Lebowski
10 MW
can now program TQFP's as well !
pins match up according to function:
pins match up according to function:
Lebowski's motor controller IC, schematic & setup, new v2.A1
- Thread starter Lebowski
- Start date
Futterama
1 kW
Yet again, excellent service from Lebowski! 
Futterama
1 kW
Lebowski, when calculating the RC filters for the current sensors, what is the "pi" in the formula in the schematic?
If we are talking RC filter cut-off frequency, what should that value be with respect to f_pwm and other parameters?
If we are talking RC filter cut-off frequency, what should that value be with respect to f_pwm and other parameters?
nieles
10 kW
My guess would be 3.14, and also the formula states: r*c smaller or equal to 1/(2*3.14*f_pwm )
With r*c smaller, this will give you a higher cut-off frequency. So the cut-off frequency of the low pass should be equal or higher than f_pwm.
With r*c smaller, this will give you a higher cut-off frequency. So the cut-off frequency of the low pass should be equal or higher than f_pwm.
zombiess
10 MW
Futterama said:
This kind of questioning is why I just started studying mathematics the last week (calc/trig/algebra) so I can better understand what the math is actually doing. I can visualize how frequency and time are connected but lack the proper skills to fully explain it, but I only started learning the math last week and have only put about 7 hours into learning so I don't feel too bad if I'm off the mark. Here is the best way I can see the relationship.
Frequency = velocity / wavelength
A sine wave is a plot of a rotating circle. If the circle rotates 1 time per second than it has a freq of 1hz. This can be illustrated by rotating a circular coil inside of a magnetic field, the output will be sinusoidal. This is circular rotation speed also known as angular frequency. How many degrees per second at 1hz = 360.
[youtube]Ohp6Okk_tww[/youtube]
Since we are dealing with circles, pi comes into play, circumference / diameter or 2*radius = 2*pi*r.
Angular frequency is w=2*pi*freq
So to calculate the first order (-3dB) point of a RC low pass filter you use f=1/(2*pi*t).
t= a unit of time which is determined by the RC network so the formula then becomes f=1/(2*pi*RC)
Lebowski solved for RC. Since we know t=RC we can treat them as a single variable and solve for both at the same time, i'm not going to do the work, but the result is his formula.
RC= 1/(2*pi*f)
I hope I got this mostly correct. I tried to break down each component until I could understand it which took me 1.5 hrs to type this. I can visualize the relationship perfectly in my mind but it's very hard to describe. There is relationship between circles, sine waves and triangles. To get a sine wave, imagine uncoiling a spring. When it's all together it's a circle (imagine it has no depth like a slinky does in real life), but as you u pick it up stretching it out it uncoils into a sine wave from a side view.
Even cooler is that if you let the spring hang in free air and stretch it out, then let it go, it will oscillate in a motion that is sinusoidal. The frequency it moves at is it's resonant frequency.
I'm missing one part of the relationship which relates to the right triangle functions (sin/cos/tan) and how it ties into the frequency formulas. Need to think on that a little more.
I am such a nerd.
nieles
10 kW
Jeremy, have a look at the imaginary number i (or j for the electrical engineers) and how it is used to solve any circuit made of R,C,L components.
hint: the real and imaginary voltages/currents will form a triangle
hint: the real and imaginary voltages/currents will form a triangle
zombiess
10 MW
nieles said:
Thanks nieles. Does my above explanation make sense to you?
I found a good explanation of caps and inductors operating by lagging the voltage -90 behind current. This makes more sense now. I was missing reactance +90 and -90, resistance 0 made sense.
Fig 2. at this link makes the relationship make more sense now, I got my triangle.
http://onderwijs1.amc.nl/medfysica/doc/ElectricalResistCapacitInduct.htm
The formulas I'm using all the time are starting to make so much more sense! I'm starting to visualize the mathematical relationships, this is awesome and kinda trippy! I didn't get the chance to take trig or calc in high school. Maybe I'm better off for it because I hated almost all my math teachers because they couldn't answer the questions I needed to understand how things worked, it was mostly just a set of rules one had to follow. I looked up trig the other day and saw SOA CAH TOA and an interactive circle plot showing the relationship with a right angled triangle and it just clicked, took about 2 mins and suddenly the Sin/Cos/Tan all made perfect sense.
