Efficient and fast math?

[astus]

Hi,
Currently I'm trying to measure the RPM of a turbine. I am setup to use the capture mode on the pic16f877. I have the period of the turbine rotation but how do I determine the RPM so I can display it on an LCD display? The math is

RPM = 60 / ( CCP_Value * Ts) where Ts is timer period.

Do I have to convert to floating point or is there a sneaky method that I'm not aware of ?

Thanks

Adam

[kypec]

Maybe we could give you some suggestions but you have
to let us know what values of RPM you expect/need to measure.
What is your supposed range? 100 - 1000 rpm?
Or is it 1000-10000 rpm?

I did a sort of DC motor regulation with feedback of optical tacho
encoder. The basic I needed to know was how many pulses
I am able to capture at minimum speed and at maximum speed.
Given that I could determine my measurement period and
eventually all necessary calculations as well.

kypec

[Ttelmah]

[SherpaDoug]

Most of Ttelmah's ideas are good. I would definately scale things so I could use fixed point arithmatic. But he lost me at
[quote="Ttelmah]
60000000.0/(float)CCP_value
Next, the 'second comment' then applies, and you could instead use:
1.666667E-8 * (float)CCP_value.
[/quote]
(maybe it is just Monday morning). But the important thing is to make sure YOU understand the code you are using.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Not yet registered]

When I learned at uni we looked at the amount of processes involved in Math functions, a division takes somewhere aound two - four times as many clock cycles to complete, so a multiplication is more efficient,

if the turbine is high speed these few extra milli seconds might make all the difference.

I'm not sure of the exact capabilities of the 16f877 but we did something simillar for a uni project on a motorolla 16hc11.

as a low speed we measured Timer over flow flags an made a heart rate monitor, (rate is circa 60 - 150 BPM anything over or under is measured for a while and critical status can be reported, but in extreemes no measurement takes place)
as a high speed application we made a fuel injection system taking throttle positions on an ADC and measuring engine speed to determine a duty cycle, in this case there were two possible ways to complete the task.
The first was to use the Timer output capture to make an interrrput driven system, in this way when a pulse comes from the crank shaft a timer is stopped and a speed can be measured, this can only measure high speeds.

a second method is to count both clock over flows and clock ticks, in this way both fast and slow speeds can be measured (0 - ~2Mhz).

depending on the speed of theo turbine will depend on the speeds that the code needs to excecute.

(i'm thinking of registering, I've been posting as a guest for a week now, and everyone seems quite nice and helpfull here abouts!)

[Not yet registered]

that should have said motorolla 68HC11.
(getting confused with the Microchip PICS I'm using now)
Definitly must register so I can at least edit my own posts!

[Ttelmah]

[astus]

Thanks for the helpful responses

I was actually using the first method suggested by Ttelmah, it takes about 1.3ms (depending on values) at a clock spped of 4Mhz (1us clock). I haven't tried the double reciprocal thing I will test it and get back to you on that one. The integer method seems interresting but I'd have to change my clock freq or add a new oscillator in to drive the timer 1 module (Possibility in next revision).

Oh, Kypec the turbine runs from ~1000 to 150,000 RPM (Fast) currently I use the ccp and timer 1 module. After the turbine reaches idle (~35000 RPM) I then capture every 16th edge from the shaft. To keep the resolution high I only use the RPM function on the LCD screen. The turbine control algorithm uses the captured value from the ccp.

If anybody was wondering I'm working on a turbine controller for a small jet engine used in the hobby model aircraft area.

Thanks Again

Adam