Elliott Sound Products Multimeters

# Multimeters, Analogue Vs. Digital, And How To Use Them

Main IndexArticles Index
Contents
Introduction

One of the first pieces of test equipment bought by anyone interested in electronics is a multimeter.  These are also referred to as a 'VOM' - volt, ohm meter (or volt, ohm & milliamp meter).  The range is bewildering to the newcomer, but at least 99% of buyers will choose a digital meter.  One of the reasons is that they are far more readily available than analogue (with a moving-coil meter movement), and usually a great deal cheaper.  Analogue meters also require a bit more skill to operate, as you have to select the scale appropriate for the selected range.  You also need to minimise parallax error - looking at the pointer and scale from an angle other than 90°.  Better moving-coil meters have a mirrored scale, and the reading is within the rated accuracy when the pointer and its reflection are superimposed.  In the days when analogue meters were the only choice, there were many people who could never get to grips with the ranges, scales and multiplication factors needed to obtain a meaningful result.

An auto-ranging digital meter only requires that you select a range (e.g. DC volts, Ohms, etc.) and the readout shows you the value in the correct units.  For most measurements you don't need to think about the units, as the display shows the measured value and its units (AC volts, DC volts, etc.).  There is an 'implied' accuracy that's often misleading, because the reading may show four or more digits, and users almost always assume that the value displayed is exact.  In reality, there's a stated accuracy (typically 1% for low cost meters), but the last digit may be ±2 digits off the true value.  When you see the accuracy described as 1% ±2 digits, that means that a voltage of (say) 10.00V could be displayed as anything from 9.88 to 10.12 volts.  However, users (and that means all users, even those who know better) tend to take the displayed value as 'gospel'.

In a well equipped workshop, it's very handy to have both types of meter available.  Analogue meters are far better at displaying a changing voltage, current or resistance.  Even fairly large cyclic changes are easily averaged by eye.  You simply look at the pointer and find the geometric mean of the maximum and minimum pointer swings to obtain the average.  A digital meter will usually just show a reading that changes (at the sampling rate), and it's usually impossible to determine a true average.  It's become common for (LCD) digital meters to have a 'bargraph' along with the digital display that supposedly gives you the best of both worlds, but I find them next to useless when taking measurements.  Note that some digital meters will show the true average of a cyclic voltage, but not all.  Some will display garbage - no use to anyone.

Despite the apparent simplicity of multimeters, many people still have difficulties interpreting the results.  This is true of both analogue and digital meters.  Analogue types are always harder to interpret, but digital meters have a very high input impedance, and it can sometimes appear that an AC voltage is present, even though the available current may only be a few microamps.  This can lead to great frustration, with the user wondering where the voltage is coming from, when it may only be due to capacitive coupling between insulated conductors.

If one has been caught out by this a few times, it can become dangerous.  The user assumes that the measurement is false (having been caught out before), then discovers that it was very real!  'Standard' analogue meters are better in this respect, because they have a relatively low impedance.  This will load very high impedance 'leakage' paths, but show the true voltage if it's present.  Great accuracy isn't important, but getting a definite answer (safe/ unsafe) is important.

 WARNING:  Always take great care when measuring high voltages (AC or DC).  Use only probes and test leads that are rated for the voltage being measured, and do not attempt to measure any voltage that is (or is suspected to be) greater than the meter's maximum voltage rating.  The information here is provided in good faith, but does not (and can not) cover every eventuality.  Safe work practices are the reader's responsibility, and must be applied at all times.  If unsure, always seek professional assistance before risking your life!  Never use test leads that show signs of abrasion, damage, or that have been modified or mistreated.

You always need to know what to expect before taking a measurement, otherwise you'll never know if the measured value is alright or not.  Your guess doesn't have to be particularly accurate, but it should be based on the component values in the circuit, or (in very few cases these days) the voltages may even be shown on the circuit diagram.  In the 'old' days, this was common, and in many cases they even described the type of meter used to take the measurement!  Alas, this is no more.  In most cases it's expected that a digital meter will be used, but you're usually left to work out what voltage(s) should exist in a circuit.  In some cases (such as power amplifiers and other circuits that use DC feedback), the voltages will only ever be 'sensible' when everything is working normally, when measurements aren't really needed.

