No soldering read. Usual community-driven expectations around sharing, etc etc ;-)

- - - Updated - - -

90 is close to a square from 65. I think 80 is just measurement error.

Could you do some light acceleration and regen let's say 0-30-0 with range mode on/off with f/r torque?

You didn't read to the bottom of the entire post carefully enough! :p. Heh, that's the wrong table with correction at end. Even so, the more I look at the correct results, the more I'm believing I need to control the experiment more - i.e. Exactly the same stretch of road. Ugh.

There are a number of range mode graphs earlier in the thread, with more in the range mode directory in excel form. Any chance you could see if one of this suffices first? If these font work, I'll fetch it for you.
 
... TPS appears as zero because I'm using TACC -lol.

So I should read this as 15, 17.5, 23,8, 31.5 kW as the estimated for 65,70,80,90 respectively? Maybe 90 was up more of an incline than I believed?

Thanks Mike, i didn't realize about the TACC and TPS. Now i have to wonder what command signal is being used for speed control during TACC when TPS is out of picture.

Yes that's the ideal power required for those speeds--add any hills or winds and the power requirement goes up quickly.
 
90 is close to a square from 65. I think 80 is just measurement error.

Could you do some light acceleration and regen let's say 0-30-0 with range mode on/off with f/r torque?

reading my post again, I might have come across as snarky. Not the intent, rather overloaded myself.

I believe 80mph was likely down hill and 90 was uphill slightly, the more I look at the data vs the estimates Kenny provided.

Try these links for more info on range mode:
Chassis CAN Logging To ASCII Text Plus Graphing - Page 24
Chassis CAN Logging To ASCII Text Plus Graphing - Page 27
And, my personal favourite: Chassis CAN Logging To ASCII Text Plus Graphing - Page 29

Also, direct link to range mode data sources: Range Mode

This help?
 
Next comparison between this Supercharging session and CHAdeMO. CHAdeMO is longer and although started at lower SoC, still shows benefit of SpC over CHAdeMO. (Dotted is CHAdeMO, translucent lines is SpC)
Thanx for the graphs Mike. If you take suggestions: we know SpC offers a higher power and current than ChaDeMo, so obviously it will charge faster. Maybe a good comparison would be SoC on the x-axis vs. charging power (V*I since this is DC) for both SpC and SoC? Altentatively (or in addition to) you could put Vbat on the x axis instead of SoC (I wonder whether SoC is derrived purely from Vbat open circuit, if it is, graphs should be identical shape). :)
 
Thanx for the graphs Mike. If you take suggestions: we know SpC offers a higher power and current than ChaDeMo, so obviously it will charge faster. Maybe a good comparison would be SoC on the x-axis vs. charging power (V*I since this is DC) for both SpC and SoC? Altentatively (or in addition to) you could put Vbat on the x axis instead of SoC (I wonder whether SoC is derrived purely from Vbat open circuit, if it is, graphs should be identical shape). :)

sure, it will be later tomorrow before I can get to this though sounds like an interesting alternative. These charts are on the Google drive if you want to play with them before then. Power is already on the chart, and I purposely left I and V separately there as for other ways to rep this :)
 
... you could put Vbat on the x axis instead of SoC (I wonder whether SoC is derrived purely from Vbat open circuit, if it is, graphs should be identical shape). :)

From looking at the previous charging data from lolachampcar i would guess that SoC is calculated by integrating the current up until it reaches the 93% mark. Then the current gets strongly tapered and controlled with a fine-toothed filter with an emphasis more on voltage than current, and the balancing circuits on the BMU boards start to come into play. Given a long enough time at the SpC station it will try to fill the cells and will definitely need to use OCV (by cutting off the current momentarily) as it gets near the critical top voltage. Otherwise it can read the OCV at any time when there is no charging or discharging load and calculate/report the SoC based upon that value and the integrated current measurement. Just my 2¢, kb
 
Guys, what a fantastic thread! I didn’t read it until this past weekend, but having done so now I think the collected data is fascinating. What I would love to do – although I’m not sure I have the competence to do it – is develop a dragstrip model/simulator that would predict how times would vary based on all the relevant variables. It would begin with electrical inputs, do some empirical fitting to estimate the conversion to mechanical power/thrust, then calculate (drag, rotational inertia of wheels & tires) or empirically fit (driveline friction, rolling resistance) factors accounting for retarding forces, and integrate acceleration runs with the goal of reproducing the recorded data. Then electrical and other parameters can be varied to see what it would take to do a 10.9 run. Or we could estimate what the P100D would run based on lower pack resistance. Or how the lower moment of inertia of carbon fiber wheels would affect our results.

