Dick Cappels' project pages http://www.projects.cappels.org
return to HOME
Precision Audio Frequency Peak Detecting Probe
This is a handy companion for a digital voltmeter. Its allowed me to do a lot of things I used to use my oscilloscope for, and in addition it measures voltages to much greater precision.Using an LM324 quad op amp, this peak detector provides 2% measurement accuracy on square waves from 5 Hz to 20 kHz.
Circuit description: Taking the circuit from left-to-right, the first op amp (lower-left corner, inputs pins 8 and 9) is a power supply splitter. It splits the 9 volt battery to make plus and minus 4.5 volts, providing a virtual ground at its output. At the upper-left a voltage divider allows the circuit to operate with up to a 15 volt full-scale input. The 10k resistor in series with the switch following the voltage divider limits current in case the input voltage exceeds the power supply current. The two diodes after the 10k resistor limit the input voltage at the input (pin 12) of the op amp. This op amp is a unity gain buffer and serves no immediate use, though it could be connected as a gain stage to increase the peak detector's sensitivity, at the cost of bandwidth.
The next op amp (pins 1,2, and 3) drives two diode peak detectors. The dpdt switch selects which of the two peak detectors is buffered by the last op amp, thus closing the loop around the output voltage for either the positive peak or the negative peak. (Note that in the schematic there appear to be two SPDT switches, but they are really ganged as DPDT.) The other section of the dpdt switch connects a diode from the output of the op amp to the input to prevent the op amp from going open loop -doing so would cause an accuracy problem at high frequencies because the op amp takes some time to come out of saturation and close the loop. The tricky part of the circuit is the two-stage peak detector.
So...what's special about this peak detector? This peak detector's quick response time while maintaining a low droop rate, which is necessary to maintain accuracy at low frequencies is owed to the presence of the dual time constant peak detector. The 100 uf capacitors in series, which equals 50 uf and 100 K resistor provide a 5 second time constant peak detector. This 5 second peak detector is coupled through a diode to the 22 uf capacitor pair, that act as an 11 uf capacitor. Normally, the 100 uf capcitor pair is kept charged at one diode drop above the signal's peak voltage, which is the same voltage as the 22 uf capacitors are chaged to. The time constant on the 22 uf capcitor is very long, primarily determined by the capacitors and the input leakage on the op amp. This long time constant is what makes the high accuracy possible at low frequencies. The voltage on the 100 uf capacitors droops beween pulses, but not until the frequency drops below approximately 5 Hz does the ripple get to be large enough to "reach through" the series diode between the 100 uf pair and the 22 uf pair, and affect the voltage on the 22 uf pair.
Another ugly but functional circuit.
Layout is not critical. Its a good idea to keep the capacitance low on the output of the voltage divider so that high frequencies aren't rolled off. The diodes are small signal types like 1N194. This was built on a piece of punched fiberglass board. A printed circuit board is too much trouble and not worth it for the frequencies and signal levels involved. Different types of op amps will give different results. The LM324 allows this to run the battery down to 7.5 volts. Newer op amps will allow higher frequency operation.
It might be a good idea to add an LED as a pilot light so that you don't leave the switch on and run down the battery. A higher voltage power supply and a faster quad op amp would allow this thing to run at higher frequencies before accuracy degraded.
Please see the liability disclaimer on the HOME page.
Contents ©2002, 2013 Richard Cappels All Rights Reserved. http://projects.cappels.org/
Corrections July 2013 with thanks to Sergey in Russia.