Frequency Multiplier problem for a guitar tuner

[steveyesterda]

hello everyone

I am a physics student who is currently writing his dissertation on designing a guitar tuner.
So far I have designed my circuit to have a microphone that inputs the sound played from the string. Then a 6 way switch connected to low-pass filters filters out the unwanted higher harmonics. I have used a crystal oscillator at 32768Hz and divide it down to give me my 1 second pulse and used a commparator to change my input string frequency to a square wave.

The problem I am having is with the +/-1 count ambiguity of a frequency counter of this style. Currently, if I were to count that signal I would only be able to determine the unknown frequency to 1Hz, not accurate enough for a guitar tuner. I need to be counting to 2 decimal places, so my plan was to multiply this frequency by 100, but this is where I become stuck.

I have tried researching using a PLL (phase-locked loop) but I don't really understand which components and values I need to get this to work. Can someone please help me. :(

Or if anyone knows of an IC that would do this job for me, that would be even better! I have looked at this frequency multiplier
http://parts.digikey.com/1/parts/1174449-ic-pll-freq-multiplier-28-tssop-sn65lvds150pwr.html
but i don't know if that will give me the desired multiplication or if it i would need to add any other components to make this work.

Thanks PhysicsForum, any help is much appreciated.

 

[revnaknuma]

1hz sensitivity is enough, I think.
commertial tuners are sensitive to about 0.5 hz.
1 hz is fair for an amateur design.

 

[steveyesterda]

hey, thanks for your reply.

Problem is, i have stated already that the human ear can tell the difference in frequency of about 5cents which on the low E string of a guitar corresponds to about 1/4Hz, an accuracy I need to be able to replicate.

Is there not a simple way to just increase the frequency x10 twice?

Thanks

 

[Sciurus]

steveyesterda, cool project.

It sounds like you're on the right track.

Of course, you could get to 1/4 Hz accuracy by triggering on the up and down portions of the square wave then sampling for 2 seconds, but I suppose you want even better than 1/4 Hz (plus faster sampling). PLL sounds like the way to go, I don't know component specifics, but there's got to be sample frequency counter (essentially what you're building) schematics out there. Best of luck, steveyesterda.

 

[MATLABdude]

Rather than use a PLL (and being completely ignorant of your design) can you just use a higher-frequency crystal for your time-base? Even the internal oscillator of most microcontrollers can be calibrated fairly well, assuming fairly consistent voltage, and temperature.

EDIT: And welcome to PhysicsForums!