Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
  • Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
  • The final cut of TMC Podcast #29 is available now with topics time-stamped. We discussed the Tesla Cybertruck's expected 1 MW Ultra-Fast Charging capability, the Tesla Semi Delivery Event, the coming Model 3 refresh (project "Highland"), and more. You can watch it now on YouTube.

Testing for Input lag (both touchscreen and pedal) in a Tesla Model 3 Performance

This post below is a heavily condensed version of the article here, which features more graphics, information and background.

I wanted to test for input lag in its various forms on my Model 3 performance and scrutinized not just the hardware angle (pedal lag & gear stick), but also Tesla's touchscreen to see how responsive the various buttons, menus, sliders and scroll widgets are. Tesla is renowned for its snappy interface, at least on its newer cars, but is there any room for improvement?

Pedal lag

In an old post, some users on this forum noticed a difference in latency between the Roadster and Model 3 pedal's response, so this is the first thing I looked at. We're looking explicitly for the time gap between pressing on the accelerator pedal, and when any amount of power is applied physically, i.e. when the car starts to move. (This doesn't necessarily equate to throttle response on ICEs!). With electric cars, this may even become primarily a software problem, rather than necessarily a hardware issue (where motors are renowned for their 'instant torque'.

Anyway, on a low-speed wind day in a safe area, I placed a modern phone camera (capable of 60 frames per second) resting in the driver side footwell, placed it against the side, and used the mat to secure it in place (see photo). I then strategically angled the phone so that it has a simultaneous view of the aluminium pedal AND the outside, assuming the door is left open:

r7tQL4B.png


Here is the video from one of the multiple tests I did.

Zooming into the video, and finding the colour difference between images, I was able to nail a latency down to around 50 milliseconds. This represents an upper limit as noise becomes an issue at these precise levels of time and resolution.

50 milliseconds is not a bad result! Quite a good one actually, and better than I originally expected. Room for improvement? Sure, though it's tricky to gauge how noticeable it would be. If we take a look at this video from a Microsoft researcher, we can see that even 10 milliseconds is noticeable in some situations, and 1ms is close to ideal.

Touchscreen latency

Secondly, I tested to see how responsive the touchscreen was. Hardcore gamers who try to shave milliseconds off their input lag during play will know what I mean: You feel more connected to the game, and it generally makes for a smoother and more enjoyable experience.

Unfortunately, touchscreens are particularly susceptible to lag, and the aforementioned video is well worth a watch to demonstrate how even 10 milliseconds can be too much.

I tested various often used buttons (200ms lag), the Climate button (185ms lag), map scrolling (110ms lag), Menu tab changing (185ms lag), the 'Start' button (170ms lag), temperature changing (205ms lag), window scrolling (105ms lag) and the keyboard (100ms lag):

Aex5I52.png


The original videos are much longer and higher quality. Overall, I found a latency of around 100-200 milliseconds which isn't great to be honest. Current touchscreen technology is at least partially to blame, but Tesla can immediately improve the experience on many occasions by at least 50 milliseconds if that click is registered at the moment the finger touches the screen ("finger-down"), RATHER than when it leaves the screen ("finger-up"). This won't be ideal for some elements that are immediately draggable, but for controls like simple buttons, opening windows/menus, and changing the temperature, the opportunity is there to shorten the perceived lag by around half!

Testing startup to Drive mode latency

Finally, some users (including myself) have noticed that the car frustratingly doesn't immediately allow you to switch into the Drive gear after you press the foot brake pedal. There's a short delay that you need to wait out before adjusting the gear lever.

To test for latency, the obvious idea is to keep hammering the gear lever (approx every 0.2 sec) until we see the Tesla show the 'D' icon (Drive) change from the 'P' icon (Park). Unfortunately, for some reason, the car doesn't like this very much and as a result just causes further delay (around 3.5 seconds based on this video before allowing you to switch into Drive).

Better to use a higher delay (say 0.4 seconds) between gear lever presses. To accurately keep interval timing, I used a metronome and timed my gear lever presses to the sound of the metronome beat, with the first press to happen simultaneously with the foot brake push (and hold). Using this technique, I found the vehicle switched to drive mode on the 3rd lever pull at around 0.8 seconds.

Experimenting with other metronome timings, 84bpm specifically (which failed on the second lever pull), I found a lower bound of 0.7 seconds, so we can estimate the true lag before the car allows you to switch to Drive mode is around 0.75 seconds. This is sometimes a painful wait, especially if you pull the gear stick too early, and find it doesn't switch at all.

Summary and conclusions

We found a pedal lag of around 50 milliseconds, a touchscreen lag from around 100-200 milliseconds and a "preflight check wait to allow Drive gear" wait of around 0.75 seconds (750 milliseconds). The above three summaries are the areas where I think Tesla can most improve in regards to latency. I hope Tesla takes note of these concerns to help make driving the car a more pleasurable and 'snappy' experience!

Full article here.
 
Last edited:
The Ryzens are supposed to be noticeably quicker on touch and general interface reaction times but I have no idea if there has been any change to throttle lag.

For me throttle feels gloriously fast, coming from a CVT+Turbo car the M3P feels like a lit rocket being held in place until you press the juice pedal haha
 
The Ryzens are supposed to be noticeably quicker on touch and general interface reaction times but I have no idea if there has been any change to throttle lag.

For me throttle feels gloriously fast, coming from a CVT+Turbo car the M3P feels like a lit rocket being held in place until you press the juice pedal haha

Right, I think the throttle has a near direct connection to the motor which bypasses the main infotainment computer, otherwise the car would be inoperable if the screen crashed during driving!

like a lit rocket being held in place until you press the juice pedal haha

And then you get the nitrous hit :D
 
Right, I think the throttle has a near direct connection to the motor which bypasses the main infotainment computer, otherwise the car would be inoperable if the screen crashed during driving!
Yeah, throttle definitely isn't processed by infotainment. I imagine it's pretty direct to the ESC for that exact reason.

And then you get the nitrous hit :D
YEEEEET!
 
Perhaps not related to your test, but I felt my M3LR has a noticeable delay to start driving upon cold "startup" in my garage. I'd shift into Drive, and step on the accelerator only to find it being sluggish to move. Same with reverse.

Once the car is fully "warmed up" or driven, this lag doesn't exist and is more responsive.
Even fully warmed up, the car's tactical applications are limited. Stepping on the accelerator in reverse (to perform a J-turn) shows noticeable lag and then the ridiculously low speed limit are huge bummers.
 
Nice testing and writeup @dspwhite! I'd certainly expect the Gauss pedal latency to be around 50ms but I sure didn't expect the car to move from a stop that quickly! The brakes have to be released first!

I'd be really interested to see you analyze the turn signal latency issues. A very long press is needed to trigger the signal and only a slightly longer press to trigger a single blink. And curiously this latency seems completely different from the gearshift which detects input no matter how brief, and presumably with the same hardware. You tested the "startup" latency from brake pedal-to-gear selection which is likely a hard-coded timer to allow ample time for all critical systems to report safe-to-drive status.

I'd guesstimate that the turn signal has to be held for >250ms to trigger a response, but less than 1000ms to actually latch into triple or continuous blinking rather than simply releasing after 1 blink.
 

About Us

Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.

Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


SUPPORT TMC
Top