Partial throttle and motor efficiency

[John in CR]

Yes I got a new 18fet Sinewave controller 100v 80amp controller 200.00 it's smooth and deadly quiet. Love seeing how that hubmonster quite down. Like magical.

Unnecessary, you can't hear anything from the motor while riding unlike the rattletrap inefficient hubbies you're used to that sound horrendous on startup. I can get 120V with 24 irfp 4568 mosfets with better programmability and switch on the fly current limits, and another switch to go between 2 throttle setups, in addition to common 3 speed switching, all for half the price of FOC stuff. The only thing missing compared to FOC is proportional regen and 150V nominal voltage or higher.

 

[John in CR]

Looks like you want an FOC controller, John. Or maybe a beast of a 60V trap controller and a 1-turn motor

Looks like you want a 6 phase motor, PunxOr.

 

[Martin A]

Yes I got a new 18fet Sinewave controller 100v 80amp controller 200.00 it's smooth and deadly quiet. Love seeing how that hubmonster quite down. Like magical.

Unnecessary, you can't hear anything from the motor while riding unlike the rattletrap inefficient hubbies you're used to that sound horrendous on startup. I can get 120V with 24 irfp 4568 mosfets with better programmability and switch on the fly current limits, and another switch to go between 2 throttle setups, in addition to common 3 speed switching, all for half the price of FOC stuff. The only thing missing compared to FOC is proportional regen and 150V nominal voltage or higher.

So was it just a bad non FOC controller in the video or are things different when the wheel is on the road? Is your controller different to the original one in the video?

 

[Alan B]

Most of the nonsense surrounding things like 3 speed switches goes away with a properly programmed torque controller. That stuff is bandaids for incorrect control programming.

But if you constrain the price to artificially low values you won't get good design and components until much later in the game. Best deal is to work with EV parts from the junkyard instead of hot-rodded scooter parts.

 

[John in CR]

Yes I got a new 18fet Sinewave controller 100v 80amp controller 200.00 it's smooth and deadly quiet. Love seeing how that hubmonster quite down. Like magical.

Unnecessary, you can't hear anything from the motor while riding unlike the rattletrap inefficient hubbies you're used to that sound horrendous on startup. I can get 120V with 24 irfp 4568 mosfets with better programmability and switch on the fly current limits, and another switch to go between 2 throttle setups, in addition to common 3 speed switching, all for half the price of FOC stuff. The only thing missing compared to FOC is proportional regen and 150V nominal voltage or higher.

So was it just a bad non FOC controller in the video or are things different when the wheel is on the road? Is your controller different to the original one in the video?

What video? There's a bit of zing spinning one up no-load that's inaudible aboard the bike. In some of my videos, those with the cam attached to the bike, there's been some kind of noise coming through the connection to the frame that I've been unable to find the source. Maybe its some kind of harmonic amplified by the camera, but other than wind noise that loudest thing is the tires. Open for ventilation I can barely hear a bit of motor zing if I listen for it, but only when I'm riding near a concrete divider on the highway.

There is one video, the 103mph video, from when I had the halls each going to the wrong controller, and the motor did make a slight noise on startup from a dead stop due to the advanced timing of 3°. I rode it 2 weeks like that, because I was unsure if I was just being overly sensitive. During the same overhaul I went from 20s to 30s along with new controllers, so it really took a combination of a slight noise on starts, a bit less mid-speed acceleration, higher than expected no-load speed, along with more rapid motor heating, to convince myself to try the halls the other way. It's not a matter of just swapping plugs.

FYI, I've only used trap controllers. I have lots of motors, so I'm well aware of the rattletrap noise of common 51 slot 46 magnet motors. The only motors that come close sonically are the old style X4xx and X5xx Xlyte motors, though their quite thin and flat covers resonate quite a bit more.

When I'm riding, as long as I don't hear more than just tire noise then I'm happy. I'll have some professionally done vids soon, and unlike others we won't use music to cover up a noisy bike, but we'll also make sure only the actual noise come thru,

 

[Alan B]

Test ride a Zero if you can, see what a good control system should feel like.

