Thursday, April 19, 2012

Sensorless Start-Up / Speed Control

It's spring demo season again. I'll be at the Cambridge Mini Maker Faire on Friday, along with the MITERS crew, with an assortment of projects. Come check it out. Follow-up post to come.

For now, time for a massive multi-controller update. The Sensorless, Sine-Commutated Field-Oriented Current Control (SSCFOCC) algorithm I've been working on has been sufficient abstracted now that I can develop for it on two different motor controllers (3ph v3.1 and FFv1.0) at the same time. (3ph v4.0 is coming soon and DirectDrive v1.0 still exists, I promise...)

Before I start developing new hardware and potentially writing a entirely different sensorless control algorithm, though, there were some loose ends I wanted to clean up in the current algorithm. Once these points are addressed, I will go back and write up a summary of my first-generation sensorless code since by now it's too complex for individual blog posts.

But First...

More Position Filtering

Loose End #1 was a problem with the rotor position estimating filter used to clean up the rotor angle at high speeds. I added this at the end of the last sensorless post and even though it immediately improved the high-speed performance of FFv1.0 on the hexrotor motors, there was a weird side-effect on the flux vs. angle graph. The flux estimate would begin to lead the rotor angle estimate at high speeds, which should not happen since the rotor angle estimate is derived from the flux estimate. The flux vs. angle curve should be constant. What the?

After staring at the code for a little while, I identified this a rather complex software error on my part tracing all the way back to a subtlety of BWD's sensored FOC code... I'll spare you the details, but I fixed it and now the flux estimate is much nicer:

Pneu Scooter w/ 3ph v3.1. Compare to unfiltered flux estimate.
Much of the "noise" from the 2d plot is shown to be offset at low speed (<200rpm).
Hexrotor motor with FFv1.0. Compare to the unfiltered flux estimate.
And in nauseating pseudo-3d.
The low-speed offset is a known issue having to do with the fact that the integrator at the heart of the flux estimate is implemented as a low-pass filter. There is also a phase offset in the flux estimate resulting from this practical non-ideality:

Since this is a known offset, it could potentially be subtracted off. Or the low-pass filter could be designed such that at the start-up routine exit speed, the offset is less than 15º or so. For the data above, that would be about 150rpm. Speaking of start-up routines...


Loose End #2 was the lack of any start-up routine for either controller. I've been putting it off for a number of reasons. For one, Pneu Scooter really doesn't need a start-up routine. The sensorless algorithm has no trouble picking up the rotor speed at as low as 50rpm and that's well within single-kick speeds:

One kick gets to 75rpm, at which point the flux estimator has locked in.
Also, Pneu Scooter really doesn't have enough torque to justify a no-kick start-up routine. Even at peak current (40A), the acceleration is something like 150rpm/s. The sensorless start-up will achieve far less than peak torque, though, since it won't have the benefit of optimum current vector placement. So it could mean a couple seconds of balancing on a barely-moving scooter while the speed ramps up. It's just much more natural to kick start.

FFv1.0 and the hexrotor motors, on the other hand, almost definitely need a start-up routine. Even though in this video, a very large (14x4.7SF) prop starts cleanly and quickly with no start-up routine at all. The command PWM goes high enough that the motor is forced to do something, and by luck it gets kicked in the correct direction at sufficient speed for the flux estimator to lock in. But this is unreliable (works maybe 50% of the time) and inelegant, so it should really have a slow-ramping startup routine as well.

Maybe the real reason I haven't implemented a start-up routine yet is because it's easy. At least, the most obvious way of doing it is easy. Lacking position information, there isn't much choice in how to drive a brushless motor: it becomes a coarse stepper. Except with sinusoidal commutation it can be like a microstepping coarse stepper, which is still pretty smooth. The point is, the only way to drive it with no knowledge of the rotor position is to put current in somewhere and wait long enough to know that the rotor has most likely reached the position you told it to go to. With that in mind, I wrote a simple start-up routine with four states:
  1. IDLE: Wait at zero speed until the user commands a positive torque or speed.

  2. PARK: Place current vector at a fixed angle and wait for some time for the rotor to reach a stable position.

  3. RAMP: Rotate the current vector with a constant angular acceleration that is sure to be achievable by the rotor given the amount of current commanded and the estimated load inertia.

