Audio Millivoltmeter
Having now got an oscilloscope to view waveforms, and a signal generator to create the test signals, there needs to be a way of measuring signal amplitude.
For non-critical use, even a multimeter is usable, and these can be acceptably accurate for low frequencies. Most multimeters are designed for DC or mains frequency AC, not for audio frequencies up to 20kHz and beyond, but most will work with acceptable accuracy up to a few hundred Hz. Even beyond that, up to a few kHz, if one is doing comparative measurements, i.e. are these two voltages the same, which one’s greater, then the multimeter will be acceptable. However, for anything requiring an absolute voltage value, and for anything over a few kHz, a proper audio millivoltmeter will be needed. At least a multimeter is cheap, from around £10.
Audio voltages will range from a few microvolts, up to perhaps 50 volts or so with a high power power amp (100 watts is 28.3V into 8 ohms, 50 volts into 8 ohms is 312.5 watts). If you’re looking for an audio millivoltmeter, then a range of 15uV to 500V is ideal, 100uV to 100 V will be OK for most audio applications.
The Millivoltmeter should have an input impedance of 1Mohm, as this is the standard for oscilloscope inputs, which means that a standard oscilloscope probe can also be used for the millivoltmeter. A switchable input impedance, say 1Mohm, 600 ohm is nice to have but not necessary. Balanced inputs are also nice to have, but rare except in studio analysers like the Lindos or Neutrik. Balanced inputs are also generally of 10k-20k input impedance, not 1Mohm.
Bandwidth should be at least to 100kHz, 1MHz isn’t unreasonable, but it doesn’t have to be as wide as an oscilloscope as the millivoltmeter will be used for measuring amplitude and not for chasing spurious oscillations, or seeing square-wave risetimes. Having a switchable 20-20kHz bandwidth is very useful when making noise measurements, as that will eliminate HF noise, for example, due to noise shaping of a DAC’s output. Noise weighting filters are also a very nice facility, but again not essential unless one’s trying to confirm a detailed specification, or derive one for a new product.
Pretty much all audio millivoltmeters with analogue indicators will indicate ‘average’, a few will indicate true-RMS, and even fewer will have switchable PPM-VU-RMS etc indicators.
Pretty much all millivoltmeters work by having a calibrated amplifier with switchable gain, followed by a meter driver that activates the indicating instrument, whether moving coil meter or LCD display. Consequently, before the meter driver there is an amplified version of the input signal available that can be taken to a ‘scope and having such an output from a millivoltmeter is extremely useful. The millivoltmeter, then, can be used as general-purpose amplifier for very small signals, such as a pickup cartridge produces. It’s almost impossible to see the unamplified signal from a MC cartridge on an oscilloscope, but using the millivoltmeter as an amplifier, it can be seen easily, and the amplitude measured at the same time.
One final thought on millivoltmeters, is what is the maximum DC voltage the input will stand? This is important if one is working on valve amplifiers and want to measure an AC signal level on, say, a valve anode that could be standing at several hundred volts. Unlike oscilloscopes, audio millivoltmeters rarely go down to DC, so it’s not possible to select AC or DC coupled, they’re always AC coupled. This is fine as measuring DC is better done with a multimeter, but if you’re working on a valve amp it’s as well to be sure that the meter input won’t be damaged by the HT.
Distortion Factor meters
Distortion Factor is measured by putting a tone of a certain fundamental frequency through the Device Under Test, measuring the output then removing the fundamental tone and measuring the amplitude of everything left. This ‘everything left’ more properly called the Residual, will be the Total Harmonic Distortion of the fundamental plus noise. The amplitude of the residual as a percentage of the amplitude of the fundamental is the Distortion Factor. Distortion Meters, therefore, consist of a millivoltmeter to measure the amplitude of the fundamental, a selectable filter which is used to remove the fundamental tone, and switching to move the millivoltmeter to after the filter to measure the amplitude of the residual.
Nulling of the fundamental can be manual or automatic, and the ranging of the millivoltmeter can be manual or automatic, but the principle remains the same. Clearly, what one is measuring is not just the Total Harmonic Distortion, but the THD+N, and how low one can go with the distortion measurement depends on how low the noise is, and how many harmonics are captured, i.e. what the bandwidth is of the millivoltmeter doing the measurement. For a 1KHz tone and below, a 20kHz bandwidth will capture up to the 10th harmonic, which will be pretty much as good as it needs to be, but a 10kHz tone, a 20kHz bandwidth will only capture the second harmonic at best. It could be argued that perhaps we don’t need to know about higher harmonics of high frequency tones, as we can’t hear them, but that argument doesn’t take into account intermodulation distortion, and so most Distortion Factor Meters will operate over a 100kHz bandwidth to capture the higher harmonics of the highest audible frequency. This incidentally is also why using a sound card and software for distortion measurements is limited, and I’ll refer to this in a future post.