 

[John in CR]

Most of the nonsense surrounding things like 3 speed switches goes away with a properly programmed torque controller. That stuff is bandaids for incorrect control programming.

But if you constrain the price to artificially low values you won't get good design and components until much later in the game. Best deal is to work with EV parts from the junkyard instead of hot-rodded scooter parts.

Applying craptastic Infineon/XieChang controller experience to that of even less costly modern trap controllers is foolish. You can push boutique priced stuff all you want, and they'll all be out of business in the long run, because they incorrectly chose short-term profit over growth.

I'll have a 3 speed switch on my new bike, but that will be only to limit speeds for people I let ride my bike. The switch for the 2 different settings of throttle response will be useful for me, as well as the switchable current limits. I rode a racing moto once and darn near fell off, so the bandaid comment is nonsense. There's no way you can tune your precious FOC controllers so with no changes they're tuned for racing type response, and still be able to let someone with little experience throw their leg over the saddle. I gave up the ridiculous throttle response we suffered with for so long several years ago, and while I believe current control is best, the difference now isn't worth more than double the price, not to mention the lack of higher voltage options.

If you're bringing up "bandaids", don't forget to include flux weakening.

 

[Alan B]

Zeroes aren't racing bikes, they're tuned for extreme rideability.

No comparison to trap controllers.

 

[John in CR]

Zeroes aren't racing bikes, they're tuned for extreme rideability.

No comparison to trap controllers.

You mean no comparison to trap controllers that you've experienced. Zero uses a controller that would be impossible for me to program anyway. I don't doubt FOC controllers are better, but the current crop doesn't meet my needs and is over-priced anyway. It's a moot point, because appropriately priced FOC controllers will come soon enough, and controllers have advanced faster than batteries. It's baffling how so many people are still buying the same crap controllers I gave up on over 5 years ago.

 

[Alan B]

I tried to find the mentioned discussion on flux weakening, everything I found indicated complete cluelessness on how it works, and there is a lot of misunderstanding about flux weakening efficiency. Justin's tests don't show it to be as inefficient as people thought, and some of the current production electric cars use it as a part of their design, so apparently it can be efficient enough. Running excess voltage is also inefficient, so it is a tradeoff like so many other things. Trap is also inefficient, so using a high efficiency motor with lossy trap waveforms is also a waste (possibly 5% or more), perhaps more than the improvement from a good motor. So buying an extra efficient motor and then running trap on it doesn't make a lot of sense.

Luke is probably right, in a few years we will look back and laugh at much of what we are doing now that we think is so great. But if you want cheap good FOC motor control you'll have to wait, it isn't going to be cheap for years. The cheap trap gear is seeing to that by damaging the market value of better gear.

Those of us willing to pay a little more will get to enjoy it, and to encourage the vendors to keep working on it. Hopefully they won't give up, as many others already have because the potential for making a profit on a good controller is so poor when competing on grossly underpriced products. I like a cheap box as much as anyone, but it is hurting the development of improved technology.

 

[Martin A]

Hi John (and everyone else)

The video I referred to in my post higher up the page above was of Bluefangs bike with the two Sabvoton contollers on page 1 of this discussion. I was not referring to noise in any of your videos. I admire what you do and have followed your exploits and experiments with more interest than any others on this board, as I am looking for power in the range of 30kw+ using a couple of motors. Not for a bike I hasten to add.

The thing is that I am looking to stay under the 75V limit for electrics here in the UK. Beyond that limit the law changes and that makes things more expensive, and as Luke among others has pointed out, more dangerous too.

From what I can derive from this thread, an FOC controller would help solve the problem that you mentioned at the beginning about better efficiency at partial throttle, but you don't think the benefit is really worth the cost. I'm also on a budget so I appreciate your view on this. I also appreciate that if we are to progress then people need to invest in new tech where sensible. If however FOC controllers were the same price as your controllers, would you choose to fit them (given you could only go to 75V) or are there other benefits that I'm not understanding because I know virtually zero about controllers.

For instance I understand it is possible to go way above a motor's rated power for short bursts of a few seconds or so, which is what I'm also looking to be able to do (I want to do the odd drag race). Is this possible with either, both or neither of these controllers or would a completely different one be necessary? For instance what could the hubmonster put out using these controllers and the appropriate battery pack?