  4. RUN: After a certain speed, switch to sensorless field-oriented control using the rotor angle predicted by the flux observer.
The ramp state here differs a little from BLDC-style sensorless start-up routines that use six-step voltage-mode commutation. Since this ramp state is still current controlled, the possibility for variable user-commanded acceleration during "open-loop" start-up is present. The angular acceleration of the ramp is linked to the amount of current commanded. If you go light on the throttle, the ramp takes longer.

Since the load inertia can't be exactly known, the start-up ramp acceleration has to be conservatively chosen. The start-up torque will therefore be significantly less than the run torque, and the efficiency will be very low. This is a common trait of sensorless start-up algorithms and the best way to minimize the effect is probably to get out of start-up as soon as possible. Transitioning to closed-loop sensorless control at 5-10% of the maximum speed seems to be achievable.

Anyway, enough hypothesizing. Here is my first shot at a sensorless start-up for Pneu Scooter, the harder of the two challenges:

The ramp goes from 0 to 150rpm in about 3 seconds with 30A commanded current. This is about 50% of the acceleration that would be expected from closed-loop control, so it feels sluggish for the first three seconds. But, it is smooth and transitions cleanly to run-state closed-loop control at 150rpm. Making the ramp faster results in pole slipping, so this is about as good as it will get.

One option for improving the start-up is to get out of the ramp state earlier. The flux estimator is able to pick up the rotor movement at as low as about 25-50rpm, so waiting around until 150rpm sounds silly. But, the low-pass filter-induced phase lead is still very high at those speeds, so the transition doesn't work reliably. It's clear that the low-pass filter time constant and the ramp exit speed are coupled in this way. Something to think about for future optimization.

Another option is to use more current. This doesn't help the efficiency, but it will create a faster ramp. 60A for a short period of time shouldn't hurt the motor or controller, but I'll wait for 3ph 4.0 to test that.

Start-up for the hexrotor motors is a much simpler problem. For one, they have essentially no load at start-up when the prop is moving too slowly to create significant thrust or drag. The performance of the controller is also not at all dependent on low-speed torque characteristics since the minimum RPM seen in flight will be well above the run-state threshold. Here's a very light ramp (<5A) for the 14x4.7SF prop:

The threshold for transitioning from ramp state to run state is 600rpm, and once it transitions to run state the torque increases dramatically. The ramp can afford to be much faster without slipping poles, but for the purpose of reaching a safe idle speed for the prop, it almost doesn't matter. Note that the speed goes directly to idle speed, 1,000rpm, and sits there. That brings me to...

Speed Control

Loose End #3 is speed control for FFv1.0. Pneu Scooter, as a vehicle, is happy with current (torque) control. But for the hexrotor, the master controller may want to command an exact speed for each rotor. Speed is directly related to static thrust (though not linearly). Steady-state current is also related to static thrust (close to linearly). But with inertial transients to deal with, the performance of a speed-based controller might be better.

Actually, the simple way to do "speed" control is to directly vary the PWM duty cycle, bypassing the current control entirely. This controls speed because motor speed is approximately proportional to applied voltage, which is controlled by the PWM duty cycle. As it turns out, it's definitely possible with the modified Synchronous Current Regulator (mSCR) to use this control mode, bypassing the q-axis current controller but leaving the d-axis (phase) controller working. This gives the benefits of FOC (dynamically adjusted motor timing) but the simplicity of direct PWM control. In fact, this was what I used for the first few tests of FFv1.0. Here's some data showing it executing step "speed" commands:

At each step, the q-axis and d-axis current spike due to the inertia of the motor and prop. But, the back EMF brings the q-axis current back to a steady-state value as the prop speeds up or slows down, and the mSCR adds or subtracts from the phase to bring the d-axis current back to zero.

The benefits of direct PWM magnitude control are simplicity and robustness. With current limiting to prevent the spikes from damaging components, I think this could be a pretty good way to run a multirotor controller. It's how most ESCs operate, so it would feel normal. But, it has several downsides:

  1. It can't achieve minimum rise time. The step in voltage magnitude will give only as much current as the motor resistance will allow. A smarter controller could slew the motor faster using a voltage that is higher than the steady-state value at the target speed.

  2. It can't accommodate minor difference in motors/ESCs/props. If one motor is slightly weaker than the others, it will spin slower under PWM magnitude control. The master attitude controller's integral term can correct for this, but why rely on this?

