Quad motor vs locking diff

[R.I.P.]

Rivian could simulate it by changing from torque control to speed control, making all the tires spin at the same rate, but you're still at the mercy of the substrate you're driving on. I've watched plenty of videos of Jeeps with locking diffs where they are still spinning all of their tires trying to climb up something very steep or slippery.

Yep. I am sure you and I are not the only people to rack our brains on how to make this happen without a connecting axle. I think with _brutal_ torque+brake control, you could get close. Oscillate the brakes so fast that your "slippage" is fractional. One of the immediate problems I come up with is that you would have to be doing this on all the wheels all of the time when in this mode. This mix of massive torque and braking on all 4 wheels seems pretty hard on the equipment, but is probably doable.

Man, I just keep coming back to "give me dual motors & locking diffs." Simple, efficient, and native torque vectoring with no overhead. Hum. Half the motor weight too.

Sponsored

 

[R1Tom]

Yep. I am sure you and I are not the only people to rack our brains on how to make this happen without a connecting axle. I think with _brutal_ torque+brake control, you could get close. Oscillate the brakes so fast that your "slippage" is fractional. One of the immediate problems I come up with is that you would have to be doing this on all the wheels all of the time when in this mode. This mix of massive torque and braking on all 4 wheels seems pretty hard on the equipment, but is probably doable.

Man, I just keep coming back to "give me dual motors & locking diffs." Simple, efficient, and native torque vectoring with no overhead. Hum. Half the motor weight too.

For off road crawling, nothing can beat a physically locked center, front and rear differentials. All 4 at same speed, each delivering up to 100% or torque as surface dictates.

But that is only scenario that arrangement excels at. For everything else, quad motor is the stuff....

 

[Zoidz]

It is interesting to see in the comments that people believe this can be "fixed" with software updates. I guess it is just a factor of people not understanding the technology.

I've read through the entire thread and find it very interesting, and appreciate your thought experiment scenarios of the creek and hill traversals. Agree, at this time Rivian appears to be challenged.

Can this be completely fixed with only software? Specific to the Rivian R1T, we don't know. Perhaps not... Probably not... Almost certainly not. Could they get closer with the existing tech? Maybe. But it's likely not a use case they really care about.

Could a software locking diff be achieved with a engineered combination of four motors, encoders, sensors, drive electronics and software? Absolutely. It's done in industrial settings all the time - paper mills, sheet plastic mills, web printing presses, etc. The control technology has been around for 15 - 20 years. But it's not cost effective for Rivian for the limited number of use cases.

 

[dorsai]

The Rivian already has encoders and knows the motor positions at all times, that's just inherent to the motor controllers. You could switch the software from torque control (which has all the problems RIP is bringing up) to speed control, where forward, no, or reverse torque is applied on a per-wheel basis to achieve 4-way 'locking' within the torque availability range of any single motor. Speed control of this sort is achievable without having to get any slippage at the wheel, as it's driven by phasing of the motor.

If implemented like this, you could 'lock' the Rivian's wheels just like a locking diff and have that capability with the limit I mentioned above, i.e. you can't exceed any one motor's torque capacity on one wheel, where with a dual motor+locking you could apply all the torque from one larger motor to one wheel, and a traditional 4WD powertrain can (approximately) send all torque to just one wheel.

 

[White Shadow]

......four wheels locked into identical rotation speeds can never be approximated by a quad motor or software....

Of course it can. This should be obvious with a little common sense. I could build a quad motor EV in my garage that puts the same exact power to all four wheels at the same time. To suggest otherwise is just silly.

There's nothing magical about mechanically locking wheels on an axle with a locking diff or front and rear axles with a locked transfer case. Its entirely possible to replicate that same behavior with a motor controlling each wheel.

Prove me wrong.

 

[ssincl3]

Rivian could simulate it by changing from torque control to speed control, making all the tires spin at the same rate, but you're still at the mercy of the substrate you're driving on. I've watched plenty of videos of Jeeps with locking diffs where they are still spinning all of their tires trying to climb up something very steep or slippery.

So from what I understand, without watching the video, is that the problem is mainly about power and torque to each wheel when there is at least one wheel with traction. When a diff is locked power gets “doubled” to the wheel with traction or not lost to the loose wheel. With the quad motor you can only rely on the only power available to that wheel which is 835/4 motors=208hp. If they could be locked, theoretically you can have 416hp to the one wheel if they are locked together. That will be stronger to push the 7000lb truck over obstacles than a mere 208. Just how I think of it.

 

[R.I.P.]