 

[John in CR]

Trap is also inefficient, so using a high efficiency motor with lossy trap waveforms is also a waste (possibly 5% or more), perhaps more than the improvement from a good motor. So buying an extra efficient motor and then running trap on it doesn't make a lot of sense.

Not a chance the real world difference is 5% or more, because at the power I run my controllers would quickly BBQ themselves. I don't run high efficiency motors for the efficiency gain (ie more range). I run them for the things that make them more efficient. The far lower resistance that gives them lower copper losses means they can handle higher current per unit of torque. The better materials and design that gives them lower iron core losses makes them capable of much higher rpm. Higher current and higher rpm means drastically higher power potential.

If efficiency was my focus then I'd slow down as well as really address aero for easily double or triple the range. I'll be more than happy to make the switch to FOC when the price comes down and the max voltage goes up. I did pick up some cheapie sine wave controllers that work fine, but their current and voltage limitations are too low for my needs. Plus there's no way I'm going back to being unable to program my own controller settings.

 

[Alan B]

Efficiency loss will vary but has been measured. The trap controller fails to move the magnetic field smoothly in the motor and fails to keep the torque generating fields at 90 degrees, this promotes losses in the system, not just the controller. Predominantly these would manifest in the motor. The manufacturer's claimed efficiency may depend on smooth control to achieve the ratings.

It is similar to phase angle power factor loss in the AC distribution grid. Operating at incorrect phase angles cause losses even though the phase angle error itself isn't lossy. It requires more current to make the torque than it should because the angle is wrong. This extra current makes loss everywhere, and it's an i squared loss.

Trap controller technology is a good fit for low cost, low power and low speed applications. It has no place in higher power. It is a coarse approximation to what is needed. Like using a fixed distributor with no spark advance.

FOC isn't the only way to get good smooth control and maintain proper field angles, a direct rotary position encoder is in some ways even better. Trap is just too low an encoder resolution to do the job well. Six steps per e revolution is very coarse and jerky. Smooth control is better.

 

[John in CR]

Thanks for the kind words Martin. I am just a frugal ebike hack. The motors do have a bit of zing with trap controllers, and when ventilated they're probably 3 times as loud, but you honestly have to strain to ever hear the motor while riding and there's virtually none of that noise on launch like most 3 phase hubbies that sound like something is broken. If I was closed up in a noise amplifying velo, then Im sure I'd want sine wave controllers, since I prize silence.

It's better to look at current and voltage instead of power. eg I ran my first HubMonster in stock form at 16kw peak input, and I now run that motor at 27kw, but I'm only using slightly higher current from each controller. I went from 220A to 245A combined. The lions share of the power increase comes from going to a 111V nominal battery pack from 74V, so going to extreme power on tap cost little in terms of heat, and since I now run ventilated my motor runs far cooler than it did at 16kw. I never got into risky high temps and never had to back off the throttle before, but now I have no heat concerns at all. Plus lower temps mean higher efficiency too.

Heat is our limitation, and with hubbies the overwhelming majority of the heat comes from the copper, except at high speed cruise where the iron losses become more prominent. Copper losses go up with the square of current, so you really want to set current limits as low as possible but still get the performance you want. Hillzofvalp managed to pump 30kw into his hubmonster with voltage sagging to 74V, but he ended up having heat issues running over 400A. It's not just current limits that you have to consider either. Load plays a big factor, whether it's accelerating the rider and bike or climbing hills, or gearing higher with a larger diameter wheel. The bigger the load the greater the time spent at high current.

WRT to what I would do, my issue isn't speed, and 74V nominal is fast enough. Like I said early in the thread I'm not giving up the acceleration passing through 50mph or so. That's why I need the higher voltage with the trap controllers. If FOC fixes that issue with field weakening, so I can still have extreme acceleration in the 50-60mph range using a 74V pack, then I don't need high voltage. That's running in-wheel though. With HubMonster in a mid-drive, then I want to run as high a voltage as possible to take advantage of the untapped potential of higher rpm. The I can gear it down for a lower effective load and run even lower current while still maintaining extreme power.