  3. Motor speed is only roughly proportional to applied voltage. The actual speed depends on the back EMF, which is V-IR. At higher loads, the speed per applied volt is lower. Does this matter? Maybe not, since none of this is linearly proportional to thrust anyway.
A classical approach to true speed control would be to wrap an outer PID control loop around everything:

The speed control loop, a generic PID with saturation, feeds the q-axis current reference a value between the maximum braking current and the maximum accelerating current, depending on the speed error. The d-axis controller is still simply trying to maximized torque per amp by adjusting motor phase. (The more I think about it, the more the mSCR strategy makes sense in this regard.)

This video shows the start-up and PID-based speed control for FFv1.0 with the 14x4.7SF props:

The nice thing about PID control is that it's easy to understand and easy to tune. Unfortunately, the load in this case is very nonlinear. As the speed increases, the damping due to propeller drag goes up. So, the step response changes depending on the speed:

Real data from the 14x4.7SF prop.
From 1,000-2,000rpm, the chosen gains exhibit a lot of overshoot. From 4,000-5,000rpm, they're overdamped. With this data as a starting point, I made a quick simulation including the nonlinear prop load. Another nice thing about PID is that it's very easy to work with in a Simulink:

This is what grad students do all day.
After some shady parameter estimation, I could almost predict the step response shapes:

The real-life response is actually a bit faster than the simulated response, for some reason. Not going to complain about that. But, they're both pretty slow overall. The motors never reach peak current during speed steps. The slew rate can and should be much faster for the best possible speed command tracking. By tweaking the gains in Simulink, I can now get a feel for which way they should be pushed in real life to speed up the response and kill off some overshoot.

I also think there's a better way. This post is already quite long so I'll have to save that for next time, but there is a hint of it in the Simulink diagram...

For now, I leave you with a Kramnikopter that I bought for basically nothing and will be doing absolutely zero custom development for:

Yay, not-complicated flying things.

Friday, April 13, 2012

4pcb: 3S/370mAh Testing

Previously, I promised some PCB quadrotor testing on a 3S/370mAh battery instead of its former 2S/460mAh. They're both Turnigy nano-tech packs with a 25-40C discharge rating, so I am only trading mAh capacity for more voltage. In actuality, the 3S does store more energy (1110SmAh vs. 920SmAh to use a nonexistent unit of energy). Accordingly, it's also physically larger:

3S/370mAh (left) vs. 2S/460mAh (right)
And, it weighs more. The 2S/460mAh weighs 30.0g, while the 3S/370mAh is 36.6g. (In both cases, this comes out to about 113Wh/kg, which seems reasonable for a small power-cell pack where much more of the weight is packaging and connector.) Generally, adding 6.6g of weight to a micro quad is going to hurt a little, but if you're going to add weight somewhere, the battery is a good place to do so.

Theorem 1: "LiPoCopter Theorem" As the ratio of battery weight to total weight (LiPoCopter ratio) of an RC helicopter/multirotor increases, it is increasingly likely to be able to fly with properly chosen motors, props, and ESCs.

Proof: Consider the extreme case: a helicopter made out of a LiPo (possibly one of these, with two rotors attached to it like a LiPo-bodied Chinook). It is certainly possible to choose a set of components {motor, prop, ESC} which will allow this to fly. As any helicopter/multirotor approaches this ratio of battery to total weight, there will be some set of components which allow it to fly.

A respectable LiPoCopter Ratio of 26.4% (formerly 22.8%).
Of course, I don't get to choose an arbitrary set of components for 4pcb; I'm stuck with the existing motors, props, and controllers. I had a hunch that it would fly better, though, since the ESCs would very much like to see more voltage. They're rated for 7-42V, so running on 2S (7.4V) is very borderline. 3S (11.1V) gives much more margin for voltage sag under load. The weakest motor would tend to reach command saturation well before the 2S battery was dead, leading to drifting and ultimately instability.

