P mode vs Park mode

[Mario Veras]

What’s the difference between putting the Model 3 in P versus Park mode? I’ve seen some YouTubers talk about that it makes no difference, so why is there a logo displayed on the screen if it makes no difference?
I can’t seem to find this in the manual.

 

[gearchruncher]

It's too bad this guy didn't hold down P for a lot longer:

I hear if you type in "ICE" in morse code into the P button it won't do this though.

 

[mtmra70]

Don't be silly. This happened because the USS were disabled.

 

[DownshiftDre]

Nice data. Anybody think there is a time element to this? Noticed on schematic that there is an APPLY command and RELEASE command from the left and right body controllers. Does the APPLY command get commanded for a specific length of time based on different modes? The RELEASE may just be a single one time 50-100ms command to release entirely (depends on implementation within EPB). Didnt see a currrent feedback anywhere in the schematics. So doesnt look like an LVDT or RVDT implementation current feedback is implemented based on position feedback.

The voltage of 13.88 looks to be the entire system voltage on the LV side. Every command in the car will probably show around 13.88 if thats what your LV bus is at.

I think next step would be to show if the command profile all look the same for each mode?

Very much interested in your responses and Im all in to get to the bottom of it.

 

[DownshiftDre]

So, here's my data collection plan for the community to review:
1) Collect current vs time data when going into park from neutral
2) Collect current vs time data when going into parking brake (press and hold) from neutral
3) Collect current vs time data when going into parking brake (via touchscreen) from neutral.

If there is a difference in the stall current amount or duration between #1 and #2 or #3, then there is a difference. If there isn't, then there's not.

Agreed with approach. This is key.

 

[DownshiftDre]

Also compared implementation of EPB to the power window motors and the power window implementation is much more complex with motor drive commands, and hall sensor feedback. Same with mirror motors. The EPB APPLY/RELEASE philosophy Tesla has implemented is very crude and simple so hopefully this will be an easy one for us to crack implementation wise.

 

[freed0m]

Same for me.

Turn your wheel all the way to the right...haha

 

[Gauss Guzzler]

No "Regular" voltmeter updates at 4,000Hz. Point me to one that does. Even if it did, your eye couldn't see it, and no LCD is that fast. Just how the heck do you think someone can use a voltmeter to repeatably measure and see the stall current phase of something that only lasts 200ms?

Huh? I never said any voltmeter display could update at 4kHz nor that you should watch it with your unblinking eye. I just assumed you were familiar with common voltmeter functions and would be able to locate the min/max button to capture the peak. Most cheap household meters have a peak capture function but since you said you're an EE I assumed you would have a basic Fluke or something with the option to go to 4kHz, even if the standard 10Hz capture would surely be more meaningful.

Nice data. Anybody think there is a time element to this? The RELEASE may just be a single one time 50-100ms command to release entirely (depends on implementation within EPB). Didnt see a currrent feedback anywhere in the schematics. So doesnt look like an LVDT or RVDT implementation current feedback is implemented based on position feedback.

Recall that Tesla's "fuse-less" electrical system relies on current feedback from every subsystem. That top-level schematic is just saying that each brake is connected to an H-bridge on its corresponding body side controller. It doesn't get into the board level components.

Reliably releasing the brake is an interesting challenge indeed. This could be done with a timer or by watching for a certain current pattern, but I'd guess a microcontroller or something in the caliper inserts a diode into the system after a certain number of jackscrew revolutions.

 

[gearchruncher]

but I'd guess a microcontroller or something in the caliper inserts a diode into the system after a certain number of jackscrew revolutions.

Geez, you just can't help yourself making up stuff, can you? Anyone that has actually worked on the brake system can tell you there are only two wires going to the caliper, it runs direct to a DC motor, has no active electronics inside, and Tesla's own process for backing off the caliper to do a pad replacement just involves applying reverse power from a 9V battery until the internal jackscrew is unwound all the way.

Internal microcontroller with an encoder to count revolutions and short out the external current controller that magically keeps track even when it has no power and shorts out it's own power? What a yarn....

 

[Gauss Guzzler]

Yes, I know there are 2 wires and that the H-bridge is in the body controller so these wires are likely just simple reversable power. I never "made up" anything otherwise.

Inside the housing there could be a rectifier to power a microcontroller which counts revolutions and flips a persistent transistor such that a diode blocks the motor from further retraction with negative voltage, yet still allows compression with positive voltage. This is a trivially simple circuit that almost any EE could sketch on a napkin after 16 beers within a few seconds. I'm surprised that you describe it as "magic".

I also wrote that it could be something other than a microcontroller. For example, a mechanical limit switch could flip a diode into the circuit after some distance to accomplish the same thing.

Did you say that you're an electrical engineer?

Anyway, the brakes release pretty quietly so it doesn't seem that they're just driven all the way to a current-limited hard stop. And it's really important that they retract reliably at any temperature or state of disrepair so a simple timer or current profile is unlikely. Plus, the mechanism is pretty slow so if it retracts too far the car could roll downhill quite a ways before the brakes engage. Put the car in Park *while* driving very slowly (<1mph) on a steep hill to see how far it freefalls and then imagine how much further it might go with overly worn brake pads after a full retraction of the jackscrew.

 

[DownshiftDre]

EPB caliper has no comms back to body controller, so the EPB caliper (no matter how simple or complex) will not know if the driver pushed P once or held P. That smarts is in body controller and if there is a difference, it should come out in the 2 signals outputs.

If there was a LIN or CAN bus wire here, different story. But there isnt, so safe to assume any smarts around P push vs P hold would reside in the body controller and if there was a difference, the effect would be seen in variation of the signal output.

