Re: pump control
- From: "Fred Marshall" <fmarshallx@xxxxxxxxxxxxxxxxxxxx>
- Date: Fri, 30 May 2008 23:52:55 -0700
"Tim Wescott" <tim@xxxxxxxxxxxxxxxx> wrote in message
news:1NOdnTSUA5ldfKPVnZ2dnUVZ_g2dnZ2d@xxxxxxxxxxxxxxx
sorenbirk wrote:
This is actualy a question regarding control theory...
I am controlling a motor by sending a reference between 0 and 1, where
one
corresponds to full speed and 0 to stopped, i.e., the motor only turns in
one direction. It is an analog value, so any value between 0 and 1
corresponds to a motor speed. The motor drives a pump, which produces a
pressure. My application is to keep the pressure constant, by controlling
the motor speed. The pressure is measured in N/m^2, (e.g. 100000 N/m^2)
which gives rather large error signals. My problem is how to scale my
signals, as the control signal must be between 0 and 1? How is this
commonly done?
One way or another you have to scale things to get them into the
processor. Your 0-1 command to the motor probably becomes a voltage to
the motor, with some scaling from 1 to the motor's maximum voltage (e.g.
the motor sees 0V to 12V).
If I'm working with a fixed-point machine I'll scale things so that the
full scale analog value on the output or input corresponds to full scale
of the data type that I'm using. Since I prefer to use fractional
arithmetic, I'd scale the pressure read back so that maximum pressure is
1. Normally I'd scale things so that 0Pa would read as 0, unless you
needed tight control over a narrow band.
If you're doing your computation in floating point arithmetic, you may
wish to just read back in Pascals (1Pa = 1N/m^2), and use small gains in
your controller. That way when you're debugging you can just compare the
actual pressure that you read by other methods with the number being read
back, to verify that all is correct. Using scaling like this can get
exceedingly weird to deal with with fractional math, but it doesn't
introduce any extra peculiarities when you're using floating point.
Tim has done more of this than I and I agree with all he's said. You do
need to drive the motor with full voltage to get the best result.
Then, the controller you describe is essentially a bang-bang or off-bang
type of controller that uses the motor as an integrator. The effective
voltage to the motor is the integral of 0's and 1's, multiplied by the
maximum motor voltage available. Said another way, the voltage is
proportional to the duty cycle.
You may want to actually have a duty cycle circuit that is driven by the
pressure error. So, instead of turning the motor completely off for longer
periods of time or driving it toward full speed for longer periods of time
you would adjust the duty cycle of 0's and 1's according to the error. What
you really want is for the motor to run at a constant speed that is the
speed needed to hold the pressure constant or as demanded. There will be a
duty cycle that will do that.
Just make sure to do a reasonable control system analysis for stability and
transient response to get what you really want.
Fred
.
- References:
- pump control
- From: sorenbirk
- Re: pump control
- From: Tim Wescott
- pump control
- Prev by Date: Re: Battery powered DSP boards
- Next by Date: Re: Spectral Analysis & Resynthesis of Audio Signals.. Anyone else into this? Applications/Tools that reliably produce good results?
- Previous by thread: Re: pump control
- Next by thread: ti's EZ-DSP STARTER KIT FOR F2812 -> just the board
- Index(es):
Relevant Pages
|