The improvement on 3S was immediately apparent, with no additional work required. Here's a comparison of the motor commands for full-length flights on 2S and 3S:

In the 2S/460mAh flight, the average command to hover starts at about 200 (out of 255) and increases as the battery voltage drops, up to about 220. The problem is the front motor: after some time, it begins to saturate and when it does, it can't produce any extra thrust to counteract for disturbances. In flight, this causes the quad to pitch forward, especially on throttle-up, and makes it difficult to fly. You can see I had to land several times to let it re-stabilize. The total flight time was about 6 minutes, but the last 2-3 minutes were me fighting to keep it in the air. Usable flight time was about 3-4 minutes and even at the end of the flight, the battery was still above 7.4V (only 50% discharged).

The 3S flight is one continuous stretch of over 8 minutes in the air. The average command to hover starts at about 165 and increases to about 185 as the battery voltage drops. This leaves plenty of margin for stability, so even though the front motor command is still highest, it never saturates. The problem of uncontrollable pitch as the battery gets discharged is eliminated. One side-effect of having higher voltage is that all the gains are now too high, causing small/fast oscillations (compare the width of the lines in the above images). This should be easy to fix. After 8 minutes, all four motors start to taper off in power. And all of the battery capacity was used:

3S/370mAh after 8 minute flight. Total voltage: 9.02V.
Luckily, I bought four of the 3S/370mAh packs, so I now have enough capacity for 32 minutes of continuous flying. Just in time for demo season.

Sunday, April 1, 2012

TI Workshop + The Joys? of High-Speed Motor Control

I spent the past week at the Texas Instruments C2000 MCU Real-Time Industrial Control Training in Waltham, MA. It was really good workshop that I would highly recommend for anyone who wants to get into real time motor control. The next events are in Michigan and Wisconsin in May and June. The three-day event covers control theory on day 1, motor control specifics on day 2, and a C28x microcontroller intro on day 3, for which you get an F28069 Piccolo controlSTICK to keep:

In the interest of fair reporting, I will say the that feature set of the TMS320F28069 microcontroller does give it a few advantages over the STM32F103C4 that I currently use. It's a 32-bit processor (not ARM based) and can run at up to 80MHz. But, it has a built-in FPU and control-law accelerator, something not found until the F4 line of the STM32s. It also has more PWMs (16) and ADCs (16) than the smaller STM32F1s. As far as the available IDEs, I've used both IAR EW and Code Composer Studio before and I find them to be equally annoying. And I hate JTAG/emulated debugging anyway. 

I don't see myself switching away from the STM32, but I've had good experiences with the TI MSP430 line and I wouldn't hesitate to recommend those or the more powerful C2000's for people looking to escape from Arduinoland.

Since I'm always talking about how convenient it is to have a compact electric scooter that folds so you can take it on public transportation, I decided to try it out on my way out to Waltham. The bus from near me stops just on the other side of Prospect Hill Park from the TI offices:

No problem. This should be a nice traffic-free scooter ride through the pa-  
I guess I neglected to consider the fact that Prospect Hill Park might actually be named for its predominant topographical feature. Pneu scooter is good at a lot of things, but hills are not one of them.

So I guess I'll be hiking.
Also, some of the paths are just dirt trails. The good news is that that's the TI parking lot right there.
Well at least I could have a fast trip down the hill on the way home - Oh no wait, it tore my brakes apart on the first two days. By the third day, with repaired brakes, I had pretty well optimized energy usage so that I could minimize hiking up the hill and coast the entire way down. Still better than driving in Boston rush-hour traffic.

Anyway, although all three days of workshop were great, I was especially interested in Dave (Wisconsin) Wilson's Motor Control Seminar. Most of the slides are available on his blog if you are interested. I learned a bit more about TI's InstaSPIN BLDC control scheme, and am beginning to sort out the various types of sensorless control methods.

One thing I picked up which I have to think more about is that what I have previously referred to as an open-loop flux estimator is in fact not as far different from a closed-loop observer as I had thought. The difference is more subtle and possibly less fundamental than it seems. For example, here is a closed-loop single-phase back EMF/current observer with a PI compensator:

With just one page of math, I can unwrap the feedback loop and rewrite it in the form of my "open loop" flux estimator, plus an additional low-pass filter. The closed-loop form still has the advantage of keeping an internal state for current estimate, which is probably way cleaner than the measured current. But they're not fundamentally different as I had once thought. Go figure... 

Maybe the important distinction then isn't between open-loop and closed-loop formulation for the observer. Maybe the fundamental difference is between observers that work only on electrical quantities and ones that include a mechanical state (speed, position). The addition of a mechanical state would almost certainly improve the performance of either observer, since it would anticipate the trajectory of I and E as the motor rotates. No wonder this is the type of observer described in the Book of Mevey, which has been right about everything else so far.