I'm unclear exactly what you're after. By "a couple of motors", do you mean 15kw to each? If so then yes that's a simple task. We would have to talk about the total load you plan and type of terrain.

John

Hi John (and everyone else)

The video I referred to in my post higher up the page above was of Bluefangs bike with the two Sabvoton contollers on page 1 of this discussion. I was not referring to noise in any of your videos. I admire what you do and have followed your exploits and experiments with more interest than any others on this board, as I am looking for power in the range of 30kw+ using a couple of motors. Not for a bike I hasten to add.

The thing is that I am looking to stay under the 75V limit for electrics here in the UK. Beyond that limit the law changes and that makes things more expensive, and as Luke among others has pointed out, more dangerous too.

From what I can derive from this thread, an FOC controller would help solve the problem that you mentioned at the beginning about better efficiency at partial throttle, but you don't think the benefit is really worth the cost. I'm also on a budget so I appreciate your view on this. I also appreciate that if we are to progress then people need to invest in new tech where sensible. If however FOC controllers were the same price as your controllers, would you choose to fit them (given you could only go to 75V) or are there other benefits that I'm not understanding because I know virtually zero about controllers.

For instance I understand it is possible to go way above a motor's rated power for short bursts of a few seconds or so, which is what I'm also looking to be able to do (I want to do the odd drag race). Is this possible with either, both or neither of these controllers or would a completely different one be necessary? For instance what could the hubmonster put out using these controllers and the appropriate battery pack?

 

[John in CR]

Efficiency loss will vary but has been measured. The trap controller fails to move the magnetic field smoothly in the motor and fails to keep the torque generating fields at 90 degrees, this promotes losses in the system, not just the controller. Predominantly these would manifest in the motor. The manufacturer's claimed efficiency may depend on smooth control to achieve the ratings.

It is similar to phase angle power factor loss in the AC distribution grid. Operating at incorrect phase angles cause losses even though the phase angle error itself isn't lossy. It requires more current to make the torque than it should because the angle is wrong. This extra current makes loss everywhere, and it's an i squared loss.

Trap controller technology is a good fit for low cost, low power and low speed applications. It has no place in higher power. It is a coarse approximation to what is needed. Like using a fixed distributor with no spark advance.

FOC isn't the only way to get good smooth control and maintain proper field angles, a direct rotary position encoder is in some ways even better. Trap is just too low an encoder resolution to do the job well. Six steps per e revolution is very coarse and jerky. Smooth control is better.

Russia and China are going through economic downturns, and the Ruble's exchange rate has fallen drastically, yet FOC controller prices remain stubbornly high. I recently got some 120V trap controllers with great features that even the FOC's don't have which are rated for 200A continuous for well under $250/ea. I can buy a nice Russian beer here in Costa Rica for less than domestic beer. FOC controller prices are artificially high, so just like I'd never buy an iPhone for $500-600, I won't pay that much for a controller even though I'm sure they cost more to make than a dummy phone.

I don't doubt they're better, but they aren't that much better. If trap controllers are so terrible for higher then someone
needs to step up and prove it with performance. Steveo is nearing the end of his build with HubMonster and a pair of Adappto's, so we'll see. I've dropped over 60lbs so the days of my bikes having an 80-100lb load handicap are over, so it's going to take a big step up for someone to surpass what my new bike is about to do.

FWIW, the factory used it's stock controllers for all dyno runs, so that's probably why Miles' spreadsheet is spitting out a slightly higher peak efficiency than the factory claims.

 

[fechter]

The main reason the cheapo 6 step controllers heat badly at partial throttle is they don't use synchronous rectification. During the off part of the PWM cycle when the current is circulating through the body diodes in the FETs is where you get the real heating. The heating will be a function of motor current for the most part, but from experience I know its worst at low speeds/high current. While the controller is wasting some heat, the total loss as a percentage is not all that bad. All the FOC controllers use synchronous rectification, so will run more efficiently. It would be possible to make a cheapo 6 step with synchronous rectification and it would heat a lot less but still be rough and noisy. Maybe not a bad compromise if cost is important.

 

[riba2233]

Trap kelly's use synchronous rectification.

