# Frequency Selective Voltmeters

Does anyone here know of any cheap-ish handheld meters that are capable of measuring the voltage at specific audio frequencies?

I'm dealing with a few constant frequency signals on the same wire - and would love to be able to tell the voltage at specific frequencies.

The answer below is a great answer - and I guess I should have clarified slightly in asking the original question.

The frequencies are specific frequencies: i.e. 1560 Hz, 1620 Hz, 1680 Hz, 1850 Hz

I have thought of using a scopemeter with fft support, but haven't though of using a microcontroller with fft! I'll have to look into that.

I've also looked into using bandpass filters, but the frequencies are so close together that the orders would be crazy high.

• I have never heard of such a device. You want something that is more like a spectrum analyzer. There are handheld o-scopes that do that, but they are expensive and might not work at frequencies as low as audio stuff.
– user3624
Aug 2 '13 at 4:26
• How close are these frequencies to each other? Do they occur simultaneously? How many frequencies etc.. Aug 2 '13 at 9:47

There are two non-commercial product options. One is that you go with something like this:

That's a MSP430 Launchpad with a Nokia LCD, but all of the signal processing is done on the MSGEQ7, a graphic equalizer display, commonly referred as a "spectrum analyzer".

MSGEQ7- The seven band graphic equalizer IC is a CMOS chip that divides the audio spectrum into seven bands. 63Hz, 160Hz, 400Hz, 1kHz, 2.5kHz, 6.25kHz and 16kHz.

There are a bunch of different projects done with it, some have led matrixes, others have leds that vary brightness instead. Arduino too. There is a 5 band version MSGEQ5A (Frequencies between the 7 band version).

Another option is determining the frequencies yourself, through a FFT process using the analog in of any common microcontroller. Especially if is a small, known amount of fixed frequencies. WAY cheaper than any commercial spectrum analyzer, but it does depend on the frequencies themselves. Your microcontroller (and code) needs to be able to sample the signal at atleast twice as fast as the desired frequency (A mcu running at 1mhz will not accurately sample a 1mhz signal, even before you get into code slowing that down)

Edit: You can even go old school discrete analog solution, by using bandpass filters to isolate the frequencies you want to measure, and measuring them individually.

• +1 for excellent direction provided... I'd enhance the answer with a link to the TI Stellaris open source FFT project I've seen around. Aug 2 '13 at 6:20
• @AnindoGhosh I didn't add a link to any specific project, because fft can be done on any brand microcontroller. One guy even did fft on the ez430 usb stick (2kb flash 128b ram) for a bass/mid/treble led display, no arm processor needed. The one in the picture is msgeq7 based, so all the fft/signal analyzing is offloaded to that, the msp430 just makes the data a pretty picture. Aug 2 '13 at 6:34
• Fair enough. I just mentioned that specific project because it is open source, inexpensively made, very configurable, with fine control on the FFT buckets, and overall neat. :-) Aug 2 '13 at 6:51

THE FREQEUNCIES ARE 1560 Hz, 1620 Hz, 1680 Hz, 1850 Hz THE DIFFERENEC IS ONLY 60 Hz and will require FFt with bandwidth of 20 hz or so to be able to distinguish these. One method is to multiply the signal with say 1620 hz and average it with a low pass filter . This is the same principle used in lock in amplifiers. You can easily get a tuned amplifier with effective bandwidth of 20 hz, 2 hz or even .2 hz, depending on the time contsnat of the averager used. you might also have to do a quadrature detection in which the 90 deg phase shifted form of the signal is multiplied with 1620 hz that you have generated, with 90 degree phase shift The product has to be averaged and added in quadrature to the in phase product. Quadrature sum of a and b is sqrt (a^2+b^2). the ethod of graphic anaylsers will have too coarse a spread of freqeuncies and you will nopt be able to distingusih presence of 1620 as against presence of 1560 or 1680 hz.