help in making engine digital RPM meter

I am trying to build an RPM meter for an engine.

The process I have selected is by Capacitive RPM sensing circuit inspire by

Now on each rotation of the engine there is one spark. That is I get one pulse in each rotation..

But how do I now calculate it in rotations per minutes..??

Is it I will have to find the time difference between pulses and multiply it with 60...??

Or there is some other way too..??

Hope, this article would help.

There are a couple common methods of calculating RPMs, and which is best to use depends on a variety of factors.

  1. Measure the time between pulses. Use that time to calculate RPMs.
    If the time between pulses is measured in milliseconds, then RPM can be calculated using:
    60000/time_between_pulses

  2. Count the number of pulses that occur in an arbitrary period of time. Use that value to calculate RPMs.
    If you count the number of pulses that occur in 1 second, then RPM can be calculated using:
    pulse_count*60

Both these calculations assume you have 1 pulse per revolution (as you mentioned). Below are versions of the formula that add a pulses_per_rev parameter (which, when equal to one, will simplify to the above formulas)
60000/(time_between_pulses*pulses_per_rev)
(pulse_count/pulses_per_rev)*60

If, in the second formula you are counting the number of pulses in a period of time that is not 1 second, then you need to normalize the formula to a 1 second count period.

jraskell:
There are a couple common methods of calculating RPMs, and which is best to use depends on a variety of factors.

  1. Measure the time between pulses. Use that time to calculate RPMs.
    If the time between pulses is measured in milliseconds, then RPM can be calculated using:
    60000/time_between_pulses

  2. Count the number of pulses that occur in an arbitrary period of time. Use that value to calculate RPMs.
    If you count the number of pulses that occur in 1 second, then RPM can be calculated using:
    pulse_count*60

Both these calculations assume you have 1 pulse per revolution (as you mentioned). Below are versions of the formula that add a pulses_per_rev parameter (which, when equal to one, will simplify to the above formulas)
60000/(time_between_pulses*pulses_per_rev)
(pulse_count/pulses_per_rev)*60

If, in the second formula you are counting the number of pulses in a period of time that is not 1 second, then you need to normalize the formula to a 1 second count period.

I feel its better to go with the 1st option...

because if I am going with the pulse per second and i have each pulse every 1.5 seconds then the calculation will be lil hard...

And more over my engine's idle RPM is 600...

How do I calculate the time_between_pulse..??

Joy:
How do I calculate the time_between_pulse..??

Magician's link provides some good information on how to do that.

because if I am going with the pulse per second and i have each pulse every 1.5 seconds then the calculation will be lil hard...

How do I calculate the time_between_pulse..??

Seems to me you already told us the time between pulses...

PaulS:

because if I am going with the pulse per second and i have each pulse every 1.5 seconds then the calculation will be lil hard...

How do I calculate the time_between_pulse..??

Seems to me you already told us the time between pulses...

yes i told but it is not same all the time it increases and decreases

yes i told but it is not same all the time it increases and decreases

@Joy

  • or - ? you talking about the engine ? car engine usally do not have acurate RPM, also Lawn Mower engine... :wink:

So what the range ? example : 450 to 650

If you have +/- 100, well that good right ?

Techone:

yes i told but it is not same all the time it increases and decreases

@Joy

  • or - ? you talking about the engine ? car engine usally do not have acurate RPM, also Lawn Mower engine... :wink:

So what the range ? example : 450 to 650

If you have +/- 100, well that good right ?

I will be reading the rpm of a single cylinder engine. That is of my bike..

the idle rpm of the engine is around 600

I tried this code idea from here Arduino Playground - Tachometer

//
// This example shows one way of creating an optoswitch
// using an IR LED as emiter and an IR LED receiver as
// light sensor.
// On this case it acts as a tachometer to count the
// revolutions per second of an aeromodelism plane's
// propeller.
//
//           + GROUND                                 +GROUND          
//           |                                        |  
//           <                                        < 
//           > 220 ohm resistor                       > 220 omh resistor
//           <                                        <      
//           |                                        |  
//           |                                        |
//         -----                                    -----
//          / \    >>IR LED emiter >>>>>>>>>>>>>>>>  / \   IR LED receiver
//         -----                                    -----
//           |                                        |
//           |                                        |
//           + +5VCD                                  +  ANALOG INPUT 0
//


int val;
long last=0;
int stat=LOW;
int stat2;
int contar=0;

int sens=75;  // this value indicates the limit reading between dark and light,
              // it has to be tested as it may change acording on the 
              // distance the leds are placed.
int nPalas=2; // the number of blades of the propeller

int milisegundos=500; // the time it takes each reading
void setup()
{
  Serial.begin(9600);
  pinMode(13,OUTPUT);
}

void loop()
{
  val=analogRead(0);
  if(val<sens)
    stat=LOW;
   else
    stat=HIGH;
   digitalWrite(13,stat); //as iR light is invisible for us, the led on pin 13 
                          //indicate the state of the circuit.

   if(stat2!=stat){  //counts when the state change, thats from (dark to light) or 
                     //from (light to dark), remmember that IR light is invisible for us.
     contar++;
     stat2=stat;
   }
   if(millis()-last>=milisegundos){
     double rps=((double)contar/nPalas)/2.0*1000.0/milisegundos;
     double rpm=((double)contar/nPalas)/2.0*60000.0/(milisegundos);
     Serial.print((contar/2.0));Serial.print("  RPS ");Serial.print(rps);
     Serial.print(" RPM");Serial.print(rpm);Serial.print("  VAL ");Serial.println(val);
     contar=0;
     last=millis();
   }
}

in the code if I change int nPalas=2; // the number of blades of the propeller to int nPalas=1;

as in my engine one pulse output is one rotation..

and the code checks for number of pulses in 500 miliseconds, the reading is not that accurate all the time..
it fluctuates a lot

and the code checks for number of pulses in 500 miliseconds, the reading is not that accurate all the time..
it fluctuates a lot

This goes along the lines of what I was saying in my first post. With only 1 pulse per rotation, and relatively low speed, you won't get many counts in a 500 millisecond window, which means any fluctuation in counts is going to create a magnified fluctuation in RPMs.