It's worth pointing out that the death of the analogue multimeter has been greatly exaggerated.  There are countless new models still available, with a wide price range.  Many vintage instruments are popular with collectors and restorers, but the usefulness of the analogue movement is such that there's still a strong demand.  I was actually surprised at the number of new models I found while researching for this article - there are far more than I ever imagined.  They don't have the accuracy of a digital meter, but mostly you don't need it anyway, and IMO no digital meter has the 'charm' of a good analogue multimeter.  A well-equipped workshop will have both.

1   Analogue Meters

An analogue multimeter uses a moving-coil movement, almost always of the D'Arsonval type (see Meters, Multipliers & Shunts for the details).  The input impedance is not easy to understand at first, because it's usually quoted as kΩ/V.  This figure depends on the range selected, not the displayed value.  Most decent analogue meters are 20k/V, meaning that on the 1V range, the meter impedance is 20k, so the movement has a sensitivity of 50µA full scale.  Very cheap meters can be as low as 1k/V, meaning that the movement is 1mA full scale.  These are not recommended, because a) they are cheap (in all respects) and b) because some circuitry will be 'upset' by the current drawn by the meter.  The more you're willing to pay, the more sensitive the meter movement, and the most sensitive I've heard of was made by Sanwa, and was 2µA full scale (500kΩ/V).  You also need to be aware of the accuracy, generally 3% for DC and somewhat worse for AC.  Scale linearity depends on the quality of the movement, and while it may be adjustable (internally) I don't recommend that you attempt it.

A properly balanced movement will show zero on the scale regardless of the angle of the meter (vertical, horizontal, 45°, etc.).  Unless you pay serious money, don't expect this to be the case.  Setting up a moving-coil meter movement to be unaffected by angle (balancing the moving parts) is a painstaking process, and unless you're an instrument technician I suggest that you don't fiddle with it.  It's far easier to make it worse than better.

To make sense of the kΩ/V specification, you look at each range, which may be 0.5V, 2.5V, 10V, 50V, 250V and 1kV full scale.  For each range, you multiply the range by the kΩ rating, not the reading.  The impedance for the 0.5V range is therefore 10k, 50k for the 2.5V range, 200k for the 10V range and so on.  This means that if you're measuring the voltage of a high-impedance circuit, the reading will change as you change ranges.  This is disconcerting for beginners in particular, as there appears to be an inconsistency in the readings.  The meter (and its reading) is operating as dictated by Ohm's law, and it's not inconsistent at all, but it does cause problems.

If a 20k/V meter is set to the 10V range (for example), the impedance is 200k.  Most circuitry won't care about the load, but it makes a significant difference if you're measuring high impedance circuits, such as the plate voltage in a valve (vacuum tube) preamp or phase splitter circuit.  For that you'd typically use the 250V range, which has an impedance of 5MΩ.  While that sounds fine, if the plate resistor is (say) 220k, the meter will cause an error.  You might expect to read 125V DC, but the meter's loading will reduce that to 122.3V.  This may be within the accuracy specification, typically around ±3% for 'mid-range' meters.  It isn't a limitation for users who understand that the vast majority of measurements don't need to be exact, but it can still be a nuisance.

There are two brands of analogue multimeters that are revered - In Britain, Australia, New Zealand (etc.) the Avometer (introduced in 1923) was the 'gold standard', and early models are sought after.  Possibly one of the most distinctive meters of all time, the dial surround was in the same shape as the dial scale (you'll have to look up a photo, but they are easy to find).  They were expensive meters, but they were one of the most common brands found in well-equipped workshops.  AVO held worldwide patents for many of the techniques used, and almost every other meter that followed was based on the original design.  To this day, no other readily available multimeter uses a current transformer for AC current ranges, along with a full-wave (bridge) rectifier.

