Quad motor vs locking diff

[Inkedsphynx]

This is probably totally bonkers but this thread still has me trying to think up novel ways you could use the technology to improve the situation.

This idea is very specific to a very specific set of circumstances so obviously doesn't address everything in this thread, but...

What if there was an ability to individually test each wheel's traction in a stopped position? You're stopped on a trail that's off-camber and before you proceed, you enable some mode that runs a traction test individually on all 4 tires. Slowly ramps up torque until the wheel starts to slip, then stops. Now it knows how much traction that specific tire has. Do that all the way around the vehicle, and now you know the traction situation and could modulate the motors in some fashion to account for that.

I guess it'd be mostly pointless since that all goes out the window once you start moving, but for those off-camber situations, you're likely riding over the same/similar conditions the entire way, so it's really about the vehicle being able to know which tires are more likely to slip and accounting for that. Maybe this would need to be coupled with a special mode for this specific situation.

I don't know, this could be entirely pointless, since the conditions could change. This obviously doesn't help if you're already moving. My brain doesn't like unsolved problems

I do enjoy though that the tech allows me to think up these crazypants ideas. The one big benefit the quads have over traditional lockers is this flexibility, so it's just a matter of whether or not a system can be reasonably designed that can perform the same functions as lockers.

Sponsored

 

[racekarl]

No, not with a locked differential. A locked differential reads nothing. It's mechanically locked and that's that. The same exact thing can easily be accomplished with a motor at each wheel. Make them turn at the same speed no matter what.....and that's essentially the EV version of a locked differential.

What you're talking about is more of a reactionary type system, which is in fact much more common in today's 4wd (and even more so in AWD) vehicles, in which loss of traction is sensed and then the system reacts by apportioning torque to where it can benefit most. That also can be done with a quad motor system.

This is the same misunderstanding that you've been having throughout this thread. Notice that I never once mentioned locked diffs, and for that matter neither did R.I.P. in his initial description of the issue.

The behavior I am describing is much closer to how an open or limited slip differential functions, which is still an important shortcoming of the current Rivian quad motor system. These differentials do not "sense" anything, the power flows through them as a natural consequence of physics.

 

[R.I.P.]

Given time and money, in theory accelerometers and cameras (perhaps even the existing ones) could be used to augment a traction control/locking diff algorithm. Lots of time, and lots of money - but it's a use case that few will use, not worth Rivian's investment.

I'd also be concerned about software flaws. We've seen major issues including fatalities with Tesla FSD. Rivian Driver+ has some "near miss" reports of erratic steering adjustments, etc. A little miscalculation in "Off Road Locking Diff Assist" mode could be disasterous...

Really good points. The more complicated we make things, the more we are at the mercy of those systems when they fail. I am not opposed to FSD, but it has, and will continue to kill people. There is not a simpler path forward that I see, so I see FSD development continuing.

My two-wheel-drive Cadillac can make it up my driveway (2.5 miles, thousand foot drop-offs) most of the year. When it is too slippery to make it, it gets stuck. What it does _not_ do is freak out and slide sideways toward the drop-offs trying vainly to put power to all four wheels. That could quite literally get me and my family killed. It is a failure of technology. Even if this were "solved" with lots of code as you suggest, a computer failure could still send me over the edge. As common as computer failures are, that does not give me comfort.

Having learned my lesson (hey, I really wanted to love this tech), I am today looking at possibly getting a Lightning to tide me over until Rivian builds me an R1S with differentials in it. I am not crazy about a number of things on the Ford (pack design being one of them), but it at least has two differentials, traction control and a locker in the back. Ironically, that should give it far more usable traction than the more "offroad" touted R1T.

We shall see.

 

[Zoidz]

Really good points. The more complicated we make things, the more we are at the mercy of those systems when they fail. I am not opposed to FSD, but it has, and will continue to kill people. There is not a simpler path forward that I see, so I see FSD development continuing.

