Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[e*clipse]

Ok - a couple more dummy questions from the peanut gallery...

I can see why one would want to *precisely* tune the PI loop for a servo motor. I WANT MY ROBOT arm ***here*** yesterday!

Other things can have much more relaxed requirements. I'm setting up a VFD for an industrial ag pump. In this case it's actually better to slowly increase speed.

It seems to me there's lots of room for a more relaxed set of requirements in a car. For example, there's all the slop (gear mesh, etc) and inertia of the drivetrain. Then there's the mass of the car. It seems that even with a perfectly tuned servo-motor type control, you'll still be limited by physics for the car.

Could the auto(mobile) tune make use of this?

Also, I'm looking for some good information about spec'ing the bus capacitor. I'll be using the SBE film capacitors. However, I haven't been able to find good info about capacitor size - just a lot of hand waving and some questionably wrong papers.

- E*clipse

[MPaulHolmes]

config (or settings) can show anything you want! How about this for starters:

config
p=xxxxx i=xxxxx
min-regen-position=xxx max-regen-position=xxx
min-throttle-position=xxx max-throttle-position=xxx
fault-throttle-position=xxx
max-battery-amps=xxx max-battery-amps-regen=xxx
max-motor-amps=xxx
current-sensor-amps-per-volt=xxxx
precharge-time=xxxx
rotor-time-constant-index=xxx
raw-throttle-position=xxx
(any others that come to mind?)

[MPaulHolmes]

I think in the ACIM case, the model isn't perfect for trying to track the rotor flux angle, so there is always some error in the feedback for Id and Iq. It may be that under dynamic situations that Id and Iq feedback get grossly pushed away from what they are being commanded to be, and maybe then it's like trying to hit a rapidly moving target while gently pushing them back to where they should be. Like a classroom of psycho kindergarteners, with a teacher that's 120 years old, and moves in slow motion. haha. This is actually a really interesting question. I don't know if Id and Iq really are that badly behaved, constantly needing a good smack upside the head to keep them from rapidly leaving their assigned post. The way I see it though is, with a perfectly tuned PI loop, you have perfect control of torque, so you can make the pump start up as gently as you want, or if you want to be all crazy, you can go from 0 to 75amp in < 0.002 seconds with no overshoot. That's how fast thingstodo's PI loop is tuned for now. I think he will beat a tesla off the line with that response. Or he will have to get a new transmission. lol

[thingstodo]

Originally Posted by MPaulHolmes

config (or settings) can show anything you want! How about this for starters:

config
p=xxxxx i=xxxxx
min-regen-position=xxx max-regen-position=xxx
min-throttle-position=xxx max-throttle-position=xxx
fault-throttle-position=xxx
max-battery-amps=xxx max-battery-amps-regen=xxx
max-motor-amps=xxx
current-sensor-amps-per-volt=xxxx
precharge-time=xxxx
rotor-time-constant-index=xxx
(any others that come to mind?)

Looks good. How about raw-throttle-position? It would help if there is a throttle fault to know whether it is high or low, and to see if it changes when you change the position of the throttle.

[MPaulHolmes]

With the 500amp DC controller, all the feedback stuff was put in a datastream, separate from the configuration stuff. Maybe we could include a separate datastream for things like Id, Iq, duty1, duty2, duty3, ... Raw throttle does seem like it would be more related to the configuration stuff than the other types of feedback though. It would help people in deciding what they should set their various throttle positions to. That's a really good idea!! Thanks!

BAM! it's now included.

[thingstodo]

Originally Posted by e*clipse

Ok - a couple more dummy questions from the peanut gallery...

If you are asking, others are wondering so thank you for asking!

I can see why one would want to *precisely* tune the PI loop for a servo motor. I WANT MY ROBOT arm ***here*** yesterday!

Agreed

Other things can have much more relaxed requirements. I'm setting up a VFD for an industrial ag pump. In this case it's actually better to slowly increase speed.

It seems to me there's lots of room for a more relaxed set of requirements in a car. For example, there's all the slop (gear mesh, etc) and inertia of the drivetrain. Then there's the mass of the car. It seems that even with a perfectly tuned servo-motor type control, you'll still be limited by physics for the car.