Although it came much later (some time during the 1930s), in the US and Canada, Simpson holds a similar position.  They were (are) more conventional, but had a well deserved reputation for reliability and performance.  From the Japanese makers, Sanwa and Micronta were probably the best known, although there were many others from manufacturers worldwide.  The one shown next is one (of a small few) that I have, and it's a good meter for a very sensible price and is currently available (at the time of writing).

Figure 1 - Analogue Multimeter

The multiple scales are probably the thing that cause most people trouble with analogue meters.  The ohms scale is reversed, with 0Ω at full scale.  There are three different scales for DC volts, with 10, 50 and 250V ranges, and you use the one that corresponds to the selected range (dividing or multiplying by 10 as needed).  AC volts has a separate scale for the 10V range, and it's not linear because the internal diode voltage drop affects the reading more at lower voltages.  The AC voltage ranges also have an impedance that's usually only 9kΩ/V, so that will impose greater loading on the circuit.  In almost all cases, the AC volts ranges are not AC coupled, so if DC is present as well, that will affect the reading.

The meter shown in Fig. 1 has extra functions, namely transistor testing.  Mostly, this is next to useless (as is the same function on digital meters), because it's often hard to use, and/ or doesn't test the devices at a realistic voltage or current.  I have several meters that include transistor tests, but they are never used.  Diode tests are another matter, but the meter does not show the resistance, but shows the forward voltage drop (but be aware that the test lead polarities are almost always reversed, so Red is negative).  Digital meters with a 'diode test' range also show the forward voltage drop, but do not reverse the polarity.

One thing that is often very handy is the lowest ohms range.  Unlike a digital meter that is 'auto-calibrated', you can set the ohms range to zero with the test leads in place, thereby eliminating them from the measurement.  This lets you measure less than 1Ω (don't expect high accuracy), where a digital meter requires that you subtract the lead resistance from the total.  Some digital meters have a 'relative' function that lets you zero the meter with the test leads in place, although it's not always obvious, and most don't have that option.  Be aware that some analogue meters have a fairly high current on the low ohms range that may damage some components.  The current can be over 150mA - look at the schematic below, in particular R11 (18.5Ω).

Figure 2 - Analogue Multimeter Schematic

The circuit shown is not meant to be representative of any particular meter, but provides the basics.  The switching is invariably convoluted, because it's almost always a simple rotary switch on the outside, but it has to make all the right connections for every range internally.  The majority now use a pattern etched into a PCB, often with gold plating to eliminate problems due to corrosion.  The rotor itself has a set of joined contacts that connect the input and output for each range appropriately.  Earlier meters used a rotary switch with four or more separate wafers, made specifically for the manufacturer.

Note that AC voltage measurements are almost always ½ wave rectified, and the meter is calibrated for RMS based on the average value.  This leads to significant errors with asymmetrical waveforms, and any input signal that is not a sinewave.  It also means that low voltages cannot be measured because of the diode voltage drop.  Although most used germanium diodes (with a forward voltage of around 150-300mV, depending on current), that still meant that measuring less than 1V AC would introduce errors.  This is why most have a separate scale to 10V AC, and it's compensated for the diode nonlinearity.

Note that the meter (and indeed many digital meters as well) doesn't include an AC current range.  This is due to the very basic ½ wave rectifier used, which won't work at low voltages.  Depending on the types of measurement you normally expect to make, this may or may not be an issue for you.  To get accurate AC current measurements almost always means using a True RMS (digital) meter.

One thing that is common with better units is the use of two batteries for ohms readings.  With a 1.5V (or 3V) supply, the maximum resistance that can be measured is limited by the voltage and the meter sensitivity.  A 50µA meter can only read up to 30k with 1.5V, or 60k (both full scale) with 3V.  Including a 9V battery (in the above it's in series with the 3V battery, giving 12V total) lets you measure up to 180k full scale (padded back to 100k with additional resistors).  This allows 1MΩ and above to be measured, but at the lower end of the scale for anything over 1MΩ.  In the heyday of 'simple' analogue meters, most resistors were generally only 5% tolerance, so any error was pretty much immaterial.  A quirk of these meters is that the terminal voltages are reversed on the ohms ranges.  If you're unaware of this, diode readings won't make sense.

