#$%@$#@ <--- (insert favorite swearword here), IT WORKS !!!!

[amberwolf]

Is there a member named 'Accountant'

Yes; that's almost certainly who he is referring to. Rarely posts.

[fechter]

This was probably posted elsewhere, but a nice application note on implementing sensorless field oriented control using a PIC30F6010A.

http://ww1.microchip.com/downloads/en/AppNotes/01078A.pdf

Perhaps a bit off topic, but I think the hardware would look almost identical. They also have a nice description of FOC and how it compares to the standard cheap 6 step control.

I think having a 'learn' mode that figures out all the relationships between the senors and BEMF waveforms would be brilliant.

As Luke found out with the Sevcon, these wonderful control loops can be quite a challenge to get dialed in. I felt the pain with a very simple PI loop on a current limiter circuit. Multiply the number of parameters, have them all interact, and I can see where the software really needs to help with dialing things in more.

If the code is available for the sensorless setup, it might be interesting to try.

[deVries]

If you want to try it out, PM me

PM sent! I gave you three options one of which allows you to borrow my motors, controllers, and power supplies as long as you wish, really, for 2012. You can actually install the motors & use them under load too. Just give the parts back to my cousin in Switzerland before next Christmas by early December or send back by post whenever you want to in 2012. That may be a hassle for you though, so consider this too...

Otherwise, I'll try to follow your instructions & buy the parts IF you want to work with me, so I can give you feedback results from my tests, etc. I'm sure I might be able to find someone else in the USA to help me with this too.

Thanks!

[Accountant]

Accountant is also offering to send you xie chang board to implement your chip inside ... Small help from Croatia.... with greetings from Tesla
If thats possible, you have the market for buying your chip ...

Is there a member named 'Accountant' or is the accountant of your company ? Tesla ?

Hello,

I do exist.

[Accountant]

Accountant is also offering to send you xie chang board to implement your chip inside ... Small help from Croatia.... with greetings from Tesla
If thats possible, you have the market for buying your chip ...

Is there a member named 'Accountant' or is the accountant of your company ? Tesla ?

Hello,

I am both.

[deVries]

Lebowski, how is your throttle working in given case? Changing multiplication factor to BEMF and then chip modifies PWM accordingly?

That is what I use currently, though the multiplication factor is fixed and I use my lab supply.
I want to go to (motor phase) current control though as close control / monitoring of the currents will prevent the killing of FET's and will make the system more user friendly (torque is proportional to current).

At your level of understanding will this be able to be accomplished within the next few months or

This has always been the major fault for performance boosters to push the limits w/o magic smoke... Probably, the most major fault on ES.

[Lebowski]

just added an option: now there is a choice between sensored and sensorless.

sensored means hall sensors are used. The motor can provide torque at standstill.
At a certain very low minimum rpm (60 or so) the motor will make a nice gradual
transition to sensorless.

sensorless means no hall sensors are used, this mode can be used when hall sensors
are not connected and/or are not available. A small nudge (of a few e-rotations)
is enough to get the motor spinning. A safety has been built in to make sure the
motor spins only in one direction, it can get very messy if you accidently push
your bike backwards while giving it some throttle
In sensorless mode pure sine waves are used to drive the motor. This because
hall sensor synchronisation is used during the calibration phase to measure the
back-emf.

[deVries]

just added an option: now there is a choice between sensored and sensorless.

Excellent Brilliant.

After reading through this thread I can see the potential this has especially for DIY motor makers. Your design is very fault tolerant AND the motor does NOT have to be precision made. Simply Awesome.

Also, and everyone should appreciate this , your progress of adding ideas with features is simply amazing. I am thrilled at the speed of potential progress you're making... ES members really need to take notice of this thread AND START TESTING this with other motors ASAP.

Lebowski, wayyy better than the movie.

[parabellum]

just added an option: now there is a choice between sensored and sensorless.

ES members really need to take notice of this thread AND START TESTING this with other motors ASAP.

Lebowski, wayyy better than the movie.

Believe me, ES members are impatiently waiting to get hands on this one.

[Lebowski]

Since I can now do sensorless I modified a small RC motor I have
lying around. It's seen some abuse (notice the wiggle in the rotor)
It's a Scorpion 4025 with the factory windings removed.
(4025 = 40 mm diameter, 25 mm magnet length). It was much harder
to start sensorless (needs messing with the parameters) and took several
attempts but I got it going. My bicycle motor starts very very easy but
after watching the video I think you'll agree this RC motor is a bit
too extreme I did this test just to indicate the capabilities of my
controller.

The video shows the motor running, first fast, then very slow. After that
it shows the (low inductance ) stator windings and the output stage.

33 V supply, 97 kHz PWM, 40kHz loop frequency, current controlled.

[Arlo1]

Since I can now do sensorless I modified a small RC motor I have
lying around. It's seen some abuse (notice the wiggle in the rotor)
It's a Scorpion 4025 with the factory windings removed.
(4025 = 40 mm diameter, 25 mm magnet length). It was much harder
to start sensorless (needs messing with the parameters) and took several
attempts but I got it going. My bicycle motor starts very very easy but
after watching the video I think you'll agree this RC motor is a bit
too extreme I did this test just to indicate the capabilities of my
controller.

The video shows the motor running, first fast, then very slow. After that
it shows the (low inductance) stator windings and the output stage.

33 V supply, 97 kHz PWM, 40kHz loop frequency, current controlled.

What inductance? 97khz? how warm do your fets and fet drivers get? You should put a load on it. Maybe put some heat sinks on the fets first.

[deVries]

Wouldn't it have been much easier to start with more poles wound? 20 pole RC motor vs 10 pole?

Sounds Sweet. You took it all the way up to 33v at how many amps?

