[etischer]

Originally Posted by Greenflight

I don't think this exactly how v/hz works. It is not related to the actual speed of the motor, just the position of the throttle sensor. So, assuming you had a minimal acceleration curve, you could be supplying 230vac to the motor at 60hz and the motor could still not be turning.

Vector mode essentially uses a measurement of motor RPM to adjust the v/hz ratio to account for variations cause by load.

That's why v/hz isn't so awesome for applications involving load.

If you have 230vac @60hz and the motor is not turning, you will be drawing a ton of current. I don't think you can just connect a motor up across the line to 230vac. Here is a graph showing how v/hz works. At low speed you need to put some DC offset to gain low speed torque, here it is called "boost"

[Technologic]

latest schematic... all of my instincts are telling me it's ok the way it is... however, the IFB0 and IFB1 (current sensor inputs) need to be reprogrammed upon a measured output to cut off at the voltage ... what voltage that is I can't say.

I'm also not 100% sure that the optoisolator will keep the gain the same for both channels... I may need a low voltage opamp in line instead... still weighing options.

Edit:
I've gone ahead and added it so you can see what's changed... the optoisolator on the current sensors won't stay (being replaced by two opamps to give a 2:1 gain)... just waiting for a reply to an email for choosing those.

http://img127.imageshack.us/img127/8287/part1m.png

I haven't decided on what outputs I want the controller to display... they can range from current/voltage sensors to full on projections onto LCDs with little to no real additional code changes, though the LCD would require decently expensive ($5-7 more) parts. But you can actually display the PWM signal, current sensor data and voltage data all at once on a screen with a DAC converter... pretty nifty.

With the option of using hall sensors OR encoders people can potentially modify industrial motors by adding hall sensors to each phase... a fairly cheap thing that could even be drawn up on another PCB for a few bucks (plug and play type deal)

[Technologic]

ok current sensing stage is finished... at least I believe It'll be hands off on it from now on.

Since you can change the registry in the MCK201 to shut off from the overcurrent at a certain voltage (say 5v from that opamp) I think I will be using the gatekill for another thing... specifically a temp sensor on the output stages...

http://img27.imageshack.us/img27/6242/part1cpi.png

Link to the nearly finished schematic (about 75%)

Gain is 2:1 on the 2.5v max outputs from the sensors (2.5v being the point the flux is saturated... or about 800A) so the opamp should output upwards of 5v. I'm much more pleased with the current sensing at this point. Large shielding will need to go around each sensor... but besides that tiny thing to remember it's workable.

If anyone wants to pitch in I need to design a 1A 20v PSU, 3.3v of like 500mA and a 5v of like 1A or so... regulated. Probably going to do that on a separate PCB and have a DC-DC supply for 12v as well (so people don't have to try and find 300+v DC supplies if they choose to run this that high). Not sure what all of the average car's accessories take up amperage wise... maybe 20A? should be fairly easy to make a SMPSU for that.

[Greenflight]

Originally Posted by etischer

If you have 230vac @60hz and the motor is not turning, you will be drawing a ton of current. I don't think you can just connect a motor up across the line to 230vac. Here is a graph showing how v/hz works. At low speed you need to put some DC offset to gain low speed torque, here it is called "boost"

That's exactly what I'm saying... Hertz is related directly to volts with a fixed ratio (except the boost, obviously). And volts is related directly to the position of the throttle pedal. Since the controller has absolutely no clue what the motor is doing, it just takes an educated guess using v/hz. So, if you floor the car from a dead stop, you will be supplying full voltage to the motor- and since the controller knows no better you will also be supplying full frequency.

In reality that's just a generalization, because most v/hz controller have a setting for acceleration so they don't go and do this. For a car, however, the acceleration would need to be extremely gentle to keep the controller from overshooting the motor.

Since vector control has a feedback system, it "knows" how fast the motor is turning, so it can still provide max voltage to the motor at low RPM at the same frequency at which a v/hz controller would only be supplying a couple volts. This is desirable because it allows near full torque and power across the whole rpm range.

Lookin good tech!

[etischer]

If you ramp up too quickly, the motor will slip and stall. Atleast that is what happens with my setup. It will probably work fine on a bench with no load, but when the inertia of the load is much larger than the inertia of the rotor, problems occur. My car feels like the motor is cogging when it slips beyond the point it can sync. Once I'm above 3% speed, everything works like a dream.