I've been obsessed with math for the last week, been watching too many Neil Degrasse Tyson youtube videos and Cosmos. About a month ago I started researching how the physics of a MOSFET work, cool stuff (if your into this sort of thing). I was trying to understand a certain phenomenon I've been seeing when paralleling.
Futterama
1 kW
Of course! I was thinking it was some coefficient from the setup menu, I didn't see at as the mathematical constant π.nieles said:
Thanks, that was what I thought.nieles said:
7circle
100 W
- Joined
- Jan 29, 2008
- Messages
- 137
@Lebowski - nice work on the adabtable hall analyzer.
Wondering whyvthe change to the Phase Voltage curcuit.
I noticed the manual V2.30 still discusses the older 2k2 divider style.
Will 70 x Max_Bat be enough to limit input on analogue input?
For 100V Bat thats 3 R_mot resitors in series 7k x 3 is 21k
And 22k with 22k divider. So with Mid-point link via 21k to say 10V the mid pint is already a little over 5V.
Also what sampling rate on the analogue inputs do you need to average out the square wave of the Bridge Powee stage.?
Or are you expexting some analogue filtering?
On a tangent... could the third Current sensor be used for Bus current ro the Battery?
Or can the use of just 2 phase current and PWM ratio be used to work it out accurately.
Wondering if it xould be reported on the CAN .messages.
7C
Wondering whyvthe change to the Phase Voltage curcuit.
I noticed the manual V2.30 still discusses the older 2k2 divider style.
Will 70 x Max_Bat be enough to limit input on analogue input?
For 100V Bat thats 3 R_mot resitors in series 7k x 3 is 21k
And 22k with 22k divider. So with Mid-point link via 21k to say 10V the mid pint is already a little over 5V.
Also what sampling rate on the analogue inputs do you need to average out the square wave of the Bridge Powee stage.?
Or are you expexting some analogue filtering?
On a tangent... could the third Current sensor be used for Bus current ro the Battery?
Or can the use of just 2 phase current and PWM ratio be used to work it out accurately.
Wondering if it xould be reported on the CAN .messages.
7C
Lebowski
10 MW
saves a capacitor,less pulling of the midpoint.7circle said:
oops, my bad
The idea was to have about 3 to 5 mA in those resistors
The algorithm doesn't sample the motor voltages, it looks at the 3 motor phase currents
no
that what it does now, for the battery current limiter function it calculates the battery current based on phase currents and PWM ratios
nieles
10 kW
awesome progress Lebowski!!
have you thought about adding a bootloader? for software updates
have you thought about adding a bootloader? for software updates
Futterama
1 kW
Lebowski, I can only get my motor spinning in toneless with recovery mode. In the other 2 modes, the chip won't leave drive_0 even when check for motor standstill is disabled. So do the controller need the input on pins 5,7 and 8 to work in HF tone and non-recovery toneless mode?
Lebowski
10 MW
the standstill disabled only works in recovery (thats why this menu is called 'recovery only'). In the other modes it looks at the motor voltages on the pins you mention.Futterama said:
what may work is a combination of a small resistor divider to put 2.5 V on all 3 pins to fool the chip, or maybe just putting a motor standsill voltage of 2.5 or higher in the misc menu.
For hf i am getting the feeling the autocomplketed current setting are not ideal, i would now set the fixed hf part at 25% of max phase amps, the proportional factor to .5 and the phase current in drive 2 to half the max phase current... make sure the motor force current is high enough and for hf you need to run the current sensor gain calibration.
