Frequency Detector Circuit

Hi there,

This is my first time on this forum and i have searched for my question but couldnt find exactly what i need

I would like to try and build a Frequency Detector circuit using an LM3915. Is this possible?

I.E i would like more LEDs to light up as the frequency gets higher. If it is possible i would preferably like it to be as small a circuit as possible so its easy to make, as i am quite new to Electronics.

Thanks

LM2907/LM2917 frequency-to-voltage converters: **broken link removed**

Ken

You could just use a diode charge pump (2 cap + 2 diodes). It's simpler/cheaper albeit less accurate.

If the input signal has a 50-50 duty cycle, the output of the charge pump will always be the same, regardless of frequency...within limits.

In order to have an output voltage proportional to input frequency you would have to change the input signal to a fixed width pulse with a monostable. That makes the monostable's output PWM proportional to input frequency. This followed by an RC low pass filter. Think....analog tachometer.

Ken

Sorry, I should have said: ... diode charge pump (2 caps + 2 diodes + 1 resistor) ...

The resistor is on the output cap to ground.

Still won't make much difference. The resistor/load will only lower the "averaged" output voltage. With a 50% duty cycle input...sine, square, triangle, whatever...the time your charge pump diodes conduct will be exactly the same...50%...regardless of the frequency. So the "averaged" output will always be the same. This is all based on the fact that charge pump caps are large, relative to the frequency range. But, we don't know what the OP's signal is....frequency, voltage, wave shape...

ken

KMoffett said:

But, we don't know what the OP's signal is....frequency, voltage, wave shape...

Click to expand...

That is true. I just checked an he wants a frequency detector (not a freq counter, which is what I read/assumed).

KMoffett said:

Still won't make much difference. The resistor/load will only lower the "averaged" output voltage. With a 50% duty cycle input...sine, square, triangle, whatever...the time your charge pump diodes conduct will be exactly the same...50%...regardless of the frequency. So the "averaged" output will always be the same. This is all based on the fact that charge pump caps are large, relative to the frequency range.

Click to expand...

The caps would have to be huge as the diode series resistance is going to be very small. As that's not the point, the charge pump is created using practical capacitor values, and the output voltage relates to the frequency. The duty cycle has little effect; the series resistance of the diodes is small and the caps charge and discharge very quickly. The charge is transferred near the edge of the input pulse.

If you are only talking peak voltage output I would agree. But, since most (and I will accept being beaten up over this) AC measuring systems deal with "average" output over time:

the series resistance of the diodes is small

Click to expand...

True

and the caps charge and discharge very quickly.

Click to expand...

True and false. The caps charge quickly, but the output cap discharges slowly. Otherwise a charge pump wouldn't work. The discharge rate depends on the RC time constant of the output cap and the load resistance.

The charge is transferred near the edge of the input pulse.

Click to expand...

. True and false. The charging current is available throughout the time that voltage, forward biasing a diode, is applied to the input. Once the voltage level on the input no longer forward biases the diode, the output cap discharges at an RC rate.

The duty cycle has little effect;

Click to expand...

False. If I apply a short pulse (1% PWM) to the the pump, the cap is charged to peak, but has a long time to discharge before the next pulse. This would result in a very low "average" output voltage. If I have a 50% duty cycle, the cap is charged to "peak", but has less time to discharge, so the "average" voltage is higher. If I have a 99% duty cycle, the cap is charged to "peak", but almost no time to discharge. So the "average" voltage is essentially equal to the peak voltage.

And by you earlier logic the higher the frequency of the input, within limits, would result in shorter discharge times and therefore higher "average" voltage.

So...The devil's in the details. But, we probably have the OP scratching both ends of their anatomy by now.

Ken

Sorry, the transfer cap charges and discharges quickly.

Attached cct shows 1% 50% 99% 100Hz charge pump outputs. A 50% 1kHz is there also. The levels are quite similar, even with the varying duty cycle.

QUOTE]Sorry, the transfer cap charges and discharges quickly[/QUOTE] Agree..I was referring to the output cap where you actually take an output.

As I said the devil's in the details. With an output RC time constant of 100mS for 100Hz, I wouldn't expect to see much difference. Try it with an output RC combination with a time constant of something between 1 and 10 mS...10K to 100K load. Just curious.

Ken

KMoffett said:

QUOTE]As I said the devil's in the details. With an output RC time constant of 100mS for 100Hz, I wouldn't expect to see much difference. Try it with an output RC combination with a time constant of something between 1 and 10 mS...10K to 100K load. Just curious.

Click to expand...

Well the devil's a pedantic little ....
As for reducing the time constant, obviously the output ripple increases (greatly). You could low-pass filter that signal I guess to get something more usable.

Yes, and that's that concept to make a tachometer with a fixed pulse width, variable frequency signal.

Ken

I've long been interested in sensing and ascertaining the frequency of a signal, and I've designed a novel circuit that accomplishes this. It doesn’t use any charge pump or time constant network, and so its response is very fast. I hesitate to post it because it's not a finished product yet. I only have a simulation using generic components. It is in need of refinement and redesign of the input circuit. If anyone wants to take a look and possible work or prototype it, I'll post the details.

BrownOut said:

I've long been interested in sensing and ascertaining the frequency of a signal, and I've designed a novel circuit that accomplishes this. It doesn’t use any charge pump or time constant network, and so its response is very fast. I hesitate to post it because it's not a finished product yet. I only have a simulation using generic components. It is in need of refinement and redesign of the input circuit. If anyone wants to take a look and possible work or prototype it, I'll post the details.

Click to expand...

hi D,
I would be most interested to see the details.

Me too.

Ken

Thanks for your interest, gentlemen. I"ve posted this as a new thread.