 

[gearchruncher]

Ok. I went and got more DATA. Like a real EE will do, instead of using a "simple voltmeter" to find the peak, I used an oscilloscope (that I designed myself when I worked at a T&M company) to look at the signal going to the EPB motor.

It's complicated. Here's what the trace looks like. The first section is the retract, the second is an activation using park.
(Can you imagine trying to diagnose this with a peak hold voltmeter!???)

Those are NOT PWM pulses. There is 3.0 seconds between the start of the retract and the end of the apply. Every pulse is 14ms on and 3ms off, and they don't change in duty cycle or frequency at any point. The first and last ones are the exact same timing, Plus, 17ms is only 58Hz, which is way too slow for any kind of useful PWM current control.

I'll admit, I don't exactly know what is going on here, but I have a theory. They are sensing the *speed* of the motor using back-EMF. They run the motor for 14ms at full bus voltage. They then turn off the drive and check the back EMF voltage to sense RPM. You can see that during the retract, the beginning has a low voltage signature until it starts spinning, and the clamp has a low voltage signature when at the end when it comes to a stop. Also, look at the signature as it retracts and runs down at the end as it slows to a stop (but thinking about back EMF is just brining quantum mechanics into a simple, low speed, brushed DC motor, amiright?)

I tried to do this in current mode as well, but when I put the current shunt I had inline, it messed up the whole logic. The motor would run WAY longer, and when it clamped, it wouldn't detect the clamp had happened. The system would peak at over 15A, and actually melted one of my wires. It might eventually time out, but I never let it, I always yanked a wire. I'm sure with a really nice setup with custom harnesses and a very small shunt you could make this measurement, but all I have right now is jamming wires into some connectors since I don't have the mates. It is possible they are *also* doing current control within these pulses. But the question remains, why bother to pulse unless you are doing something interesting in the off time? Also, why does it peak at >15A if there is closed loop current control when normal operations are under 5A?

My theory is that the RPM is a much more stable signal than current or anything else. For instance, the back EMF wouldn't be impacted by temperature, since it's a zero current state and the change in copper resistance would be irrelevant.

I'll go out later and get these clamp traces in the 3 different activation modes and we can see if they look any different.

 

[gearchruncher]

On a side note, please don't believe people that tell you there are smarts in the caliper motor. It requires a fundamental lack of knowledge of how parking brake systems work, and how they adjust for pad wear to suggest otherwise, and just how far they move relative to that pad wear in a normal park cycle.

There are zero switches, diodes, or microcontrollers in there. There are no limit switches. This is easy to prove, since the procedure to actually swap brake pads requires a service tool to actually back the caliper off all the way. The above 1 second reversal only backs it off enough to not drag the pads. It takes about 15 seconds with the tool to actually reverse the brake fully, and when it hits the end, the motor keeps spinning because the jackscrew just runs off the end of the threads.
The first step in the service instruction is to release the parking brake using either toolbox or the "special tool", which is just a power supply and some wire harnesses: https://service.tesla.com/docs/Mode...UID-C039ADBE-F3D8-4FFA-BC8E-9D7B9F241978.html

If there were smarts or limit switches in the EPB, nothing external would be able to retract it. Yet you can just hook a 9V battery up and retract it all the way, much, much, much farther than it retracts in normal use.

 

[DownshiftDre]

Really great stuff! Thank you for this and very noble of you to get this put together for the purpose of Friday afternoon science. In the end we all win with this data.

I do see the profile of the Apply voltage change near the end as a means of detecting when the clamp is starting to slow down as the pad builds up pressure once it hits the rotor. Crude feedback to works good enough to squeezr something tight. And Self adjusting for temperature, pad wear, and many other items.

It will be very interesting to see these tests from Neutral (N).

Please be safe in ensuring your car is chocked and jacked up correctly. Again thank you.

 

[TexSam]

Lots of good information here BUT I would suggest to check the 5th post in this thread too. And view the second video referenced. Good stuff.

For some reason the link to the video doesn’t seem to play the video, only the sound for me. But if you go to the 5th post it seems to play fine for me. Be sure you click on the second video.

 

[gearchruncher]

Lots of good information here BUT I would suggest to check the 5th post in this thread too. And view the second video referenced.

Like has been brought up many times, the question here is if normal Park is different than the "Parking Brake." None of this has anything to do with what the car does if you press the P button in motion, which easily could be a quite different behavior, and nobody is denying exists. However, none of those videos have any data that what happens in motion is a different clamping force than if you do it stopped either.

 

[mtmra70]

Ok, I'm scratching my head. You show a retract and a park command.

How come not:
-D to P
-D to P to parking brake
-N to parking brake

 

[gearchruncher]

I tried to do this in current mode as well, but when I put the current shunt I had inline, it messed up the whole logic. The motor would run WAY longer, and when it clamped, it wouldn't detect the clamp had happened. The system would peak at over 15A, and actually melted one of my wires.

I'll go out later and get these clamp traces in the 3 different activation modes and we can see if they look any different.

Ok, I'm scratching my head. You show a retract and a park command.

How come not:
-D to P
-D to P to parking brake
-N to parking brake

Like I said, I will get those later. When I tried to get current data, I melted my setup. I ran out of time last night to rebuild my setup and get all that data.

 

[mtmra70]

Ah, got it!

 

[Gauss Guzzler]

Great stuff @gearchruncher! I like your back EMF theory, that would be a really clever and easy (cheap) way to guesstimate the retraction position and perhaps even detect issues along the way. It looks like your scope data alternates between positive/negative voltages which also seems to support your postulate, although I would expect to see a clear taper in the back EMF voltage as the motor grinds to a halt.

I’d expect the system to be really robust and tolerant of impedance changes so I’m surprised that your current shunt threw it out of whack.

 

[T4M3]

Really great information to everyone, thanks, I learned a lot on this thread.