Of course it can. This should be obvious with a little common sense. I could build a quad motor EV in my garage that puts the same exact power to all four wheels at the same time. To suggest otherwise is just silly.

There's nothing magical about mechanically locking wheels on an axle with a locking diff or front and rear axles with a locked transfer case. Its entirely possible to replicate that same behavior with a motor controlling each wheel.

Prove me wrong.

I can't tell if you're joking lol. I'm going to assume that you are, just based on the statement.

In the unlikely case that you're not, the address is 7777 Galice Rd, Merlin OR. It is a four-wheel drive demonstration park where you can enlighten the many regular participants with your groundbreaking theories.

 

[mroe]

Very informative post. Thank you for testing these hypotheses. I've had similar thoughts.

I'm still very happy with the truck for my use case. When off-road, I'm usually always on dry surfaces and very rarely require the use of a locker.

 

[White Shadow]

I can't tell if you're joking lol. I'm going to assume that you are, just based on the statement.

In the unlikely case that you're not, the address is 7777 Galice Rd, Merlin OR. It is a four-wheel drive demonstration park where you can enlighten the many regular participants with your groundbreaking theories.

Not joking at all. The only joke here is you. For whatever reason, you fancy yourself as some old retired offroad expert, but when cornered by facts, you resort to your "I have the keys to the offroad park, come prove yourself" banter, which doesn't apply to my statement.

So again, prove me wrong that its not possible to design a quad motor setup that keeps each motor at the same exact speed, regardless of the traction or slippage at any corner. Give me one good reason why that's impossible to accomplish. For an electrical engineer, that should be child's play. But it's not because you're wrong. You can't make incorrect statements in your initial comment and still expect to be taken seriously or have any credibility.

 

[astonius]

Even if you simulate a locker by forcing the motors to turn at the same speed wouldn’t there be feedback forces you’d get from the ground back into the wheels that would present differently with a mechanically-connected axle?

I admittedly know nothing about this stuff, just spitballing where differences in behavior could still be insurmountable.

 

[White Shadow]

So from what I understand, without watching the video, is that the problem is mainly about power and torque to each wheel when there is at least one wheel with traction. When a diff is locked power gets “doubled” to the wheel with traction or not lost to the loose wheel. With the quad motor you can only rely on the only power available to that wheel which is 835/4 motors=208hp. If they could be locked, theoretically you can have 416hp to the one wheel if they are locked together. That will be stronger to push the 7000lb truck over obstacles than a mere 208. Just how I think of it.

You aren't effectively doubling power with locked diffs. If you had a 400 hp 4x4 with a fully locked drivetrain, you're dividing that 400 hp between all four wheels (allowing for drivetrain losses, of course). You can never increase power to one wheel because it's "locked" to the other wheel on the same axle.

With a quad motor EV, obviously there's no mechanical connection between wheels. So each wheel has the advantage of having the full power/torque of the motor available at each wheel.

My Jeep is said to have the ability to send all of its power to just one wheel. And while there's some truth to that statement, its also important to understand that not all of the engine's maximum power will ever get to a single wheel. And quite honestly, I'd never want it to. Traction is so much more important than power when it comes to 99.9% of offroading scenarios. With low gear engaged, the torque multiplication is insane and power isn't ever an issue.

 

[White Shadow]

Even if you simulate a locker by forcing the motors to turn at the same speed wouldn’t there be feedback forces you’d get from the ground back into the wheels that would present differently with a mechanically-connected axle?

I admittedly know nothing about this stuff, just spitballing where differences in behavior could still be insurmountable.

A tradionally locked 4x4 binds up when when turning. That's why you should always be in a loose terrain/low traction sitaution when you're locked up. Otherwise, damage to the equipment can easily occur. That's because locked diffs don't let the diffs actually differentiate, which is their entire purpose in the first place.

If one were to replicate lockers on a quad motor set up, you'd still have the same forces apply in a situation where the tires are trying to turn at different speeds. So for example, driving in a circle on blacktop. This would be brutal for the tires. I'm guessing the motors wouldn't be too happy either since they'd basically be working against the grip and coefficient of traction at each wheel. But again, that's why locked drivetrains are only used in low traction situations (or when moving straight ahead only) to prevent damage and/or wear to the equipment.

 

[astonius]

A tradionally locked 4x4 binds up when when turning. That's why you should always be in a loose terrain/low traction sitaution when you're locked up. Otherwise, damage to the equipment can easily occur. That's because locked diffs don't let the diffs actually differentiate, which is their entire purpose in the first place.