Prior to the advent of digital meters, the impedance limitation of standard moving-coil multimeters led to the development of the VTVM, typically using a dual triode valve to drive the meter movement.  This made it possible to have a constant input impedance of (usually) 10MΩ (or 11MΩ with a 1MΩ probe resistor).  This all but eliminated the problem of loading the circuit under test, and the input impedance remains constant on all DC voltage ranges.  AC voltage measurements were generally less advanced, and were equally sensitive to waveform distortion.

VTVMs usually used the same rectifier (½ wave) as standard multimeters, as did later FET voltmeters that worked the same way as a VTVM, but with much lower power consumption.  These have not disappeared, with several new FET models still available.  Original versions regularly command fairly high prices as 'vintage' test gear.  The only real advantage was a higher input impedance, and most were no more accurate than 'ordinary' multimeters.

Figure 3 - Basic FET Voltmeter (11MΩ Impedance)

You won't find a new VTVM on sale, but FET voltmeters are available, and I've included a very basic schematic above.  As shown, it's intended for DC only, and this arrangement was often sufficient when working with high impedance circuits using valves.  Most other measurements could be taken using a normal multimeter, but the vacuum tube or JFET meter would provide negligible circuit loading so that measurements on high impedance circuits would not cause circuit malfunction.  In the circuit shown, the two JFETs must be matched, and in good thermal contact with each other.  The 'Balance' and 'Cal' pots are internal, but the 'Set Zero' pot was always available on the front panel.  JFETs are more stable than valves, but both drift and require adjustment.

Note that the circuit uses JFETs that are no longer available, and quite a few changes would be needed to make it work with the few choices available today.  It's certainly not a circuit that I'd recommend that anyone try to build, and it's shown only for its historical significance.  There are many similar circuits shown on the Net, with some having a better chance of working than others.  In almost all cases, suitable JFETs are no longer available.

If you needed something like that these days, a FET input opamp would be the preferred option.  With high gain, excellent linearity, low drift and an input impedance of around 1TΩ (1E12Ω), making a meter with high input impedance has never been easier.  Of course there's far less need for a simple high-voltage DC meter any more, because most circuitry is low impedance and even a very basic analogue multimeter will measure most voltages just fine.  However, there are exceptions!  The majority of digital meters have a high enough input impedance that very few circuits will be affected.  However, the inherent limitation of all digital meters still applies - you can't read fluctuating voltages easily (if at all).

Other 'vintage' valve and FET meters used a switching system similar to that in the passive multimeter, although some only included AC and DC voltage, so the user needed a standard multimeter for measuring ohms.  This wasn't an issue at the time, since most people involved in electronics had one (or more) standard multimeters as well.  A few VTVMs included a parallel capacitive voltage divider (as shown in Project 16) so that AC voltage measurements extended to beyond 20kHz.

Note that it is possible to have an analogue readout with a True RMS AC measuring capability.  This means that it must have internal electronics, and will require power (either battery or mains), and other than a few AC millivoltmeters there are none available that I'm aware of.  The True RMS converter will typically be the AD636, AD737 or (perhaps) an LTC1967.  These are not inexpensive ICs, so don't expect to find any of them in low-cost meters (analogue or digital).  Most have an input sensitivity of 200mV.  There's an Application note (AN268) from Analog Devices that describes the use of RMS converter ICs.  It's well worth reading to find out how these devices are used.

2   Digital Meters

These days, probably 99.9% of all multimeters sold are digital.  I doubt that I need to show a photo of one, but I'll do so anyway.  Most people use hand-held meters, but a good bench type multimeter is well worth having if you can justify the cost.  I use both, but in the workshop the bench meter is always my first choice.  A unit such as that shown below is under AU\$400.00, which isn't cheap, but you do get a lot of meter for the money.  One thing that (IMO) is absolutely essential is 'True RMS'.  This means that the meter will show the actual RMS value of an AC signal, regardless of distortion.