 

[Alan B]

The main reason the cheapo 6 step controllers heat badly at partial throttle is they don't use synchronous rectification. During the off part of the PWM cycle when the current is circulating through the body diodes in the FETs is where you get the real heating. The heating will be a function of motor current for the most part, but from experience I know its worst at low speeds/high current. While the controller is wasting some heat, the total loss as a percentage is not all that bad. All the FOC controllers use synchronous rectification, so will run more efficiently. It would be possible to make a cheapo 6 step with synchronous rectification and it would heat a lot less but still be rough and noisy. Maybe not a bad compromise if cost is important.

Yes.

This is a part of the story, the other part is they don't drive the FETs very well, they spend too long in the very lossy transition region, in which loss is a function of voltage. The higher the voltage the higher the loss.

It makes sense that the loss is worst at low speed due to the high current since part of the loss is a function of the current, and the voltage transitions are maximal for the switching loss part.

Synchronous operation and better drive will both contribute to lower losses. A well made trap controller can lower losses slightly, but still the incorrect phase angle for much of the cycle reduces system efficiency, requiring more current for the same torque.

 

[Martin A]

What is it that makes FOC controllers cost twice the price, the componentry or something else?

 

[Alan B]

What is it that makes FOC controllers cost twice the price, the componentry or something else?

That's a good question.

Comparing a cheap cloned commodity knock-off of a BLDC trap controller to any first line product is fraught with problems.

And re John's comment about Russian electronics - we have that, see Adaptto. But should the market be depending on starving companies to define the value of a product?

Both controller types have (generally three) phase FET H bridges driving the motor windings. Trap controllers have a very simple CPU or chip that is reading the three binary hall inputs and driving PWM into the FET bridge, while monitoring battery current via a shunt. Literally a one dollar chip can do the brains of a trap controller.

Let's make a list of some of the things that an FOC controller has that a Trap controller doesn't.

FOC controllers have motor current sensors, at least two of them. They must have high bandwidth to be used for FOC control. They feed ADCs that need to be fast and fairly precise and synchronized to produce the data the algorithm needs for good control. These are more expensive than the shunt a trap controller uses.

FOC controllers have better drivers for their FET arrays. They are managing currents into and out of all three phase wires simultaneously, they must actively control all the FETs synchronously and not rely on the diodes to conduct motor current, so they need better control precision.

FOC controllers need to manage noise much better since they are making critical measurements while they are controlling. This requires better design and more components.

FOC controllers have significantly more computing power and complicated control algorithms to implement in real time.

Many licenses and patents exist for FOC controls, this may have impact on the cost of the software.

The FOC controller needs to know more about the motor and operational characteristics, so documentation, tuning software and support costs all go up.

The software needs to be more suited to the application, so an FOC program for running a lathe won't be suitable for an ebike. So the controller must be more customizable than a Trap controller. Again documentation and support costs go up.

I'm sure there are many more details that I haven't covered here. Basically an FOC controller is implementing a modern optimized mathematical control algorithm whereas a trap controller is basically implementing a 200 year old mechanical commutator (invented in the early 1800's) using electronics.

 

[John in CR]

What's wrong with the cheapie sine wave controllers that aren't FOC? They silence motors essentially the same, so the wave form must be matching the motor requirements better. Then it's just a matter of better throttle control and proportional regen, though a nice added touch would be at low speed switching over to plug braking to get smooth electric braking down to a stop. Other than proportional regen, which is ridiculous that it's not the norm already, I've had all these things in cheap controllers, so the expensive current sensors and control chips aren't necessary to achieve the vast majority of the benefit.

Alan you can badmouth trap controllers all you want, but it's just hot air, since even with an 80-100lb load handicap, the trap controller driven hubbie still outperforms any FOC driven hubmotored ebike. For the current price of FOC control I could do it with a motorcycle instead, making it passenger capable and extending range well beyond 100 miles. It's like eating with sterling silver utensils when stainless steel works just fine. Maybe the problem is that your only experience with trap controllers is the plastic spork type stuff.