[Lebowski]

Got it dailed in a bit more now You can see how easy it is to (sensorless)
start this 3 x 5 windings RC motor. The laptop (next to my ugly mug in the
reflection) shows the real-time phase information available in the algorithm.
Notice when i spin the motor by hand to start it, the algorithm almost immediately
knows the phase.



to Arlo: the 97 kHz is pretty random, it's a non-integer multiple of 40kHz (the main
loop frequency). In the above vid I changed it to 77 kHz. At 77 kHz the output
stage takes 10 W to switch, the motor takes an additional 5 W at 2000 rpm.
It takes (additional) unmeasurable power at slow.

[Lebowski]

Wouldn't it have been much easier to start with more poles wound? 20 pole RC motor vs 10 pole?

Sounds Sweet. You took it all the way up to 33v at how many amps?

that's the whole sport, to use a motor with minimal back-emf and low L and still get it to work

[Arlo1]

I just found these boards for someone who wants to see if they can use this chip and thier own powerstage to make somethign cool!
http://www.ebay.ca/itm/MCU-BOARD-DEVELO ... 1c200b3dcb

[deVries]

Time for a friction belt like on an exercise bike or some kind of quick & dirty load... hand with thick gloves on... rim/disc brakes... anything!

[flathill]

Wouldn't it have been much easier to start with more poles wound? 20 pole RC motor vs 10 pole?

Sounds Sweet. You took it all the way up to 33v at how many amps?

that's the whole sport, to use a motor with minimal back-emf and low L and still get it to work

Sweet! That is low L for sure...think it would work with one of these ultra low L motors:



"The stator is made from 29 gauge, M15 electrical steel laminations that are stacked and bonded. Trapezoidal bars of aluminum are inserted into the stator core, exceeding 90% slot fill (compared to less than 50% for motors with traditional copper windings)."

These type of motors would cease to give low wage countries an advantage (no hand winding)

I wonder if you could combine your tech with a current source inverter http://www.ornl.gov/adm/partnerships/ev ... ersion.pdf

[liveforphysics]

We have a solid-slot motor like that at work. It actually comes with giant inductors made to mount in-line with each phase lead to make it kinda/sorta controllable.

The work that Lebowski and RickNZ are doing here with ultra high speed PWM and current control is the key to enabling designs like solid slot motors.

[flathill]

Yup they are laying the groundwork for when SiC MOSFETS come down in price

What are you using that motor at work for?

[Lebowski]

We have a solid-slot motor like that at work. It actually comes with giant inductors made to mount in-line with each phase lead to make it kinda/sorta controllable.

The work that Lebowski and RickNZ are doing here with ultra high speed PWM and current control is the key to enabling designs like solid slot motors.

Isn't Ricky building the (known) sensorless FOC as decribed in the Microchip / NXP application notes ?
Here I'm using something that is not sensorless FOC. There are no extra inductors in series with the
RC motor. Tomorrow I'll try to see what happens if I go to 3 x 4 windings or even less...

High speed PWM... we're talking like anything over 20kHz is fast but audio class D amplifiers work at 300
to 700 kHz.

[Lebowski]

sensorless startup



It's a variant of forcing commutation. A certain torque generating current is fed to the motor.
The phase determining algorithm is very sensitive, even a small wiggle of the motor will tell
it whether we're going in the right direction or not. Wiggle in the wrong direction -> phase
remains constant. Wiggle in the correct direction -> phase is increased. The wiggle in the
wrong direction will at a certain point reverse (as the constant phase means the motor is forced
in a certain position, if it wiggle's in the wrong direction it will 'bounce' back, as if an elastic
band is attached to the rotor). A wiggle in the correct direction will advance the phase, encouraging
the wiggle to continue in that direction). Net effect of this is that the motor starts up.

The video shows both immediate startup (initial wiggle is in the correct direction) or a 'bounce back'
for an initially wrong wiggle. Notice that in all cases the motor spins in the same direction in the end.

What happens if initially there's no wiggle ? The noise in the phase algorithm will 50/50 detect a
correct / incorrect wiggle. So 50% of the time the phase is increased. This will drag the rotor along
which is interpreted as a 'correct' wiggle -> motor starts.

This is very useful for if you have a Mac motor with a one-way clutch in it, with this startup
addition the controller can now operate the Mac in sensorless mode.

It's quite stubborn and would probably be able to start under load (notice its pulling in one direction) :


HAPPY NEW YEAR EVERYBODY !

[deVries]

This is very useful for if you have a Mac motor with a one-way clutch in it, with this startup
addition the controller can now operate the Mac in sensorless mode.

It's quite stubborn and would probably be able to start under load (notice its pulling in one direction) :

HAPPY NEW YEAR EVERYBODY !

My Mac says Happy New Year to You!

I'm really very excited about your progress...

Thanks

[flathill]

Damn dude I think you are a natural! Time to build dyno

What is your special signal to initiate the wiggle on sensorless startup? You can try signal injection at 20khz or more ideally a resonant frequency of the coils to sense position so there would be no initial jerkback

[Lebowski]

powertest on my hometrainer 'dyno'

Sensorless startup, under load. The power supply that I have is almost maxed out during this test.
6 FET uncooled output stage (when you're not running it for too long of course).

I see $700 worth of batteries in my future

You can hear when the sinewaves start clipping by the sound the motor is making. The rateling
noise is the motorchain tensioner, is has one of those rattling derailleur wheels in it. In the
second half of the video the mechanical output power , powerstage and supply voltage/current
is shown. Efficiency on the flay looks good . If you're wondering, the pedals of the bike
need to turn for the hometrainer/dyno to work, else it cut out thinking you're taking a break.

[deVries]

Outlandish

At your pace within a month or two we'll have 100v 100amp out of this beast