With the averaging measurement system used by budget meters, the reading can be so far off the mark that the measurement is useless.  This topic is covered in AN-012, Peak, RMS And Averaging Circuits in the ESP app. notes section.  As an example, a symmetrical squarewave will read 11% high, and a pulse-train can measure as much as 90% low.  True RMS meters used to command a very high price, but these days a hand-held True RMS meter can be bought for less than AU\$50.00.  The meter shown below is the one that I use most of the time.  It's served me well, as I've had it for many years.  It's mains powered, but that's common for bench meters and isn't a problem because they aren't moved around.  Readouts that extend to 55,000 counts give good low-level resolution, and True RMS bench meters can be found for less than AU\$250.00 at the time of writing.  You can pay a great deal more of course, and you need to check the specifications.

The basic building block for digital meters is the analogue to digital converter (ADC).  The most common arrangement is a dual-slope integrating type [ 4 ].  These are a low-cost but very linear ADC, with the ICL7106/7 3½ digit devices (maximum reading 1.999) being very common, and the maximum voltage that's displayed is 199.9mV.  There are other ICs too, but a detailed description of those available is outside the scope of this article.  External circuitry is required to allow measurement of voltages above 200mV, AC, current and resistance.  There are application notes available should you wish to build your own, but given the low cost of 'standard' 3½ digit multimeters making one is ill-advised.  It will cost a great deal more to build than you can pay for one that's available (often less than AU\$10.00).  3¾ digit meters (3.999 maximum reading) are available for less than AU\$40.00 and there's no way you could build one for less.

Figure 4 - Digital Bench Multimeter

One thing that a lot of digital meters (and almost all analogue meters) are very bad at is measuring high frequencies.  Some digital meters include a frequency counter that may extend to 10MHz or more, but voltage measurements may be limited to only a few kHz.  'Ordinary' (not True RMS) meters are generally worse than True RMS types, with some being incapable of measuring anything beyond 2-3kHz.  To make matters worse, most don't fully specify the frequency range for AC measurements, and some don't mention it at all!  Most True RMS meters extend to at least 10kHz, and often much more.  The one shown in Fig. 3 has been tested fairly thoroughly, and is within 2% at 90kHz.  That makes it useful for audio frequency measurements, but not much more.

This doesn't mean that average-reading meters are of no use, because not everyone needs to measure AC voltages other than low-frequency sinewaves (or close to, such as the AC mains derived voltages from transformer windings).  Digital multimeters are now pretty much all that most people ever use, but there are traps for the unwary.  The greatest of these is accuracy.  You may wonder how this could possibly qualify as a 'trap', but mistakes are very common, because we believe the digits.  If the meter shows the output of a 5V regulator as 5.00V everyone is happy, but that doesn't consider the error that's inherent in all meters and regulators.

That exact 5V measured may actually be anywhere between 4.93V and 5.07V, allowing for 1% ±2 digits accuracy.  I've lost count of the number of people who've built a ±15V P05 (or other regulated supply) and said that their meter showed +14.8V and -15.3V, and wondering if this is alright.  Anyone used to an analogue meter would simply look at the pointer, see that it's 'close enough' to the required voltage and move on.  The implied precision of a digital multimeter has people wondering what's wrong when a 5V supply measures 5.1V (or 4.9V) when there's absolutely nothing amiss.

Nothing in life is perfectly precise.  Regulators have small errors, as do the meters used to measure their outputs.  All electronic parts have some leeway for supply voltages, and it should be obvious that a small error won't cause any problems.  Most opamps will happily operate with +27V and -3V if you want them to (but obviously this is not the case for those with a 5V maximum supply voltage), and logic ICs (including PICs, microcontrollers and CPUs will all handle voltage variations as shown in the datasheet.  For example, processor ICs (CPUs) are probably the most fussy as they operate at low voltages (2.7V and/ or 3.3V).  Even these have leeway, and the 5V supply is expected to be within the range of 4.75V and 5.25V (±5%, [ 5 ] ).  Unless a multimeter has been damaged, the voltage you read will normally be more accurate than the supply requirements.

If you need to make accurate measurements, a 6-digit (or 5½ digit) meter is worthwhile.  You can get more, but ultimately noise becomes the dominant factor and limits the ultimate resolution with AC measurements.  DC is usually less restricted, as multiple readings can be averaged.  The end result will be as accurate as the meter allows.  Most digital meters offer auto-ranging, so you only need to select the quantity to be measured (DC V, AC V, ohms, etc.) and the meter will adjust itself to give the most appropriate display.  This is in contrast to analogue meters where you must select the range.  If you were to select (accidentally or otherwise) the 0.5V DC range and try to measure 230V or 120V AC, the meter will almost certainly be destroyed.  Many have an internal fuse, but that's often only provided for the 10A range (this applies to both analogue and digital meters).

Low resistance readings often pose a problem unless the meter has a 'relative' function.  This lets you short the leads, then select the relative function which resets the reading to zero.  Provided you make a good connection to the device under test you'll get a fairly accurate measurement.  Don't expect to measure much below an ohm or so with any accuracy though, as for that you need a dedicated 4-wire measuring technique.  This technique is described in Project 168 (Low Ohms Meter).

3   Burden Resistance

Almost all multimeters us a shunt for measuring current.  The manual may or may not indicate the value, but it can me measured if you have another meter.  Note that this basic technique only works if the meter is not auto-ranging, and you must be willing (and able) to perform a meaningful test.  This is not likely to be easier, as you need a variable power supply that can deliver the current needed.

Connect the meter (in DC current mode) to your power supply, with a series resistor to limit the current.  Most meters only measure up to 250mA or so (not including the 10A shunt if provided).  Apply a voltage to obtain a low current (around 2.5mA or to suit the lowest range), and measure the voltage across the multimeter.  If the voltage measured is 250mV with 2.5mA displayed, the resistance is determined by Ohm's law, and is 100Ω.  The same procedure is used for higher ranges (10mA, 25mA and 250mA for example).

By knowing this, you can work out the voltage drop across the meter when measuring current.  If you ignore the voltage drop, you may get readings that don't seem to make sense.  The more you know about your test equipment the better, as you're less likely to get readings that appear to be wrong.  No-one expects the user to know everything, but there are always things that you need to know to get the best from the meter.

This technique is just as important (if not more so) for digital meters, because we all see the digits and believe the number, even if it's wrong!  With an auto-ranging meter, you need to increase the current (from the initial low value) until the range switches, and (with some) the voltage across the meter suddenly drops.  Set the current for that range to something that makes an easy calculation, and you'll soon know the voltage dropped across the meter for each range.

I tested my bench meter, and it uses a constant resistance for all current measurements.  It extends to 800mA, and shows a voltage drop of 1.28V at 800mA, a burden resistance of 1.6Ω.  At low currents this is immaterial, dropping only 16mV at 10mA, but it becomes significant at higher current.  It's never caused me any problems though, because I'm aware of it, even though I hadn't measured it before.  The burden may be specified as a mV/mA figure, which in the case of my meter is 1.6mV/mA.  Note that this does not include the test leads, and most specification that describe the burden won't include them either.

I tested another (switched range) digital meter, and it measured 100Ω on the 2mA range, 10Ω on the 20mA range and 1Ω for the 200mA range.  These correspond to the basic digital meter IC sensitivity, which is typically 200mV full scale.  With an analogue meter, the voltage across the shunt(s) must be sufficient to deliver the movement's full-scale current (e.g. 50µA).  The unknown quantity is the internal resistance of the meter itself, which is typically 2kΩ, but it can vary from around 1.5kΩ to 5kΩ depending on the way the movement was made (particularly magnetic field strength).

In the analogue meter circuit (Fig. 2), you can see that the current ranges from 2.5mA to 250mA use odd value resistors (R8, R9 and R10).  This is due to other resistors that are in series and parallel with the movement, which skews the ranges.  As a result, a little more voltage is needed at low current than at high current (there's a total of about 2.49k [R7, R22 and VR1+R25] in parallel with the movement, and 240Ω [R21] in series).  I'm not about to perform a full circuit analysis for each range here, but it all works out to better than 1.6% accuracy.

4   Taking Measurements

Taking measurements is easy for voltage (AC or DC), and if you have a meter with switched ranges and an unknown voltage, it's a good idea to always select the highest range first, and reduce the range until you have a sensible reading.  Whenever you use the milliamps or ohms ranges, make sure that you put the red plug back into the correct socket for measuring voltage as soon as you're done.  It's all too easy to damage your meter or the circuit being tested if you try to measure voltage when the test lead is plugged into the current measuring socket.  A (very) few meters have mechanical shutters that block access to the current measurement socket unless current measurement has been selected.

Current readings always require that you break the circuit, so the meter is in series with the power supply and the device under test (DUT).  This is often a real nuisance, and I will often use a series resistor and measure the voltage across it.  The resistance needs to be selected based on the expected current draw of the DUT.  For example, if you expect it to draw 100mA and the supply voltage is more than 5V, a 1Ω resistor will show 100mV across it at 100mA.  The powered circuit gets 100mV less voltage than intended, but nearly any 5V circuit will function normally with 4.9V.  The meter does the same thing, with selected shunt resistors for each range (look at R8, R9 and R10 in Fig. 2).  All 'conventional' meters (analogue and digital) use the same arrangement, so when measuring current there is always some of the supply voltage dropped across the meter.  This is known as the 'burden' resistance.  The meter doesn't measure current directly, but instead does the same as an external resistor.  The voltage across the resistor is displayed, but read as current (see previous section).

AC current measurements are subject to the same limitations as AC voltage measurements.  Almost all meters that are capable of measuring AC amps/ milliamps are True RMS types, because a simple rectifier has too much voltage loss to allow measurement of AC at low voltages.  The meter must be capable of providing an acceptable frequency response, and again, True RMS meters are likely to have better high frequency response than simple ½ wave rectifiers.  Unless you run your own tests to verify frequency response, assume that most True RMS meters will be limited to about 5kHz at most, and 'ordinary' meters usually somewhat less.  Most low-cost digital meters don't offer AC current ranges, because to do so requires a precision rectifier (although some may perform the rectification using the meter's processor IC if it has that capability).

5   Insulation Testers

Insulation testers (often called 'Meggers' after the original insulation testers - the name is a registered trade mark, but has become 'generic') are a special case.  Most are designed for one task - measuring insulation resistance.  In Australia, the standard test for household mains wiring (with the test performed before the energy supplier will connect the mains) is 500V DC, with a minimum resistance of 1MΩ.  Normally, each mains conductor (active [hot] and neutral) will be tested between each other and to mains earth.  Some insulation testers also include a high-current earth (ground) test mode, and/ or selectable voltages (usually 500V or 1,000V DC for 230V countries).

These are specialised tools, and are often rather costly.  However, the verification that all building wiring be safe is somewhat more important than the one-off cost of the tester.  Another specialised tester is a 'PAT' (portable appliance tester) unit, which is used for testing individual portable appliances, extension and removable mains leads.  Most workplaces will have a test regime set up so that all equipment that plugs into a wall outlet is tested regularly.  The electrical tests include polarity (active, neutral or earth not swapped), leakage (mains to chassis or earth connection) and earth continuity (the earth connection must not exceed 1Ω).  A more rigorous test uses a high current to verify that the earth connection can carry at least 10A without failure.

I won't go into any more detail on these testers, because they are specialised and are not applicable for normal hobbyist workshop duty.  Having said that, I do have a 1kV insulation tester that I use to verify that the old transformer I may be thinking of using is still 'safe'.  I also sometimes use it to check that transistors are truly isolated from the heatsink (25µm thick [thin?] Kapton will withstand 1kV easily).  Mine probably only gets dragged out a few times each year, but if I didn't have it I'd have to buy one, as they are useful.  This is especially true as I do a wide variety of tests and experiments, not all of which are audio related.

6   Millivoltmeters

Dedicated AC millivoltmeters are another useful tool, but they are generally fairly expensive.  Most use a moving-coil (analogue) readout, and almost invariably use a 1-3-10 range switch.  This is done to get 10dB steps, and the actual ranges are 1-3.16-10 (along with multiples and sub-multiples thereof).  Most will measure down to 3mV full scale, and have a frequency response that remains flat up to at least 100kHz.  An example of a DIY version is shown in Project 16, and it has a capacitive voltage divider in parallel with the resistive divider.  This minimises the effects of stray capacitance that causes serious errors at frequencies above around 10kHz.