Originally Posted by Greenflight

That's exactly what I'm saying... Hertz is related directly to volts with a fixed ratio (except the boost, obviously). And volts is related directly to the position of the throttle pedal. Since the controller has absolutely no clue what the motor is doing, it just takes an educated guess using v/hz. So, if you floor the car from a dead stop, you will be supplying full voltage to the motor- and since the controller knows no better you will also be supplying full frequency.

In reality that's just a generalization, because most v/hz controller have a setting for acceleration so they don't go and do this. For a car, however, the acceleration would need to be extremely gentle to keep the controller from overshooting the motor.

Since vector control has a feedback system, it "knows" how fast the motor is turning, so it can still provide max voltage to the motor at low RPM at the same frequency at which a v/hz controller would only be supplying a couple volts. This is desirable because it allows near full torque and power across the whole rpm range.

Lookin good tech!

[aeroscott]

Originally Posted by etischer

If you ramp up too quickly, the motor will slip and stall. Atleast that is what happens with my setup. It will probably work fine on a bench with no load, but when the inertia of the load is much larger than the inertia of the rotor, problems occur. My car feels like the motor is cogging when it slips beyond the point it can sync. Once I'm above 3% speed, everything works like a dream.

well put this is the area of concern . LTI says you can get a huge inductive load heating the rotor to the melting point ( extreme case ) .they went to switched reluctance to solve this ( very little rotor heating ) . My hope is this controller will be able to s. r. with little mods( diodes are added it cannot do a igbt to igbt short ) . but svm solves some of this problem with very low freq. and voltage ( like 3 hrz. at start up )

[etischer]

Originally Posted by aeroscott

well put this is the area of concern . LTI says you can get a huge inductive load heating the rotor to the melting point ( extreme case ) .they went to switched reluctance to solve this ( very little rotor heating ) . My hope is this controller will be able to s. r. with little mods( diodes are added it cannot do a igbt to igbt short ) . but svm solves some of this problem with very low freq. and voltage ( like 3 hrz. at start up )

I've never heard of Switched Reluctance mode, but sounds like the motor has to be a Switched Reluctance motor. Sensorless vector should solve the problem.

[aeroscott]

Originally Posted by etischer

I've never heard of Switched Reluctance mode, but sounds like the motor has to be a Switched Reluctance motor. Sensorless vector should solve the problem.

yes the motor is moded it's 3 phase ( or more ). my understanding sensors are a must for a ev . you need good position indication put put the wright power at the wright time . the controller needs to see the effect as the motor picks up speed . this timing needs to be vary good . check google for s.r. motors .

[Greenflight]

Originally Posted by aeroscott

well put this is the area of concern . LTI says you can get a huge inductive load heating the rotor to the melting point ( extreme case ) .they went to switched reluctance to solve this ( very little rotor heating ) . My hope is this controller will be able to s. r. with little mods( diodes are added it cannot do a igbt to igbt short ) . but svm solves some of this problem with very low freq. and voltage ( like 3 hrz. at start up )

Most 3-phase controller ICs have an adjustable parameter for dead zones to prevent IGBT-IGBT shorts... I don't think that will be a huge problem.

I can imagine that high current, low RPM situations are bad for an induction motor the same way they are bad for a series motor- just sitting there and pulling a bunch of current is bad for anything. In that sense, I can see a small amount of throttle ramping being a good thing. Current limiting might be more useful though.

[Technologic]

Originally Posted by Greenflight

Most 3-phase controller ICs have an adjustable parameter for dead zones to prevent IGBT-IGBT shorts... I don't think that will be a huge problem.

I can imagine that high current, low RPM situations are bad for an induction motor the same way they are bad for a series motor- just sitting there and pulling a bunch of current is bad for anything. In that sense, I can see a small amount of throttle ramping being a good thing. Current limiting might be more useful though.

In most situations throttle ramping would be the way to go, but in a DSP controlled AC controller that allows easy registry shifts (especially deadtime etc shifts with just inputed parameters) it's pretty easy.

On a separate note there's no way I can accurately design a Switchmode PSU without a spice program... and since the last thing I want to do is hand input MOSFET and other parameters, I need to figure out a way to download Pspice again.