While the electronics types want to focus on a niche market where high prices are ok, that does little to help the world convert to electric transportation. I for one don't want to leave it in the hands of the big car manufacturers, so that requires mass market appeal and with ebikes price is critical as demonstrated by the Sondors Indiegogo campaign.

 

[FluxZoom]

Trapezoidal driven controllers measure rotor position using hall sensors. My take is that hall sensors aren't amazingly accurate, but even if the sensor was still amazingly accurately, there is still some form of a delay in the reading and commutation that requires specific algorithms to compensate for this delay. This results is a never perfect degree of accuracy, this is largely why we see a higher total system efficiency when using a FOC controller. A controller which doesn't use FOC but still generates a sinusoidal wave rather than a trapezoidal wave is nice in that it produces less noise, but the degree of efficiency isn't better as you might find in a FOC controller, and I suspect it may be worse than trapazoidal waveforms in this way as I imagine there is greater switching losses and similar inaccuracy that you find in a purely sensor based controller.

A typical trapezoidal controller isn't really great at keeping up with the ever changing ideas like phase resistance and RPM in a motor but a FOC controller will do a seemingly perfect job of this at highish motor RPMs and above. This has a lot to do with how a FOC controller operates, kind of a long subject and a brutal amount to take in. The main point is that a FOC controller isn't some gimmick and I imagine that in the future the majority of brushless motors will be driven using FOC at some point at a price that doesn't seem so annoyingly unattainable to some.

https://endless-sphere.com/forums/viewtopic.php?f=30&t=79225&p=1169431 If you want to know more, there is a video you can find here that will go into some depth on the subject of field oriented control and BEMF. https://www.youtube.com/watch?v=cdiZUszYLiA Direct link if you don't want to hunt.

 

[Martin A]

What is it that makes FOC controllers cost twice the price, the componentry or something else?

That's a good question.

Comparing a cheap cloned commodity knock-off of a BLDC trap controller to any first line product is fraught with problems.

And re John's comment about Russian electronics - we have that, see Adaptto. But should the market be depending on starving companies to define the value of a product?

Both controller types have (generally three) phase FET H bridges driving the motor windings. Trap controllers have a very simple CPU or chip that is reading the three binary hall inputs and driving PWM into the FET bridge, while monitoring battery current via a shunt. Literally a one dollar chip can do the brains of a trap controller.

Let's make a list of some of the things that an FOC controller has that a Trap controller doesn't.

FOC controllers have motor current sensors, at least two of them. They must have high bandwidth to be used for FOC control. They feed ADCs that need to be fast and fairly precise and synchronized to produce the data the algorithm needs for good control. These are more expensive than the shunt a trap controller uses.

FOC controllers have better drivers for their FET arrays. They are managing currents into and out of all three phase wires simultaneously, they must actively control all the FETs synchronously and not rely on the diodes to conduct motor current, so they need better control precision.

FOC controllers need to manage noise much better since they are making critical measurements while they are controlling. This requires better design and more components.

FOC controllers have significantly more computing power and complicated control algorithms to implement in real time.

Many licenses and patents exist for FOC controls, this may have impact on the cost of the software.

The FOC controller needs to know more about the motor and operational characteristics, so documentation, tuning software and support costs all go up.

The software needs to be more suited to the application, so an FOC program for running a lathe won't be suitable for an ebike. So the controller must be more customizable than a Trap controller. Again documentation and support costs go up.

I'm sure there are many more details that I haven't covered here. Basically an FOC controller is implementing a modern optimized mathematical control algorithm whereas a trap controller is basically implementing a 200 year old mechanical commutator (invented in the early 1800's) using electronics.

Thanks for that Alan. I can see your side of the discussion as well as John's. Most went over my head but I'll at least be able to know what to research to improve my understanding. One of my main concerns (and that of most hobbyists I imagine) is reliability and fixability / replaceability. It seems that you both agree that a trap is cheaper to replace and can work to high voltages for those wanting a super powerful e-bike, while a FOC is more efficient but the available controllers have real world voltage limits.

However does a FOC (or Sine wave) have greater reliability that might tip the balance more in its favour from that point of view? To me there seems to be more to go wrong with a FOC, just like with a modern car over a fifty year old one. The latter I can usually fix for the price of a few beers, the former I'm usually looking at havaing to give someone else several crates of fine wine! So which one is more reliable? And which is more fixable? I mean what does go wrong with each type and can they be repaired or are they both very reliable if used within sensible limits?