A pump is not a big deal. We have applications with small pumps where we use Volts/Hz, a recipe instead of a tuned loop. You can change out a motor without retuning. In an emergency case, you can run two smaller pumps in parallel if you don't have a spare larger pump ... stuff like that.

For a screw conveyor (auger) or a conveyor, you have situations where the equipment stopped while fully loaded. So you need the regular torque (to move the product) plus some extra torque to start things moving, plus maybe more if things have cooled off and the product becomes harder to move. For this, you need the controller to track rotor position and to control the slip. If the slip gets too high (rotor not turning or torque has passed the knee of the curve) the controller needs to bump up the voltage and back off the frequency to avoid stall, then resume ramping up the frequency when the slip goes back to normal. We use a target speed instead of a target torque ... so things are a bit different ... but the controller controls torque to reach a target speed ...

Paul could likely re-phrase that description in terms of Id and Iq.

From my limited knowledge of the what's *REALLY* happening, that's the best I can come up with for the reasoning. The industrial controllers can somehow figure out rotation speed and slip based on measurements of the motor back EMF. We don't use encoders - they are unreliable in our conditions. I'm told that the sensorless control is a bit less accurate, but is accurate enough for our purposes of generating torque for startup. As you say, being out by a degree or so on the angle of the rotor is not an issue for us.

Could the auto(mobile) tune make use of this?

Not sure - just looking for the most torque I can get out of my motor

[thingstodo]

Originally Posted by MPaulHolmes

With the 500amp DC controller, all the feedback stuff was put in a datastream, separate from the configuration stuff. Maybe we could include a separate datastream for things like Id, Iq, duty1, duty2, duty3, ... Raw throttle does seem like it would be more related to the configuration stuff than the other types of feedback though. It would help people in deciding what they should set their various throttle positions to. That's a really good idea!! Thanks!

BAM! it's now included.

Thanks for the addition .. same day service ... you're SPOILING me!

A data stream would be a great addition as well.

Adam was going to send me a beta of his new RTDExplorer - I have not received a copy as yet but will be trying that out as soon as it arrives.

[e*clipse]

config (or settings) can show anything you want! How about this for starters:

config
p=xxxxx i=xxxxx
min-regen-position=xxx max-regen-position=xxx
min-throttle-position=xxx max-throttle-position=xxx
fault-throttle-position=xxx
max-battery-amps=xxx max-battery-amps-regen=xxx
max-motor-amps=xxx
current-sensor-amps-per-volt=xxxx
precharge-time=xxxx
rotor-time-constant-index=xxx
raw-throttle-position=xxx
(any others that come to mind?)

Originally Posted by thingstodo

Looks good. How about raw-throttle-position? It would help if there is a throttle fault to know whether it is high or low, and to see if it changes when you change the position of the throttle.

Maybe Temperature sensor stuff?

Also, I'm using a Prius throttle - there's both a high and low output fault zone. It also has two sensors that it uses *somehow* to detect if the sensor's working.

Also, you might have different configurations based on the configurations... For example, I'm going to use this with a 'Yota Interior Permanent Magnet motor. As far as I know, the rotor time constant isn't an issue. So, there might be a "motor type" configuration, with each motor type having specifics like a rotor time constant.

[e*clipse]

Originally Posted by thingstodo

Thanks for the addition .. same day service ... you're SPOILING me!

A data stream would be a great addition as well.

Adam was going to send me a beta of his new RTDExplorer - I have not received a copy as yet but will be trying that out as soon as it arrives.

You know, the OBDII type data stream would be really nice. Would it be possible to use the OBDII protocol? Maybe even use some of the voltage, temperature, etc sensors - it's all worked out and an industry standard. It seems working with the standard might actually save some effort vs inventing a new protocol.

[MPaulHolmes]

Is that CAN? I "can" include a "CAN" transceiver on the next board for just such emergencies. Did you hear about the guy who had too many peaches?

"We'll eat what we can, and what we can't, we'll can." haha