My two-wheel-drive Cadillac can make it up my driveway (2.5 miles, thousand foot drop-offs) most of the year. When it is too slippery to make it, it gets stuck. What it does _not_ do is freak out and slide sideways toward the drop-offs trying vainly to put power to all four wheels. That could quite literally get me and my family killed. It is a failure of technology. Even if this were "solved" with lots of code as you suggest, a computer failure could still send me over the edge. As common as computer failures are, that does not give me comfort.

Having learned my lesson (hey, I really wanted to love this tech), I am today looking at possibly getting a Lightning to tide me over until Rivian builds me an R1S with differentials in it. I am not crazy about a number of things on the Ford (pack design being one of them), but it at least has two differentials, traction control and a locker in the back. Ironically, that should give it far more usable traction than the more "offroad" touted R1T.

We shall see.

I work in factory automation and control systems. My favorite saying?

"Just because you CAN automate something does not mean you SHOULD."

 

[MaskedRacerX]

Imagine if someone took a chainsaw, didn’t start it, and began bashing the bar and chain end against a tree. After a while they’ll declare an axe victorious at cutting down trees.

Dumb old country boy comes into the town general store for the first time, picking up some supplies, owner comes up to him ...

"Howdy! New in town?"

"Howdy Mister! Sure am, I bought that Johnson place up yonder, trying to make a go at selling wood."

The owner replies, "Sounds like a good plan, lots of folks need wood around here, how's it going?"

Dumb old country boy replies:

"Not so good, it's harder work than I figured, my pa's axe is slow doings."

Owner, shocked that he's using an axe says, "Let me set you up, you can pay me later." Goes out back, get s a nice new chain saw he just gassed up, hands it to the kid, says, "This will make that work much easier".

The kid, excited things would get easier, takes his supplies, new chain saw, leaves.

About a week later, the get comes back, frustrated look on his face, "Mister, this thing made it even harder, I had to switch back to my axe"

The owner, rightfully confused, says, "Dang, really? Let's go out back, and test it out!" The kid, says, sure, they head out back.

The owner, pulls the cord, 1-2-3 times, the chainsaw fires up.

The kid jumps back, shocked, "WHAT THE HECK IS THAT NOISE?!?!"

 

[R.I.P.]

This is the same misunderstanding that you've been having throughout this thread. Notice that I never once mentioned locked diffs, and for that matter neither did R.I.P. in his initial description of the issue.

The behavior I am describing is much closer to how an open or limited slip differential functions, which is still an important shortcoming of the current Rivian quad motor system. These differentials do not "sense" anything, the power flows through them as a natural consequence of physics.

EXACTLY! Maybe it would help if I were to say hey, lets take the lockers completely out of the equation. I don't want lockers, not in that rig, all I want is for it to do what a 200-year-old differential does and allow static rolling traction on the two wheels with the _most_ traction at all times. Does that help? The system would have to have a way to know which two wheels have the most traction _at all times_, and then never apply any torque, regen or resistance of any kind to those wheels.

How is that, does that get us away from the lockers and back on the problem?

 

[R.I.P.]

I work in factory automation and control systems. My favorite saying?

"Just because you CAN automate something does not mean you SHOULD."

... which is why my actual off-roader has zero computer control. I can manually lock, unlock or LS to any degree on any of the differentials, giving me absolute control over where the power is going to go. There is no guessing, and no surprises.

Now, to the point of others on the forum, there is a place for automation. My wife's Model Y is a beast in the snow and ice, incredibly predictable and controllable, and that is exactly what I want for her. It is a well done mix of old-school (differentials) and of course Tesla motor and traction control.

 

[R.I.P.]

Dumb old country boy comes into the town general store for the first time, picking up some supplies, owner comes up to him ...

"Howdy! New in town?"

"Howdy Mister! Sure am, I bought that Johnson place up yonder, trying to make a go at selling wood."

