Design a wideband analog multimeter to measure AC voltages and currents from 20 Hz to 1 MHz with a selectable sensitivity from 1 mV to 100 V, while the ammeter has a selectable current sensitivity of 10 mA to 10 A.
It may be surprising to offer a purely analog meter when digital meters are now so widely used. However, it isn’t well-known that digital meters have analog circuits preceding the analog-to-digital converter (ADC). Wideband analog voltmeters, with frequencies from below 20 Hz up to 200 kHz, have been widely used in the past and are still very useful.
Key specifications for a wideband analog meter
With modern operational amplifiers (op amps), the design of our analog meter can be greatly simplified. This design provides a flat bandwidth of 20 Hz to 1 MHz (–1 dB at 10 Hz) using reasonably-priced components.
It is not difficult to add the capability of measuring currents in circuits in which the insertion of a resistance of 0.1 Ω at the grounded end does not reduce the current significantly. If the meter is battery-operated or of safety Class 2 construction, the grounded connection does not have to be an actual ground.
The six voltage ranges are 1, 10, 100 mV, and 1, 10,100 V full scale, while the four current ranges are 10, 100 mA, and 1, 10 A full scale. The design incorporates a wide-band peak detector, whose sensitivity can be switched to read the peak voltage or the root mean square (RMS) voltage of a sine-wave signal.
There is a provision to insert external filters to provide special frequency responses. This feature also allows the meter to be used as two separate amplifiers with switched variable gain.
The instrument will run from two 9 V batteries or a 9-0-9 V mains power unit. The current drain is below 25 mA for each battery (without any light-emitting diode (LED) indicators), so long life can be predicted.
Analog meter project – general description
The block diagram of the instrument is shown in Figure 1.
 |
|
| Figure 1. | Block diagram of the wideband analog meter. |
The input attenuator/current mode selector is necessary since we want to measure voltages up to 100 V, and they can’t be applied directly to the amplifier. The current mode is selected so that there is no switch contact in series with the current-sensing resistor, thus eliminating a potential source of error. The small price to pay is that a separate connector is required for current-mode operation.
Using two op amps to reach gain
The amplifier requires a gain of around 1000 (60 dB) to give a full-scale reading for 1 mV input. This is achieved by using the two op amps in the LM4562, with gains of 31.62 each. The LM4562 is actually intended for hi-fi preamps, so it has low noise and low distortion, together with a wide bandwidth. It isn’t so hot on offset voltage, but that can be overcome in this application.
Since there are two amplifier stages, it is easy to provide connectors and switching so that other circuits can be inserted between the two amplifiers, and they can also be used as independent amplifiers if a further connector is added at the output of the second amplifier.
Choosing your detector: half-wave, "true-rms," and full-wave peak
An important decision has to be made concerning the type of detector provided to turn the amplified signal into DC for operating either a digital display or a pointer instrument. The simplest detector is a half-wave average type, which tells us little about the signal and may hide a significant voltage excursion in the undetected half-cycle.
At the other end of the complexity scale is the "true-RMS" detector, labeled ‘true’ to distinguish it from an average detector whose gain has been tweaked to read the RMS value of a wave signal but not of all others. This is good if you want to know the RMS value of the signal, but the only reasonably-priced device is the AD736, which is limited to 200 kHz, whereas the amplifier works up to 1 MHz and beyond.
The third type of detector is the full-wave peak detector, and this can be constructed at a reasonable price. It is very often that the peak value of a signal needs to be measured, as it may indicate that something in the signal chain is overloaded and peak-clipped. It is also easy to find the RMS value of a sine-wave signal (even with up to about 10% distortion) by dividing the peak value by 1.4 or multiplying by 0.7, as long as the waveform isn’t significantly clipped. This is easily arranged because it requires a 3 dB attenuator to be switched in.
Another simple attenuator interpolates between the 20 dB (10 times) steps of the range switch, which allows all indications above 0.5 mV on a pointer instrument to be in the upper half of the scale. Both attenuators can be applied.
Searching for published full-wave peak detectors that work up to 1 MHz proved unfruitful, but by merging techniques from two unsuitable detectors, a solution has been found, using another LM4562 and two RF bipolar transistors BF140. Other transistors with similar or better characteristics can, of course, be used, but general-purpose transistors such as BC547/847 are not suitable.
Project's circuit design aspects
For this experiment, the blocks are described individually, with their schematics (except for the amplifiers), which use a dual op amp and a 2-pole switch, thus, separating them would be confusing. The whole schematic is too big to be legible if shown on a single page. The power supply connections at the op amps are shown only as V+ and V– and in the power supply block to avoid too many long wires.
The input attenuator and current-mode switch
The schematic for the input attenuator and current-mode switch is shown in Figure 2.
 |
|
| Figure 2. | Input attenuator and current-mode switch. |
The switch is shown in the highest sensitivity position. Beginning at the lower position and moving clockwise, the switch steps are current mode (10 mA), 100 V, 10 V, and 1 V.
The current-sensor resistor (and all resistors used in the project unless otherwise stated) should have ±1% tolerance, so it is not a cheap component but affordable. An alternative is to use a 0.15 Ω resistor with another value in parallel, chosen to give close to 0.1 Ω. It should not be wirewound unless metal-cased because otherwise, the inductance might introduce error at high frequencies: 100 nH is 628 mΩ at 1 MHz.
Note that the current sense resistor is connected as closely as possible across the input connector so as not to introduce extra resistance. There is also no DC blocking capacitor in series for the same reason. If a component with a power rating of 1 W or more is used, any DC component less than 10 A should not cause a problem.
C2 should be a polyester film capacitor of 250 V rating or higher if you use higher DC voltages.
The trimmer capacitor is there to adjust the attenuations to be correct at high frequencies. These values worked for me, but the capacitances are very construction-dependent, so you may need different values or configurations.