 

[parabellum]

To Alan B. Is there any real FOC (controller) on the market anyway, for Johns big hub motor, not considering those hall sensor start-up pussies?
Than, speak numbers, there will be 25% more component cost compared to trap and rest 75+% "intellectual property" behind. (My poor experience)
I also love to overestimate my engineering value, it just ends when someone asks for concrete numbers and work hours.

 

[Alan B]

A sinewave controller is making a smoothly rotating magnetic field in the stator of the motor. A trap controller is making a magnetic field that jerks 60 degrees every now and then when it decides the rotor has moved into the next sector. Most of the time the fields are not perpendicular for maximum torque efficiency with trap.

In order for a sinewave controller to make the magnetic field at right angles to the rotor for maximum torque efficiency, it needs to know where the rotor is. FOC is one way to do that. Digital Halls such as used by trap controllers are another way to do that, but the accuracy isn't good enough. Time estimates interpolating from digital halls can be used. Resolvers are common on higher end systems, or optical encoders. Analog halls have been used by some motors. There are many ways to do it, however they often require making changes to the motor so are best done by manufacturers making a system of both motor and controller. Random combinations of motors and controllers don't necessarily mate up well.

As I said earlier, even if FOC is not used, the sinewave controller will be an improvement if it has a reasonable estimate of rotor position. It needs to be significantly better than 60 degrees.

The other side of the coin is excellent FET Power Stage design. Trap controllers generally have low quality value engineered Power Stage designs. They don't have to, but the cost constraints of the market have driven them to a very low standard. A trap controller can be made with very good Power Stage design, but who does that? Such a controller would cost almost as much as a Sinewave controller. So we don't generally see those, they exist in products that are too expensive for folks like us to look at. They were more common in earlier ebikes when ebikes were quite expensive. Cheap ebike motor controllers have driven them out of this market segment.

What would a quality Power Stage do for us? It would reduce the losses and provide features that would make the controllers much more reliable. A good Power Stage will not fail just about no matter what. It will shut down and blink an error code in the face of shorted wiring or motors, miswired motors, etc.

If you want reliability the quality Sinewave controllers are built to much higher standards and are extremely reliable. They have considered the failure modes and designed in protection for them, if at all practical. Since they are measuring motor current directly they can respond to shorts in the motor wiring that would blow a trap controller up, because by the time it senses the current at the shunt the capacitor bank energy has already destroyed the FETs.

People ask about costs. I was looking into high performance integrated circuits designed to drive FETs in controllers awhile back, and the good ones have a ton of features built into them to monitor the FETs for safety and protection and will shut the FET down very quickly before it can be damaged, these are very impressive parts. Each chip costs five bucks in quantity (almost ten dollars each in low quantity) and it takes six of them to make a controller. That's $30 just for that one part. You will never see that part in a $100 controller.

Putting constraints on the cost of the electronics based on the cost of the motor is not a sound strategy. The costs of the two items are unrelated and uncorrelated in so many ways. Electronics costs will go down over time while mechanical fabrication costs with rare earth materials will tend to go up. In the long term good electronics will likely be as cheap as we all would like, but I'm not going to wait to enjoy better efficiency and reliability for an artificial and arbitrary cost constraint. Life is too short.

I buy and use trap controllers, but I also buy and use sinewave controllers and the difference is significant. Don't avoid a major improvement if you can afford it. If you can't, figure out a way to make a few bucks to get a good one.

I'm not familiar with all the products out there, the market is changing fast and we are seeing a lot of new Sinewave controllers coming out in the ebike market. In the motorcycle and large scooter market that John's looking in there are probably new products as well, but I'm not up to date on everything in that area. The Sabvoton has been excellent, they were coming out with a higher voltage model at one point, but I haven't followed it lately. A good Power Stage design doesn't need as many FETs either, so you can't really compare that way.

It is really best to stay under 100 volts by a reasonable safety margin, the efficiency of higher voltages is much harder and - more expensive.