Of course the devil is in the details and I am not sure I will get very far at all: I am no engineer, my pre-retirement career was in financial services, and it’s been a long time since I was a physics major. What I have done so far is take Bill D’s 2016-03-01 run 1 spreadsheet – which I like because it’s a full quarter mile run – and import it into Mathematica 10. The first step is to understand the conversion from electrical power into mechanical power. Rather than divide one by the other to get percentage efficiency, I think it’s instructive to take the difference and look at the actual conversion dissipation in watts. Below, blue is electrical power (v*I), brown is mechanical power (summed torque in newton-meters times rear axle angular velocity in radians per second – all data converted to SI units) and green is the difference. All are in watts. I did not spend much time making the chart pretty.


BillD QM.jpg


Because I add the front and rear torques when calculating mechanical power I am assuming the two motors are turning at very close to the same speed, so the two motors are geared the same. The MS owners manual implies this when it claims a single final drive ratio of 9.73:1. I checked this for the rear by calculating the diameter of 245/35-21 tires and calculating MPH per rear motor RPM given 9.73:1. The result is interesting because the RRPM data fits the MPH data very closely after twelve seconds when the car is coasting. During the run, rear motor RPM was 3% too high about two seconds in, followed a curved descent to 1% too high at the end of the quarter mile. In other words, it shows the tire slippage one would expect under hard but decreasing acceleration, where slippage goes slightly the other way once decelerating. It raises the question of how MPH is measured. Personally, I suspect it’s integrated from an accelerometer, which is periodically calibrated against motor RPM when the car is cruising without accelerating.

Turning to the chart, note the 40 kW of static dissipation in launch mode. That’s a lot of heat into the alternator power semis, and the coils in the stator and rotor. There is a blip in dissipation just after launch but if you take out the blip there is a ramp from the pre-launch level up to roughly 7 seconds in where it levels off at 100 kW and stays flat until 12 seconds when the run ends. Notice how the leveling off in dissipation coincides with the knee in the blue electrical power curve, which results from a knee in current. My suspicion is that the current knee is programmed at that RPM so that heat dissipation is held to a 100kW limit, which is an awfully round and deliberate-looking number.

One factor that might explain some of the blip after launch is that electrical power not only has to produce useful torque, it also has to spin-up the rotor itself, which is a hunk of metal turning 14,000 RPM at the end of the run. That takes a lot of power, and the angular acceleration is highest early in the run where the blip is. If any of you guys have a guess for the weights and diameters of the two rotors I could treat them as solid cylinders, calculate their moments of inertia, figure out how much power is needed to spin them up, subtract that from the dissipation to get the real dissipation as heat, and see how the result fits to motor torque and speed. Or at least that would work if the reported motor torque is at the output shaft, net of the torque needed to spin the rotor. If reported torque is a gross number based on power and phase between rotor and stator (and I am no expert on induction motors) I would have to add the rotor moments to the tire and wheel moments (axle moments are insignificant since the axles are narrow) instead. The effect of rotational moments of inertia is to increase the effective mass of the car, so they need to be taken into account.

I’ve made this post long in part to give a sense of some of the complications (devil in the details) necessary for a drag strip simulation. Any comments from the engineers among you are very welcome.

Eric


Edit ... Looking at photos of the large-motor rotor it's about 6" dia and a foot long, so 42 Kg at the specific gravity of Iron. Radius of gyration 0.053m. Moment of inertia 0.12 Kg m^2. Probably 0.2 for the two motors together. Recorded angular acceleration over the first second is 324 radians/s^2. Rotational kinetic energy is thus ~10,500 Joules after 1 second, so 10 kW of the difference between electric and mechanical power during the first second is spent spinning up the rotors. Explains about half of the roughly half of the 20 kW jump in the difference right at launch, which sustains given fairly constant angular acceleration in the first second. Or it might explain the whole jump if my eyeballing the rotor size is merely 15% low -- in scaling the cylinder proportionately, mass goes as the cube and moment as the 5th power. So it probably does account for the whole 20% jump in dissipation right at launch.
 
Last edited: