|Elliott Sound Products||AN-002|
This app. note is adapted from the AC millivoltmeter described in the project pages, as well as some additional ideas. while none of the information here is original, it is offered as a potentially useful collection of different metering amplifiers. Meter amplifiers are a special variation of the precision rectifiers described in AN-001, and typically they need extended frequency response. Very high linearity is nice to have, but in reality few analogue meter movements will match the accuracy of any of the circuits shown here.
The discrete version is shown in Figure 1. This is almost identical to that shown in Project 16, with the addition of an input resistor to ground, and a higher value cap between the FET preamp and the discrete opamp formed by Q2-Q4. This circuit has the advantage of wide frequency response, and the gain is high enough to enable full scale deflection with signals as low as 3mV RMS. Making the circuit less sensitive is quite simple, and more information is given below.
Recommended supply voltage is ±15V, although the circuit works well with ±9V as shown in Project 16.
Figure 1 - Discrete Meter Amplifier
If you cannot obtain the 2N5459 JFET, you can substitute a BF244. Almost any other JFET can be used, provided the source resistor (R2) is changed to suit. Because this resistor sets the bias conditions for the JFET, you may need to experiment a little to get best performance. Ideally, the voltage on the drain should be half the voltage between the source and the positive supply. If the JFET has 2V on the source and you use a 15V supply, the optimum voltage is therefore ...
Vdrain = ((+V - Vsource) / 2) + Vsource = ((15 - 2) / 2) +2 = (13 / 2) + 2 = 8.5VAlthough it is possible to improve the circuit in terms of linearity, this is not necessary for metering applications. A small amount of signal distortion will cause a very small overall error - usually better than the accuracy of the meter movement itself.
As shown, the circuit has a -3dB frequency of 1.17Hz, and according to the simulator the upper -3dB frequency is about 1MHz. I tend to think this is rather optimistic, however it is certainly possible with careful layout.
The meter movement should ideally be either 50uA or 100uA, however it is possible to use less sensitive movements. I would not recommend anything above 100uA though, because the drive circuit has limited current capability. The maximum with the circuit as shown is capable of an absolute maximum of 300uA output, but this is just below the amplifier's clipping level.
The maximum input level is limited to around 75mV RMS, although you can increase that by removing C1 (which reduces the gain of the JFET amplifier), or if you need even higher input levels the JFET stage can be removed altogether. The absolute maximum recommended input voltage is 2V RMS - if you need it to be less sensitive, it's easy to add a simple attenuator in front of the circuit.
To obtain better performance than the standard circuit, replace D1-D4 with OA91 (or OA95, 1N60, 1N34A etc.) or similar germanium diodes, or you could use BAT43 or similar Schottky diodes. These are all faster than the 1N914/1N4148 silicon diodes, and they also have a lower forward voltage drop. It may appear that this would be of no importance, but the low voltage drop is beneficial. Any speed limitation of the amplifier circuit causes a measurable time delay as the signal goes from one polarity to the other. Lower forward voltage means there is less 'dead time', where neither diode is conducting.
The next circuit is based on one that has been used by Hewlett-Packard (which became Agilent and is now Keysight) in some of their older instruments. It has been modified to use standard E12 value resistors and more readily available transistors. One feature of the circuit is the use of R9, and it is used to partially overcome the forward voltage drop of the diodes. The circuit was designed with germanium diodes, because these are fast, and have a very low forward voltage drop. While it is possible to increase the value of R9 to enable the use of silicon diodes, this is not really recommended. Schottky diodes would (probably) be a suitable alternative, however germanium diodes (such as the OA91) are still available if you look around.
Figure 2 - Discrete Meter Amplifier (after Hewlett-Packard)
The sensitivity of the meter amp can be such as to obtain FSD (full scale deflection) with an input of 5mV RMS, and as shown the maximum is around 28mV with the sensitivity pot at maximum resistance. This can be changed, but frequency response will almost certainly suffer because of limited open loop gain and insufficient feedback.
This circuit doesn't have any particular vices, but it is completely unsuitable for DC operation because it uses only AC feedback. While circuit #1 (above) can be modified to allow DC operation, the DC stability almost certainly will not be good enough for precision work. These two discrete meter amps can be expected to perform well up to at least 100kHz, and with some tweaking can probably exceed that quite easily. Recommended power supply is 15V, although it should work well with lower (or higher) voltages if needs be. Some modifications may be required
Opamps are very convenient, but unfortunately are not always suitable as meter amps. One thing they do offer is simplicity and great flexibility, with potentially much wider input range and the ability to drive less sensitive meter movements. Very fast opamps should be able to give good frequency response, and up to 100kHz is possible with some care, or if the circuit is designed to have a relatively low sensitivity.
While it may appear that the circuit shown below cannot work properly because there is virtually no DC feedback path, it actually functions fine even without R4. The electrolytic capacitors have a very small leakage, and the circuit topology generally means that the bias point of the opamp's output will be well within limits. It will make you feel better if you include R4, although it really doesn't do a great deal.
Figure 3 - Opamp Meter Amplifier
By using low forward voltage diodes (germanium or Schottky), this circuit is capable of very good results, and will be around 0.6dB down at 50kHz. With small signal silicon diodes (e.g. 1N4148), it is almost worthless. To maintain flat response, it is necessary to keep the gain fairly low, otherwise the internal Miller cap in the opamp will cause premature roll-off of high frequencies. By using an uncompensated opamp, the necessary stability cap becomes an external component, so it can be selected to give the required bandwidth. A TL071 opamp (for example) has 13V/us slew rate, and this is fairly fast ... despite this, it is much too slow to be useful at higher frequencies. Consider the NE5534 with an external compensation cap, as it should be possible to obtain flat response to 100kHz fairly easily.
The issue with opamps in this role is simply one of slew rate - the amplifier must be able to overcome the diode voltage drop as quickly as possible. Ideally, the opamp would offer an infinite slew rate, but such opamps do not exist, so it is necessary to make do with what we have.
The gain of the opamp version is much lower than the discrete versions - about 80mV RMS input for 50uA meter current. This can be varied by changing the value of the sensitivity pot and associated series resistor, but I do not recommend using anything less than about 600 Ohms.
Any additional gain needed may be supplied with a preamplifier circuit, to lift the typical 3mV signal to 100mV. Consider using a JFET in front of any BJT input opamp if high input impedance is needed, otherwise noise will become a problem.
To get a real-life idea of performance, the opamp circuit was built on an opamp test board. This uses LM1458 dual opamps (equivalent to the uA741), and as predicted it was useless with 1N4148 silicon signal diodes. However, using Schottky diodes and set to have FSD at around 1V (using a 250uA meter that I had handy), response was 0.2dB down at about 35kHz - not a bad effort for a very slow opamp. Performance was degraded significantly if sensitivity was increased. Virtually no opamp circuit is likely to be quite as good as discrete for sensitivity, but given the low cost and great simplicity of the opamp approach it is certainly worth considering.
Instrumentation meter amplifiers are a special case of rectifier, and present the designer with a great many sometimes conflicting requirements. Because measurement instruments are expected to perform well below and above the audio frequency range, it becomes a challenge to design a circuit that has sufficient gain and wide enough bandwidth to cover the required frequencies accurately.
Sensitive meters make the design much easier, and in nearly all cases the lowest diode voltage drop possible is highly desirable. Metering amplifiers such as those shown in this article are used in a wide variety of test instruments, including AC millivoltmeters, distortion analysers, impedance meters, etc. They are my no means limited to audio usage, and are used in almost every area of electronics and engineering where analogue metering is required.
1 - Hewlett Packard instrumentation manuals (various).
|Copyright Notice.This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2004. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference while constructing the project. Commercial use is prohibited without express written authorisation from Rod Elliott.|