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

[Arlo1]

Hey lebowski. At low rpm can we program a max on time for both fets??? I am wondering how this all works... If the max on time was say 20uS then they both shit off for ~5 uS to cool down its not the same as the PWM of the one side... Maybe I need to studie the center aligned up down PWM control to understand this all better.

[Lebowski]

Hey lebowski. At low rpm can we program a max on time for both fets??? I am wondering how this all works... If the max on time was say 20uS then they both shit off for ~5 uS to cool down its not the same as the PWM of the one side... Maybe I need to studie the center aligned up down PWM control to understand this all better.

The nasty bit is that the FET's don't determine the current, the inductors do. So you can shut off the FET's but then the current
will just take the path through the protection diodes, dissipating more than if you would have just kept the FET switched on.
So, to answer your question, no this option is not there as it does not make sense...

[Lebowski]

Sorry for the slow reporting, I've been very busy with adding the setup menu's and swapping my current sensing
resistors out for proper current sensors. I'm thinking about spending this (rainy windy snowy) weekend swapping
around all the analog channels and adding 'room' for a third current sensor so that I have sensors in all motor phases.

I have to see whether the third current sensor makes sense... theoretically you would say 'no' but practically, the
third sensor makes it possible to compensate for the common-mode voltage offset (which may be temperature
dependent) and common mode voltage spikes on the sensor outputs / ADC reference voltage. If i put in the third
sensor it'll be an option in a setup menu, you can choose whether you have 2 or 3 sensors.

I also just got myself a present, my GF for some reason 's been comparing me to Sheldon from Big Bang Theory
after i showed her this

DSC00705.JPG (169.72 KiB) Viewed 337 times

[Arlo1]

Hey lebowski. At low rpm can we program a max on time for both fets??? I am wondering how this all works... If the max on time was say 20uS then they both shit off for ~5 uS to cool down its not the same as the PWM of the one side... Maybe I need to studie the center aligned up down PWM control to understand this all better.

The nasty bit is that the FET's don't determine the current, the inductors do. So you can shut off the FET's but then the current
will just take the path through the protection diodes, dissipating more than if you would have just kept the FET switched on.
So, to answer your question, no this option is not there as it does not make sense...

If the fets are shut off for long enough then yes they stop the current from flowing other wise my bike which has had a powered up controller for 6 months without use would be dead. When the fets are on the current climbs but when they are off the current falls and the inductance determins how fast!

[bearing]

the third sensor makes it possible to compensate for the common-mode voltage offset (which may be temperature
dependent)

These Allegro sensors (at least the bigger ones that I have been playing with) have a built in ferrite core. Remanence in the core makes the voltage at 0 amps differ from 0.5Vdd. If the last direction of magnetization was positive, sensor will output some mV in the positive direction at rest. This makes it hard, or maybe even impossible, to measure small currents. If you are integrating current over time (like I did) it will accumulate to a big error.

[c_a]

We need a quick solution because my 135-50 inrunner is on the way:
http://www.powerditto.de/monster.html

We will run it at 24s lipo @450amps = 40kW

[c_a]

Is one enough or should I buy two?
http://www.mitsubishielectric.com/semic ... u-3a_e.pdf

[Arlo1]

We need a quick solution because my 135-50 inrunner is on the way:
http://www.powerditto.de/monster.html

We will run it at 24s lipo @450amps = 40kW

Good luck man i have needed something fast for 2 years!!!

[Arlo1]

Is one enough or should I buy two?
http://www.mitsubishielectric.com/semic ... u-3a_e.pdf

You will be lucky to get 1/4 there rating so... Get 6. How much are they?

[c_a]

I trust in mitsubishi spec sheets, found them for 347€ each in germany. I really like the big blocks

[bearing]

Is one enough or should I buy two?
http://www.mitsubishielectric.com/semic ... u-3a_e.pdf

Do you know if anyone have used those for high frequency switching? They have a pretty big gate capacitance, so you'll need a stiff gate drive.
Spec says it can handle 1kW of losses, but since thermal resistance is 0.1°C/W, that means the junctions will be 100°C above case temperature. So, I don't think you can use them close to their power rating.

I wonder if every "transistor" is made of several chips in the "normal" power mosfet size, or if it's one giant chip per transistor.

[Arlo1]

I trust in mitsubishi spec sheets, found them for 347€ each in germany. I really like the big blocks

IR and ixys have the best fets most have us have tested so far and their spec sheets are very optimistic. Remember when the fet is on it can only touch the max ratings then you have to turn them off and wait till the current falls enough so you can turn them on again so you wont see 300 amps 150 v from each unit more likely ~80amps at ~130v each unit and thats 80 phase amps!

[Arlo1]

I wonder if every "transistor" is made of several chips in the "normal" power mosfet size, or if it's one giant chip per transistor.

I took some big ixys fets apart and they were 8 smaller fets inside.

[Lebowski]

the third sensor makes it possible to compensate for the common-mode voltage offset (which may be temperature
dependent)

These Allegro sensors (at least the bigger ones that I have been playing with) have a built in ferrite core. Remanence in the core makes the voltage at 0 amps differ from 0.5Vdd. If the last direction of magnetization was positive, sensor will output some mV in the positive direction at rest. This makes it hard, or maybe even impossible, to measure small currents. If you are integrating current over time (like I did) it will accumulate to a big error.