Like insulation testers, these are fairly specialised, and are only useful if you have an audio oscillator (with flat frequency response across the range) so you can perform frequency response tests.  They are also at the heart of distortion meters, with the final measurement calibrated in % THD (total harmonic distortion plus noise) - see Project 52 for an example of a distortion meter.  The circuit shown needs a millivoltmeter to measure the THD.

I've been using my audio millivoltmeter for around 40 years, and couldn't be without one.  There are currently three that get used, two of which are in distortion meters, and the Project 16 version which is stand-alone.  The P16 unit is particularly useful with high impedance circuits, as it has a 2MΩ input impedance.  Most hobbyists won't need one, but for design work a millivoltmeter is an invaluable tool, but the same tests can be done using an oscilloscope (albeit with a few calculations to convert to dB).  Although my digital bench meter (Fig. 3) can also measure millivolts, it's very slow, and it's completely useless for many tasks.  Reading a steady-state low voltage works well enough, but having to wait for up to 10 seconds for a stable reading is sub-optimal (to put it mildly).

Conclusions

Both analogue and digital meters are useful, and while I don't use my analogue meter a great deal, it's often much faster to see that the pointer shows 'about right' than to have to read the digits.  Slowly moving voltages or currents are easily visible, and when cyclic it's easy to see the average with an analogue meter, but almost impossible with digital.  If you plan on getting yourself an analogue multimeter, you generally should expect to pay at least AU\$30.00 or so (very expensive ones are also available).  Avoid anything advertised for less than ~\$15.00 or so, as you'll almost certainly be bitterly disappointed.  Many of the very cheap ones aren't even useful to use for spare parts - they are complete rubbish and not fit for purpose.

For a digital meter, get (at least) one with True RMS.  They are usually fairly inexpensive, and even a cheap unit is handy to have as a spare, or to measure current while you use another to look at voltages.  A well-equipped workshop will have several, and it's a very good idea to compare them regularly so you know that they all read the same voltage, current and resistance, within their accuracy specifications.  If you find one that reads vastly differently from the others, you know it's out of calibration.  Most can't be re-calibrated because the information you need isn't provided, so it may well end up as scrap, or a source of spares (handy if you have a couple that are the same make and model).

You always need to make sure that you've selected the right range for the intended measurement.  Many (but by no means all) digital meters have some degree of protection built in, but if you try to measure 230V AC with the meter set for milliamps (AC or DC), expect instantaneous failure.  Fuse protection is unlikely to save the meter from destruction, but it reduces the risk of fire or melting the case which may expose live parts.  If you're not vigilant, this can be surprisingly easy to do, and I recommend that the meter is set for AC or DC volts (with a high range selected for analogue meters) when it's not being used.  Naturally, the probes should be plugged into the 'normal' sockets (if a separate connection is used for milliamps).  Make sure that the meter is used in a position where it can't fall.  Even 'cushioned' cases won't save an analogue movement if it falls, with the most common failure being that the moving coil assembly 'jumps' out of its jewelled bearings and jams.  This can be fixed if you have a good eye and a steady hand, but it's often quite tricky and you're dealing with very small (and delicate) electromechanical parts.

One final point: many (mainly cheap) analogue meters use a small (typically 2mm diameter) pin on the test leads, rather than (now almost always shrouded) 4mm (nominal) banana plugs and sockets.  Avoid the small ones if at all possible (or at all costs), because it means that you can't use your other 'general purpose' test leads.  I have at least ten (maybe more) leads with banana plugs and alligator clips that are used with a variety of meters, power supplies, loads, etc, and I very rarely use meter probes.  This isn't a recommendation, it's just the way I've always worked - clip leads stay where you put them, but I do have an insulated 'probe' that I can attach a clip to when necessary.  Most people will use the probes supplied with the meter, and if that works for you then it's all good.