The owner replies, "Sounds like a good plan, lots of folks need wood around here, how's it going?"

Dumb old country boy replies:

"Not so good, it's harder work than I figured, my pa's axe is slow doings."

Owner, shocked that he's using an axe says, "Let me set you up, you can pay me later." Goes out back, get s a nice new chain saw he just gassed up, hands it to the kid, says, "This will make that work much easier".

The kid, excited things would get easier, takes his supplies, new chain saw, leaves.

About a week later, the get comes back, frustrated look on his face, "Mister, this thing made it even harder, I had to switch back to my axe"

The owner, rightfully confused, says, "Dang, really? Let's go out back, and test it out!" The kid, says, sure, they head out back.

The owner, pulls the cord, 1-2-3 times, the chainsaw fires up.

The kid jumps back, shocked, "WHAT THE HECK IS THAT NOISE?!?!"

HEY! I resemble that remark!

But on a different note, now do you like your 4xe?

 

[racekarl]

What if there was an ability to individually test each wheel's traction in a stopped position?

Yeah this is where I was going - it's not so much "stopped traction" that you need to measure as static friction.

Tires, even when they are rolling, are encountering static friction; when they break loose and slide across the ground (as in a skid, or locked brakes, or a burnout) they are encountering kinetic friction.

The static fiction of a tire on the ground is almost always higher than the kinetic friction (this is why "threshold braking," which is the maximum braking force before the wheels lock up, will stop faster than locking up the brakes).

A pure open differential will essentially (oversimplifying here) send power to the tire with the least static friction (the wheel that is easiest to turn). As a consequence the other tire remains in its static friction "envelope" and does not break traction with the ground.

The challenge with the quad motor design as it is currently formulated is that it "tests" static friction by exceeding it. It applies power to a tire until it exceeds static friction and enters kinetic friction (wheel spin) then detects it and reduces power. This is totally fine 99% of the time, but there are conditions in which moving into kinetic friction is very undesirable (e.g. moving sideways across a slippery surface, where gravity would be able to move the car if it enters the kinetic friction zone)

To overcome this you need to be able to measure static friction in way that ensures that you do not exceed the static friction envelope of the tire. This is the key enhancement that would be needed to make the Rivian quad motor system work in the situation described in this thread (NOT the ability to to "lock" the motors together at the same speed, though that could be helpful in other situations).

 

[Nix]

It really depends on how the control is implemented. ICE motors are generally driven based on torque control and requested torque. You can take this same idea and implement it with EV motors but it will have extremely delayed reaction to tire slip in the way that you are suggesting.

Oversimplifying the issue, in that situation, the inverter is told to produce X amount of torque and will rapidly increase motor speed to maintain that torque when a tire slips. Typically in the span of a second, an inverter is capable of sending somewhere on the order of ten thousand different "commands" to the motor. If you're waiting to react to readings that say your tires are spinning faster than you want them to, you've already missed hundreds of opportunities to reduce the output. You can have limitations on ramp rate versus what torque you're commanding, but generally those are set much higher than you would set if the system was expected to always act as a locked differential.

Electric motors can be controlled to react to unexpected changes in speed nearly instantly if you are instead have an inverter design that can be commanded specific speeds instead of torque. Every individual pulse sent to the motor can be varied so that you are maintaining the specific speed instead of a specific torque. Typically these types of controls are used in things like industrial machining servo motors, but there's no reason they couldn't do the same thing in an automotive space.

To simulate true 4 locked wheels with very low gear ratios, it would indeed need to ignore reacting to tire slip, and simply continue to turn at the demanded RPM with very little reaction to tire slip. If we strip down the behavior of a 4-wheel locked mechanical system, it has the following characteristics:

1) On a smooth level surface, all 4 tires will slip when the vehicle is turned. But at moderate throttle this tire slip is slight and the tread blocks are constantly reengaging after slipping, anchoring the vehicle. For an electric motor system to work like a true 4 wheel locker, it has to ignore this slip and not cut power or react at all to this type of slip. It needs to simply continue to maintain the same RPM, allowing the tire tread blocks to do their job of constantly reengaging traction then slipping.

