CCP - Problem with capture mode

[hemnath]

I'm trying to measure and display the frequency value in LCD. I'm using PIC18F2520. Internal oscillator 4Mhz. Frequency from function generator. I have connected output of function generator to pin C2(CCP1) of uC. Using capture mode, I'm trying to capture the signal and displaying. But in the display it shows fixed value and value is not changing in the LCD when i change the frequency in function generator.

Please help.

[PCM programmer]

Regarding the signal on Pin C2:

Look at the signal with an oscilloscope and post the answers to
these questions:

1. Is the signal a squarewave or a rectangular waveform ?
Or, is it a sinewave or some other shape ?

2. What is the frequency of the input signal ?

3. What are the low and high level voltages of the input signal ?
For example, does the signal go from 0v to 3.3v, or 0 to 5v ?

[hemnath]

Sorry. I noticed now only, that output level from function generator was about 1V. I increased to +4.5V. Now the code works good. Thanks for the valuable information. But now I faced another problem, when i give input frequency from function generator say about 5Khz. LCD display value keeps changing. So i thought to use PLL.
I declared in the fuses line as H4 and used external crystal 4Mhz.

Now the LCD fluctuates from 5Khz to 5.025Khz.
Can i control with any methods?

code changes

[Ttelmah]

You need to start by looking at your input signal.

Now, unless this is a TTL source, you have a waveform (possibly programmable in shape), which will almost certainly be going +ve and -ve of it's ground. You are then feeding this into a logic input, with clamp diodes, which stop the -ve excursions going more than about 0.6v below ground, and the remainder of the waveform then goes +ve of ground. The logic input detects 'high' at a particular voltage about ground. So unless everything is very stable indeed, there will be variation with this point, and the signal coming in.

To feed an AC signal like this into the processor, you really need to probably clamp the incoming to half the supply rail, and process it with something like a zero crossing detector and comparator to give a a nice logic square wave to feed the PIC.

You need to get this done first.

Then look at smoothing the count result.

Best Wishes

[Mike Walne]

OK.

Have a play at being computer.

When your input frequency is 5kHz, the period is 200us.
With your original 4MHz clock that's 200 counts.

Suppose the real period was 199.5us:-

1) Your count would be 200 one time. Displays 1e6/200 -> 5.000kHz
2) Your count would be 199 next time. Displays 1e6/199 -> 5.025kHz

Your display is doing exactly as what you're telling it.

Your method for calculating frequency is great for low frequencies, & falls over for higher ones.

So, you need to either:-

a) Modify your present method to deal with higher frequencies
OR
b) Use a different approach for higher frequencies.

BUT first, do as Ttelmah suggests and clean up your input signal.

Mike

[hemnath]

Thanks Ttelmah for the information. Now i'm giving input to PIN C2 from Pulse generating circuit which will be clean 5V and 0V. There will no negative pulse. I did everything whatever you said.

But still the problem remains the same. Do i have to change anything in the program?

I have one doubt. Please clarify me. If i use PLL, my program instructions will be executing fast right?
say #fuses H4 . and crystal 4Mhz. So the frequency could be 16Mhz? is it right?

Also i have did in my program

[Mike Walne]

[hemnath]

Sorry for the late reply.

I have measured the frequency.
With crystal 4Mhz and for #fuses XT and 5Khz from function generator.

LCD display 5000 Hz and sometimes 5025 Hz. When i increase the frequency little higher, say about 5002 Hz, LCD still displays 5000 Hz and 5025 Hz. It seems that LCD counts every 25 Hz as stated in below post

[Mike Walne]

I thought I'd explained in my previous post.

You're trying to measure ~5000Hz frequency to one part in 5000.
That's 0.02%.
You're doing it by measuring period to 1 part in 200.
That's 0.5%.

The LCD is displaying what you're telling it to do!

So like I said before either:-

a) Modify your present method to deal with higher frequencies
OR
b) Use a different approach for higher frequencies.

You can do (a) by measuring several periods, then take average.
You do (b) by measuring frequency (i.e. No of cycles in 1s) directly.

It's been done loads of times here on the forum.
Do a search.
I know I've covered it myself at least once.

Mike

[hemnath]

How can i modify or to use different approach?

I'm totally clueless. Can you please change my above code to error of 1HZ? So that i ll learn from it.

Do i have to change the crystal frequency to minimize the error?

Can you please direct me to those links?

Thank you .

[Mike Walne]

[hemnath]

Thank you. I ll look into those links.

[Mike Walne]

To calculate for 20MHz crystal you simply change these lines in your code

[hemnath]

Thank you very much.

It took some time for me to understand why you have mentioned 100Mhz crystal.

If possible, Can you please explain with small example program with 0.0001% with my existing 4Mhz external crystal?

Everything is good when i work out in theoritical. But practically, it's not attaining for me.

Please bear with my ignorance. Sorry .

[Mike Walne]

Before we go any further please tell us:-

1) Total range over which you are trying to measure frequency.
2) Required accuracy over each range.
3) Purpose of the exercise.

What I'm trying to avoid is, showing you how to do one thing, then you coming back and saying "Oh! But I want to do this elsewhere".
i.e. moving the goal posts.

Mike