Design Ideas: May 9, 1996

Figure 1,

The function of the circuit is to measure the peak voltage of the input sine wave and then reduce the amplitude of the measured peak by 0.707, so that the dc voltmeter can display the ac rms value. The circuit scales the output for 0 to 2.000V dc. This output range corresponds to an input range of 0 to 2.000V ac or 0 to 20.00V ac, depending on whether you set the scope probe to the 1X or 10X position. An input impedance of 1 M Ohm paralleled with 15 pF allows you to use a standard scope probe, which provides the familiar 10X-range multiplier feature when measuring voltages.

R₁ and C₂ terminate the scope probe into a 1-M Ohm/15-pF impedance to assure correct 1X- and 10X-range calibration. C₂ must be a temperature-stable ceramic to prevent calibration drift in the 10X range. C₁'s dc-blocking capacitor assures that the circuit measures only the ac portion of the input signal. A scope probe's high source impedance requires the use of a FET input-buffer stage (IC₁) to prevent input-bias currents from causing errors. An LF411 amplifier wired as a voltage follower performs this purpose well, because this amplifier has a good slew rate (approximately 15V/µsec) and a very low offset voltage (approximately 1 mV).

The heart of the peak detector is IC₂, which tracks the positive excursion of the input signal and charges C₄ to this value through D₂ and R₂. When IC₂'s input signal falls below the voltage across C₄, this op amp's output drops until D₁ turns on, causing IC₂ to become a voltage follower during this time period.

The action of D₁ is important, because this diode always holds the output of IC₂ one diode drop below the input voltage during the time it is below the peak value already stored on C₄. Therefore, when the input signal exceeds the voltage across C₄ (and IC₂ needs to charge it to a higher value), the output of IC₂ has to slew only a total of two diode drops to become active. D₁ is a 1N34A diode and not a 1N4148, because the 1N34A produces better performance at high frequencies. You can also use a small-signal Schottky for D₁. Note that you should use a 1N4148 for D₂, because lower off-state leakage is necessary to prevent charge loss from C₄ and resulting measuring errors at low frequencies.

For high performance, both IC₂ and IC₃ must be very fast. The LM6171A is a good choice for IC₂ because of its 3600- V/µsec slew rate and its ability to drive capacitive loads. When the input signal swings positive, the output of IC₂ rises high enough to charge C₄ to the peak value of the input. As the input swings below the peak value, the output of IC₂ follows the input voltage by one diode drop. The final amplifier stage, IC₃, is a buffer that holds charge on C₄ and provides feedback to IC₂.

As the output of IC₂ swings down, the output of IC₃ must stay locked at the voltage across C₄. To stay at this voltage, IC₃ must source the current that flows through R₃ and D₁ into the output of IC₂. To prevent slowing the frequency response of the circuit, IC₃ must also be able to react quickly to any changes in the output of IC₂. The LM310 works well for IC₃ because of its 30V/msec slew rate and input-bias current of only a few nanoamperes, which minimizes charge loss from C4. Note that this design eliminates the input offset voltage of IC₃ as an error source because IC₃ is inside a feedback loop with IC₂.

The peak value of the measured input appears at the output of IC₃. R₄, R₅, and R₆ form a voltage divider that reduces the peak value by 0.707. R₆ calibrates the output reading for the 1X range. You can calibrate the 10X range using the trimmer capacitor built into the scope probe and using an input sine wave of about 50 kHz. You must perform 10X calibration after 1X calibration to get proper performance.

C₃ is a peaking capacitor that improves accuracy at frequencies of approximately 1 MHz. The particular value of C₃ in Figure 1 produces the most accurate response between approximately 500 kHz and 1.5 MHz. R7 and R8 provide an optional dc trim to eliminate the offset voltage of IC₂.

Tests compared this circuit's performance, using the Tek P6129B scope probe, with that of two high-end voltmeters with published ac-performance specs, the HP34401A and HP3468B. The 2V-ac range tests used a 2V-ac sine-wave test signal, and the 20V-ac range tests used a 10V-ac test signal, because of the function generator's (an HP8111A) output-amplitude limitations. All measurement errors are expressed as a percentage of full scale, which is 2V ac for the first and 20V ac for the second test.

In the 2V-ac range, all meters showed measurement accuracy better than ±0.3% at frequencies as high as approximately 50 kHz. Between 40 and 100 kHz, all meters stay within an envelope of ±1%. Above 100 kHz, each meter displays distinctly different accuracy characteristics (Figure 2). Assuming a maximum acceptable error of 2% on the 2V-ac range, the respective bandwidths of the meters would be as follows: 250 kHz for the HP3468B, 900 kHz for the HP34401A, and 800 kHz for the circuit under test.

In the 10V-ac range, all meters showed accuracy better than ±0.3% at frequencies as high as approximately 100 kHz. At approximately 100 kHz, the HP3468B begins peaking, and the other two remain basically flat out to approximately 1 MHz. Assuming a maximum acceptable error of 2% on the 20V-ac range, the respective bandwidths of the meters are as follows: 250 kHz for the HP3468B, 1.2 MHz for the HP34401A, and 1.5 MHz for the circuit under test. (DI #1867)