If one were to replicate lockers on a quad motor set up, you'd still have the same forces apply in a situation where the tires are trying to turn at different speeds. So for example, driving in a circle on blacktop. This would be brutal for the tires. I'm guessing the motors wouldn't be too happy either since they'd basically be working against the grip and coefficient of traction at each wheel. But again, that's why locked drivetrains are only used in low traction situations (or when moving straight ahead only) to prevent damage and/or wear to the equipment.

I do understand that part. I’m just unsure if in an identical low traction scenario would a simulated locked diff with no mechanical connection behave differently compared to a physically linked axle due to the forces coming back from the surface into the wheel(s). In the former the return forces would stop at the motor and not reach the other wheel, whereas in the latter the forces would travel through the axle/diff over to the opposite wheel.

 

[SeaGeo]

I think this is a mostly silly argument honestly. Locking diffs may be better for a very specific set of circumstances that a Rivian driver may encounter. For many of the other use cases that the Rivian is capable of, the quad motors are a massive improvement over a solid axle.

Quad motors allow for optimum flexibilty depending on use case. Sure that makes the Rivian more of a Jack of All Trades, but that's also what makes it so unique.

For example... ripping around and drifting or generally just having fun on a dirt road.

I can see Rivian offering a virtual locker mode where all 4 wheels are kept at exactly the same speed regardless of reaction.

I think this is generally what they're already trying to do to some extent. Think about two rear wheels where one has a lower slip force. You ramp up the the energy applied to each motor until wheel A slips. Wheel B doesn't slip. If the R1 isn't moving then Wheel A has to stop then stop that wheel after the slippage occurs and temporarily remember the torque limit that it can apply with 0 RPM. You then incrementally increase energy to wheel B until it spins.

It's a nasty little problem, but the boundary condition isn't actually speed partially because of different slippage forces at different wheels forcing the motor to "catch" them once you exceed the static coefficient of friction.

With a quad motor EV, obviously there's no mechanical connection between wheels. So each wheel has the advantage of having the full power/torque of the motor available at each wheel.

My Jeep is said to have the ability to send all of its power to just one wheel. And while there's some truth to that statement, its also important to understand that not all of the engine's maximum power will ever get to a single wheel.

Well, but with your Jeep in this stance keeps whatever it's total geared down torque and can apply it to three wheels when one wheel lifts off. The R1 reduces it's torque by 25%, and with as heavy as it is we've seen that power can be the limiting factor.

 

[Zoidz]

The Rivian already has encoders and knows the motor positions at all times, that's just inherent to the motor controllers. You could switch the software from torque control (which has all the problems RIP is bringing up) to speed control, where forward, no, or reverse torque is applied on a per-wheel basis to achieve 4-way 'locking' within the torque availability range of any single motor. Speed control of this sort is achievable without having to get any slippage at the wheel, as it's driven by phasing of the motor.

If implemented like this, you could 'lock' the Rivian's wheels just like a locking diff and have that capability with the limit I mentioned above, i.e. you can't exceed any one motor's torque capacity on one wheel, where with a dual motor+locking you could apply all the torque from one larger motor to one wheel, and a traditional 4WD powertrain can (approximately) send all torque to just one wheel.

I agree that if slipping is acceptable, your solution will work. But your speed control implementation would not ensure all 4 wheels start turning at the same speed in all possible conditions, which was the OP's concern and point, and is what a locking diff does.

Specifically from a standstill, with a mechanical locking diff, all 4 wheels will receive the exact same amount of power. On 4 wheel electric motors, from a standstill, there is no reference as to how much power to apply to each wheel. For example, If one wheel is in the air, and three others are firmly on ground, starting from standstill with equal and even minimal power applied will result in the tire in the air free spinning (no load on it) and the others will not spin (too much load for the minimal power applied). That's not emulating a locking diff. This same scenario extends to any situation where 3 wheeels have better traction than one. Even with speed control, different traction loads on each wheel will require a different amount of power applied to maintain the same speed and not spin, but you don't know how much until it starts to slip. This is OP's arguement.

With a high resolution encoder providing multiple pulses per motor revolution to the VFD - arbitrarily let's say every 5 or 10 degrees, you could detect the spinning wheel in a few milliseconds and adjust the power accordingly. Due to the way ABS sensors work, I don't think you could detect it quickly enough to satisfy the OP.

Do you know for a fact that Rivian has high resolution encoders on each motor? VFDs for DC motors can use other techniques to detect and control motor speed by using a single shaft sensor and back EMF detection.

Sponsored