Back to reality.

My immediate task is to get this thing spinning:

After realizing that I wouldn't be able to hear anyone coming over the noise of the propeller, I added this:


In the previous post, I had gotten FFv1.0 up and running using Pneu Scooter as a test motor. The TI DRV8301 gate driver/everything chip performs well. But, the first time I attempted to run the motors on the hexrotor, I ran up against some not entirely unexpected issues with high speed control. At full speed, the electrical frequency of these motors can be 1,000Hz or higher. Every part of the control loop from the currrent sensors to the digital filters has to keep up with this, and in fact exceed it by a large margin to minimize phase lag. This is a tall order with a 15.625kHz PWM-limited control loop.

What I found on the first attempt is that at high speeds, the flux estimate becomes noisy and its zero-crossings, which are used to mimic Hall effect sensor transitions in my sensorless method, become blurry and unreliable. As a result, the position estimate would become noisy, which would lead to incorrect voltage vector placement. This leads to even more noisy current which leads to an even more noisy flux estimate, and so on. The physical result would be a motor that sounded very unhappy and current measurements that are all over the place starting at just under 5,000rpm. Under load, it could lose sync and crash entirely, surviving only because it was running on a current-limited power supply.

As an attempt to remedy the problem, I'm trying an idea I first thought of a long time ago for use with Hall effect sensor-based field-oriented control. The algorithm already uses a time-based interpolator to work between one flux zero crossing (or Hall effect transition) and the next. Instead of trusting the new zero crossing's position estimate 100% of the time, it's possible to use an IIR-style filter to merge the interpolator with the new position. Maybe this will help:

When the new flux zero crossing or Hall effect sensor transition comes in, it generate an angle (blue) which previously would be taken as the true angle. Now, that blue angle is averaged with the current estimate based on time interpolation (brown) to create the new estimate (purple). The relative degree of weighting for the average is set by a parameter, A, just like and IIR filter.

At high speed, when the blue vector is popping up at the wrong time due to noise, this filter will help stabilize the rotor angle estimate. And what a difference it made, without any tuning whatsoever. Here's some video of spinning up the larger prop, a 14x4.7SF:

One interesting thing is that I haven't even written a startup routine yet, but most of the times it starts just fine. I guess I'm used to vehicle controllers where if you don't write a startup routine they just sit there and buzz angrily. Also interesting is that once it does start, it can run at as low as 370rpm, which is about 7.5% of the top speed. So that would be my target speed for exiting the startup routine and entering run mode.

I can tell from the data that the angle filter helped a lot - the current is much cleaner. Here's a plot of the flux and phase currents with respect to the estimated angle:

Things are mostly in the right place and the flux estimate has the correct magnitude. (The flux is related to the motor constant or the rpm/V constant.) Something fishy is going on with the phase of the flux estimate. Its peak should stay right on 180º, since that's being used to define the d-axis. But instead it starts to shift forward at high speeds. As it turns out, I messed up the filter code a bit, so I'll have to go back and fix this one later.

For now, I also did some benchmarking against a Turnigy Sentilon 100A HV, a very good brushless motor controller. I measured (using a clamp multimeter) the AC RMS phase current and the DC bus current at different speeds running with the Sentilon 100A HV and with FFv1.0:

The not-very-interesting result is that they pretty much overlap so far. So I am at least not doing worse then good-but-dumb brushless DC controllers. It's definitely possible that any advantage of FOC will be lost in these motors because they are so efficient in BLDC mode already. Typically, high-speed low-inductance motors can operate very well with square wave drive. 

But there are other advantages that I'm still counting on. One which I was reminded of accidentally is that this controller can do four-quadrant drive. It can, in fact, brake so hard that it can unscrew the prop adapter and send a propeller flying across the room. Having braking should improve the response time of the motors to negative speed steps, which can help the control be snappier.

I've only just barely scratched the surface of the number of parameters that can be tuned. My next task will be to explore the effects of changes to the various motor parameter estimates, the filters, and the limits. Then, to see how it responds to more dynamic inputs. Then collapse everything into a simple enough interface that it could be used by other people. But so far, it's been a relatively problem-free controller build for once...