2) On uneven surfaces, if a tire is unweighted it will continue to turn at the same RPM on a locked ICE vehicle as the tires that have weight on them. Here again we don't want the motor to react to lack of traction, but we also don't want it to spin wildly either. We want it to continue to turn at the same forward RPM so that the moment it gets some weight transfer it is already under power and engaging forward. Locked jeep owners often leverage this phenomena by turning their front wheels back and forth while chirping tires, or by rocking the weight of the vehicle back and forth while slowly chirping tires.

It is truly a different way of thinking about wheel slip. Instead of focusing on cutting power to eliminate wheel slip, the system would need to embrace slow, controlled wheel slip prioritizing continued tire rotation at a slow controlled speed. The goal would be to keep slow rotation going at all times and let the tires do their job of continually regaining traction, while also making sure the tires don't spin wildly fast where they have no chance to regain traction.

There is also drivetrain wind-up on 4-wheel locked ICE vehicles, that acts as a bit of a pulse in power when tires slip and reengage. All the parts in the drivetrain have some amount of deflection, bend, gear lash, etc that constantly has a pulsing effect on the rest of the tires when 1 tire slips. Something would need to simulate that too. Something still needs to "tie" all the wheels together to simulate the interaction between wheels that slip and wheels that grip that come with locking all 4 wheels together.

But to be clear, I know nothing about how to program electric motors. But the traditional idea of cutting power to stop slip won't simulate how a fully locked ultra low geared vehicle operates. They operate by continuing to turn at a controlled rpm that gives the tires a chance to claw for traction, regardless of whether the tires slip.

NOTE: others are saying they don't want lockers, but going back to the OP's video, the problem of tire spin and lack of traction over obstacles would be entirely solved by lockers. I turn off my lockers with a rocker switch when they aren't needed or if it starts side-slipping too much. I wouldn't turn on my lockers on a snowy off-camber road.

 

[brancky3]

Wait for Jeep to do it then. I can 100% guarantee you that Jeep will be building very offroad capable EVs once they really get into the EV game. I mean, that's what the Jeep brand is built on in the first place.

Today, you can get a Jeep Wrangler or a Jeep Grand Cherokee that's a hybrid electric vehicle. And on the Grand Cherokee 4xe (their hybrid designation), it comes with the top 4wd system available in a Grand Cherokee, including their eLSD, which essentially can act like a truly locked rear axle.

Yep I'm waiting to see what Jeep does but I have faith they're going to create a really great EV off roader. I would never consider cross shopping a Rivian and a Jeep though, there's absolutely no way (source: I owned a Gladiator)

 

[kurtlikevonnegut]

To simulate true 4 locked wheels with very low gear ratios, it would indeed need to ignore reacting to tire slip, and simply continue to turn at the demanded RPM with very little reaction to tire slip. If we strip down the behavior of a 4-wheel locked mechanical system, it has the following characteristics:

It is truly a different way of thinking about wheel slip. Instead of focusing on cutting power to eliminate wheel slip, the system would need to embrace slow, controlled wheel slip prioritizing continued tire rotation at a slow controlled speed. The goal would be to keep slow rotation going at all times and let the tires do their job of continually regaining traction, while also making sure the tires don't spin wildly fast where they have no chance to regain traction.

Listening to Max Koff on the podcast with MT, they have effectively already done this with Soft Sand mode. As I understand his explanation of it, they effectively tuned the motors to ignore wheel slip and continue to turn (albeit at a very high RPM) to maintain forward progress. They could very easily just adjust the pedal mapping and max RPM for a simulated Low Range experience.