The small Allegro current sensor work in the same way as the big ones. I'm not integrating currents over a long time so it's not a big issue, it's more that I want to build the best controller IC possible, and for this reason I am considering adding the option to use 3 current sensors. It adds the cost of a 3rd sensor but makes it more robust against (drifting, maybe temperature dependent) offsets. The effect of offset is that, because the throttle is current based, at throttle closed you get a tiny bit of either regen or power. Not a big issue but something I want to fix.

[Alan B]

Phase currents are AC so compensating for DC offset should be possible.

[Arlo1]

Lebowski I was thinking I was going to put a current sensor in each phase wire anyway I just wanted them to be ballanced.

[Lebowski]

Phase currents are AC so compensating for DC offset should be possible.

I thought about that one. It is possible to run a low pass filter and get the DC offset. Another possibility
would be to run a calibration routine where you make sure all FET's are off (so no phase current)
and measure the DC offset.But I realised going the 3 current sensor route actually has the additional
advantage of also eliminating high frequency noise common in all 3 sensors.
Imagine, the current sensors are in the motor phase lines. all the PWM signals are generated
by the same source meaning that there will be instances where all high side FETs and at other times
when all low side FETs are on. The voltage on the current carrying path of the sensors goes up and
down by about 100 V (I'm planning to run 100V battery). The isolation from current input to voltage
output is not specified but is not going to be an infinite amount of dB. Another sources of noise
which 3 sensors can compensate for is ripple on the 5V supply.
Something else which speaks against using just 2 sensors it that in this case the 3rd motor phase current
is compute as the negative of both other currents combined. So this computed value is going to carry
the combined error of both sensors. While with 3 sensors part of the error is common to all 3 and will
be cancelled out in the 30F.

If you're not versed in analog design it might all sound complicated and stuff but differential signal
processing is very common in integrated analog circuits...

I just ran the motor with 3 current sensors, results are excellent. The 30F will have a menu option
where you can chose 2 or 3 current sensors. For the case of 2, there'll be a small potentiometer
in the schematic that will need to be calibrated. The main difference between 2 and 3 sensors ?
2 sensors will result in a bit less accurate control when motor rpm and currents are low (so low
motor speed and low throttle). Any other case there won't be much of a noticable difference (maybe
some small amount of torque ripple in the 2 sensor setup, but there are coefficients that you can adjust
to reduce this)

Engineering pride though requires me to strongly recommend 3 current sensors

[Lebowski]

Lebowski I was thinking I was going to put a current sensor in each phase wire anyway I just wanted them to be ballanced.

Good thinking... But looking at my own motor, there's so much variation in magnet strength between the magnets
that in practise you probably won't notice the missing current sensor in 1 motor phase... In my case beer was consumed
during the motor's construction, not all motor phases have the same amount of windings

[Lebowski]

independent auto calibration added to the 2 current sensor setup

This removes the need for voltage calibration with a small potentiometer,
this simplifies the schematic and removes the need for accurate voltage
measurements / adjustments.

[Lebowski]

Been very busy adding more and more menu options. So much in fact I'm starting to get lost

this is the main menu:

Code: Select all

a] calibrate hall sensors
b] determine coil positions
c) PWM parameters
d) current settings
e) control loop parameters
f) throttle setup
g) running modes
h) CAN bus setup
z) store parameters in EEPROM for motor use


this is the 'running modes' submenu

Code: Select all

a) sensored or sensorless: SENSORED
b} sensorless startup: PUSH START
c} e-rpm limit sensorless self start: 999
d} e-rpm reached before transition: 89 %
e} minimum current push start: 0.9 A
f} push start current, error allowed: 20 %
g] erpm sensored to sensorless transition: 500
h] transition time sensored to sensorless: 249 milli-sec
i) return to motor start below 200 erpm
j) motor maximum: 6.99 k-erpm
k) motor standstill voltage threshold: 1.49 V
z) return to main menu


Some menu's have gotten so long that I've decided to switch all to letter instead of numbers. Also every option
is indicated by a ']', '}' or ')', the meaning is:
]: option only used for sensored
}: option only used for sensorless
): option used for both sensored and sensorless

This week I hope to be able to start drawing the schematic....

[Arlo1]

6.99k erpm max??? Thats only 699rpm for colossus....

[Lebowski]

6.99k erpm max??? Thats only 699rpm for colossus....

These numbers are for my own motor, you can type whatever you want here. This
particular entry I think goes up to 255.99k and is not limited by the controller algorithm
but by the computation which takes the number you enter and transforms it into the
16 bits integer the algorithm uses. Is there a need to go higher than 256 k-erpm ?

[Farfle]

The joby motors are some insane E-rpm iirc, but I cant find it.

[Arlo1]

6.99k erpm max??? Thats only 699rpm for colossus....

These numbers are for my own motor, you can type whatever you want here. This
particular entry I think goes up to 255.99k and is not limited by the controller algorithm
but by the computation which takes the number you enter and transforms it into the
16 bits integer the algorithm uses. Is there a need to go higher than 256 k-erpm ?

I doubt it but colossus will need <100KERPM so we are good!

[mauimart]

The joby motors are some insane E-rpm iirc, but I cant find it.

From the Joby thread it appears we are looking at 26 pole pairs. That's 9800 mechanical rpm with the 255k-erpm. Seems like that will do...