Futterama
1 kW
If I enter 2 or anything above, I'll end up in the erpm menu and the setting won't be saved.Lebowski said:
Code:
########################################
# (c)opyright 2014, B.M. Putter #
# Adliswil, Switzerland #
# bmp72@hotmail.com #
# #
# version 2.21 #
# experimental, use at your own risk #
########################################
0) mode: Toneless, with recovery
a) PWM parameters
b) current settings
c) throttle setup
d) erpm limits
e) battery
f) current sensor calibration
g) control loop coefficients
h) filter bandwidths
i) FOC motor impedance
j) CAN setup
k) recovery only
l) miscellaneous
z) store parameters in ROM for motor use
------> l
a) autocomplete
b) motor standstill voltage threshold: 1.98 V
c) low side pulsing in drive 0: enabled
d) low side pulsing rate: 20 Hz
e) low side pulsing width: 20 usec
f) wiggle range: 9 deg
g) wiggle rate: 99 Hz
h) minimum # of cycles going from drive 2 to 3: 1000
i) number of cycles going from drive 3 to 2, HF only: 200
z) return to main menu
------> b
new value -> 2
a) erpm limit, forward: 99.97 k-erpm
b) erpm limit, reverse: 19.98 k-erpm
c) accept direction change below: 75 erpm
d) transition erpm drive 2 -> 3: 976 erpm
e) transition erpm drive 3 -> 2: 488 erpm
z) return to main menu
------>Lebowski
10 MW
hey thats funny, 'll have to look into that i think you'll have to use the resistive divider as mentioned before
i think you'll have to use the resistive divider as mentioned beforeFutterama
1 kW
I'll add the resistive divider to my next brainboard then, thanks 8)
Futterama
1 kW
How is the erpms measured using the PWM output from the chip (see the scope shot in my thread)?
Lebowski
10 MW
you cant really as the pwm is a fixed frequency... what you can try is to filter the pwm signal with a 2khz RC filter so see the sine content.Futterama said:
or you look at the system variable for phi_int with the rs232 realtime data out.
about the resistive divider, you can try to fool the chip as mentioned earlier, no need to add all the diodes and power resistors for this.
Futterama
1 kW
I have tried, but I just can't get anything meaningful out. When the databytes are sent so close to each other, I can't sync them when trying to print something human readable at the same time as receiving the data.Lebowski said:
How do you use the debug output from the RS232?
Lebowski
10 MW
i plot real time running graphs using a small labview program. If you run linux you can find labview 6.1 for free on the internet, this is what I use...Futterama said:
another method is to receive the data with your terminal program and lookmat the data in hex format.... though nothing beats a running graph
Futterama
1 kW
Doh, I was trying to use another PIC to translate from 16bit data to human readable chars.
I can run Ubuntu in VMware Player and have a look at that labview thingie, thanks
I can run Ubuntu in VMware Player and have a look at that labview thingie, thanks
Futterama
1 kW
Maybe you could share your wisdom? I have labview installed on Ubuntu, but have no clue how to use it.
Lebowski
10 MW
labview is a programming language tailored for equipment control.Futterama said:
first thing, if you're using a usb-rs232 cable, is to add the cable to the list of ports.
the program to do this is visaconf, which can be started like this:
Code:
run these command lines:
export VXIPNPPATH=/usr/local/vxipnp
sudo chmod 666 /dev/ttyUSB0
sudo chmod 666 /usr/local/vxipnp
to show unknown dependencies:
sudo ldd /usr/local/vxipnp/linux/NIvisa/visaconf
to add:
sudo gedit /etc/ld.so.conf
add lines:
/usr/local/vxipnp/linux/lib
/usr/local/cvirte/lib
to process, run:
sudo ldconfig
run visaconf with:
sudo /usr/local/vxipnp/linux/NIvisa/visaconfi'll post the labview program tomorrow, it's getting late...
Lebowski said:
Awesome! Real sensorless startup from 0 rpm without going backwards by mistake (and no need for jerky alignment)! Only a few others have figured this out with signal injection
Have you played around with the frequency of the HF tone? The optimal tone freq varies from motor to motor....there is a resonant frequency where you don't need a lot of amps to measure the non-linearity (above 20kHz usally)...might want it adjustable.Interesting you need 700A to matched the strength of the fixed magnets. People don't seem to realize how amazing neodymium is. Take a look at this:
Lebowski
10 MW
Lebowski said:
Similar threads