Yep I'm waiting to see what Jeep does but I have faith they're going to create a really great EV off roader. I would never consider cross shopping a Rivian and a Jeep though, there's absolutely no way (source: I owned a Gladiator)

I have lost all faith in Jeep's ability to innovate. IMO they have become purely reactionary and much more interested in the market segment of mall crawlers with 4x4 badges than the market segment of true off-road enthusiasts.

 

[White Shadow]

This is the same misunderstanding that you've been having throughout this thread. Notice that I never once mentioned locked diffs, and for that matter neither did R.I.P. in his initial description of the issue.

The behavior I am describing is much closer to how an open or limited slip differential functions, which is still an important shortcoming of the current Rivian quad motor system. These differentials do not "sense" anything, the power flows through them as a natural consequence of physics.

Okay, so two things:

1. I never said that you mentioned lock diffs. My statement was based upon the fact that the majority of this thread was comparing quad motors to fully locked diffs. That much should be obvious. And this is why I started my response with 'no, not with locked diffs'

2. Our buddy R.I.P. ABSOLUTELY mentioned locked diffs multiple times throughout the thread. In fact, here's a direct quote from HIS VERY FIRST SENTENCE IN THIS THREAD THAT HE CREATED: " Good review, illustrating very well why quad motors can't really compete with full locking diffs."

So really, I think the misunderstanding is yours.

As far as the behavior you're describing....I already said it's more like open diffs. And FWIW, Rivian obviously doesn't have ANY diffs because there's a motor at each wheels. The real question is the one I've been asking all along---why did they program the software to allow the vehicles to BEHAVE like an open diff vehicle in the first place. There's no reason why you should ever see a quad motor Rivian spint a left front and right rear tire while the other two tires aren't getting any power at all. To me, that's insane, but that's exactly what we saw in Kyle's video with his own R1T. This can easily be fixed via software because it is software that is controlling the quad motors. Why is that so hard to understand?

 

[kylealden]

Listening to Max Koff on the podcast with MT, they have effectively already done this with Soft Sand mode. As I understand his explanation of it, they effectively tuned the motors to ignore wheel slip and continue to turn (albeit at a very high RPM) to maintain forward progress. They could very easily just adjust the pedal mapping and max RPM for a simulated Low Range experience.

The challenge here is that as others have pointed out, with current hardware, the only real way to make sure all wheels actually turn at the same speed would be to use wheel speed sensors and aggressive application of brakes against the motors.

That's doable, but sounds really unpleasant and like it has a high potential for damage if misused. (To be fair - you could say the same about lockers).

The root problem here is that it's easy to send the exact same power to all four wheels. It's very difficult to ensure all four wheels turn at the same speed, and even more difficult to ensure that wheels with the least static friction receive the least power. Differentials (plus lockers, in the former case) get those powers basically for free.

In most driving, and even the majority of offroading, they're not really the most important scenarios and quad motors will win on balance. But if you specifically care about preventing slip on loose terrain/high-camber - yeah, current quad motor implementations and hardware are intrinsically disadvantaged in ways that are exacerbated by high weight. You could certainly design hardware ways to overcome this - I bet they'd start to look a lot like differentials.

I want to shout out to @R.I.P. for the patience on this thread. I think this is mostly 16 pages of people talking past each other but you've commendably stayed calm and on point throughout. It took me a minute to fully grasp the disadvantages you're articulating and frankly they mostly don't matter to my use cases (where I care a lot about being able to clear obstacles, crawl, etc. but minimizing slip isn't often my #1 top priority), but I'm glad I took the time to understand.

 

[White Shadow]

Yep I'm waiting to see what Jeep does but I have faith they're going to create a really great EV off roader. I would never consider cross shopping a Rivian and a Jeep though, there's absolutely no way (source: I owned a Gladiator)

Gladiator isn't really a vehicle that I'd compare to a Rivian, but a Grand Cherokee is a very good comparison to an R1S and I could see lots of people cross shopping them.

Sponsored