The easiest way to remedy that is to increase the duration in which you're counting pulses, possibly increasing it to several seconds. Unless you are using the RPM as an input to some other high speed process, like a PID, that shouldn't be a problem.

@Joy

First. Check your sensor and make sure it is "clean" <--- no bouncing

So you are using one cylinder engine <-- Just like a lawn mower :smiley:

My Theory :
Let assume the RPM is 600. 600 pulse / min / 60 second = 10 pulse / second +/- 100 rpm or 1 pulse / second ( the variation )

If using 500 ms, you should get 5 pulse / 500 ms = 10 pulse / second

In you code you have "contar" meaning count, can you modify so you can see the variale "contar" to check your "pulse", if you have 5 pulse +/- 1, well it work fine, if not, check the calculations, maybe check sensor, and maybe re-modify the code.

int val;
long last=0;
int stat=LOW;
int stat2;
int contar=0;

int sens=75;  // this value indicates the limit reading between dark and light,
              // it has to be tested as it may change acording on the 
              // distance the leds are placed.
int nPalas=2; // the number of blades of the propeller

int milisegundos=500; // the time it takes each reading
void setup()
{
  Serial.begin(9600);
  pinMode(13,OUTPUT);
}

void loop()
{
  val=analogRead(0);
  if(val<sens) // Software comparator
  {
    stat=LOW;
  }
   else
  {
    stat=HIGH;
  }
   digitalWrite(13,stat); //as iR light is invisible for us, the led on pin 13 
                          //indicate the state of the circuit.

   if(stat2!=stat)
 {  //counts when the state change, thats from (dark to light) or 
                     //from (light to dark), remmember that IR light is invisible for us.
     contar++;
     stat2=stat;
   }

if(millis()-last>=milisegundos)
{ // isolated the calculations to check the pulse count and check the pulse counting routine
/*
     double rps=((double)contar/nPalas)/2.0*1000.0/milisegundos;
     double rpm=((double)contar/nPalas)/2.0*60000.0/(milisegundos);
     Serial.print((contar/2.0));Serial.print("  RPS ");Serial.print(rps);
     Serial.print(" RPM");Serial.print(rpm);Serial.print("  VAL ");Serial.println(val);
*/
     Serial.println(contar); // to check the contar variable
     contar=0; // reset contar variable
     last=millis();
  }
}

You could feed the pulses to a frequency-to-voltage converter chip:

http://www.national.com/ds/LM/LM2907.pdf
http://parts.digikey.com/1/parts/483487-ic-converter-freq-volt-14-dip-lm2907n-nopb.html

...and then the rpm reading becomes just an analog read and scale factor & no need to worry about timing pulses. Note the "minimum component tachometer" on p. 8.

Joe

jmknapp:
You could feed the pulses to a frequency-to-voltage converter chip:

http://www.national.com/ds/LM/LM2907.pdf
http://parts.digikey.com/1/parts/483487-ic-converter-freq-volt-14-dip-lm2907n-nopb.html

...and then the rpm reading becomes just an analog read and scale factor & no need to worry about timing pulses. Note the "minimum component tachometer" on p. 8.

Joe

Thats a great thing I have come through,..

It will be great..
But can u help me out like how to do with it...??

how do I know what frequency input is what output voltage..

The datasheet is not clear to me...

Joy:
how do I know what frequency input is what output voltage..

The datasheet is not clear to me...

Check out the equation given:

Vout = fin * Vcc * R1 * C1

So just for the sake of argument say you wanted the full-scale reading (Vout = Vcc) to be at 2000 Hz (2000 rpm in your case as I understand it):

R1 * C1 = 1/2000 = .0005

On p. 7 of the data sheet there are some guidelines for choosing R1 and C1, although as you say it's not exactly clear--maybe just experiment.

Joe

I want pick the capacitive pulse with a ne555 like here http://www.sportdevices.com/rpm_readings/index.htm
and feed it to the LM2917..

now in my single cylinder bike there is one spark in every rotation..

so if the engine is running at 2000 rpm then what will be the frequency (Fin) ..??
2000 Hz or (2000/60 )Hz...??

Joy:
I want pick the capacitive pulse with a ne555 like here http://www.sportdevices.com/rpm_readings/index.htm
and feed it to the LM2917..

now in my single cylinder bike there is one spark in every rotation..

so if the engine is running at 2000 rpm then what will be the frequency (Fin) ..??
2000 Hz or (2000/60 )Hz...??

Oh, sorry--right you are, it's 33 1/3 Hz. Funny, that!

Joe

first I want to ask that should I convert the frequency to 0 - 5 volt ( taking Vout = 5 volt at 7000 Rpm) and read it with the analog input pin..??

so my 0 - 7000 RPM will be analog read rpm = map(rpm, 0, 1023, 0, 7000)

OR

There is other better way to do it for better precession and accuracy..??

There is other better way to do it for better precession and accuracy..??

None of what you suggest here will make for better precision or accuracy. Multiplying an inaccurate value won't make it more accurate.

Getting the signal each and every time, at the exact time it is sent, is how you improve accuracy. Precision is rather meaningless in this context. Either you got the pulse or you didn't. There isn't any middle ground.