If you’re looking to buy a Distortion Factor Meter, you’ll need to check at what frequencies the tuneable filter works at, whether continuous, or just a few switched frequencies. Also important is what the incorporated millivoltmeter goes down to in terms of level, as this will determine what the minimum measurable distortion will be.
Distortion Factor Meters sum together all the harmonics, so it’s not possible from just the % number to know whether the distortion is a relatively harmless 2nd harmonic or a nasty 5th. However, my view is that provided the sum of all harmonics is low enough, then even if the residual is made up of the nastier higher harmonics, it’s not audibly important. The debatable point is, though, how low is low enough? Conventional wisdom used to say that we can’t hear even 1% distortion on music and speech provided the harmonic content isn’t extreme, even 3% is OK, given that Reel-Reel tape recorders run at around 3% THD at peak level...in fact, that’s how Peak Level was defined, as the 3% point. So if 1% is OK, an amplifier with 0.1% THD at all frequencies and power levels will be completely transparent, at least as far as distortion is concerned. I have not seen any evidence to suggest these numbers are wrong even today.
This leads me to mention that one needs to understand for what purpose one is buying a Distortion Factor Meter. If for checking that an amplifier is working ‘well enough’ or for working mostly on valve amplifiers, then a meter that will go down to 0.3% at Full Scale, means that one can check whether distortion stays below 0.1%. For SS amps, or for confirming specifications, then a meter that can go down to 0.1% Full Scale or lower, becomes more useful. My meter indicates 0.1% full scale, and with the generator I have, I can measure down to about 0.02%. My view is that if the distortion is sufficiently low that I can’t measure it, then it’s nothing to worry about.
If one wants to do an analysis of the harmonic structure of the distortion residual, then Spectrum Analyser is needed rather than a nulling filter, an altogether different instrument, which again, I’ll refer to in a future post.
S.
Having now got an oscilloscope to view waveforms, and a signal generator to create the test signals, there needs to be a way of measuring signal amplitude.
For non-critical use, even a multimeter is usable, and these can be acceptably accurate for low frequencies. Most multimeters are designed for DC or mains frequency AC, not for audio frequencies up to 20kHz and beyond, but most will work with acceptable accuracy up to a few hundred Hz. Even beyond that, up to a few kHz, if one is doing comparative measurements, i.e. are these two voltages the same, which one’s greater, then the multimeter will be acceptable. However, for anything requiring an absolute voltage value, and for anything over a few kHz, a proper audio millivoltmeter will be needed. At least a multimeter is cheap, from around £10.
Audio voltages will range from a few microvolts, up to perhaps 50 volts or so with a high power power amp (100 watts is 28.3V into 8 ohms, 50 volts into 8 ohms is 312.5 watts). If you’re looking for an audio millivoltmeter, then a range of 15uV to 500V is ideal, 100uV to 100 V will be OK for most audio applications.
The Millivoltmeter should have an input impedance of 1Mohm, as this is the standard for oscilloscope inputs, which means that a standard oscilloscope probe can also be used for the millivoltmeter. A switchable input impedance, say 1Mohm, 600 ohm is nice to have but not necessary. Balanced inputs are also nice to have, but rare except in studio analysers like the Lindos or Neutrik. Balanced inputs are also generally of 10k-20k input impedance, not 1Mohm.
Bandwidth should be at least to 100kHz, 1MHz isn’t unreasonable, but it doesn’t have to be as wide as an oscilloscope as the millivoltmeter will be used for measuring amplitude and not for chasing spurious oscillations, or seeing square-wave risetimes. Having a switchable 20-20kHz bandwidth is very useful when making noise measurements, as that will eliminate HF noise, for example, due to noise shaping of a DAC’s output. Noise weighting filters are also a very nice facility, but again not essential unless one’s trying to confirm a detailed specification, or derive one for a new product.
Pretty much all audio millivoltmeters with analogue indicators will indicate ‘average’, a few will indicate true-RMS, and even fewer will have switchable PPM-VU-RMS etc indicators.
Pretty much all millivoltmeters work by having a calibrated amplifier with switchable gain, followed by a meter driver that activates the indicating instrument, whether moving coil meter or LCD display. Consequently, before the meter driver there is an amplified version of the input signal available that can be taken to a ‘scope and having such an output from a millivoltmeter is extremely useful. The millivoltmeter, then, can be used as general-purpose amplifier for very small signals, such as a pickup cartridge produces. It’s almost impossible to see the unamplified signal from a MC cartridge on an oscilloscope, but using the millivoltmeter as an amplifier, it can be seen easily, and the amplitude measured at the same time.
One final thought on millivoltmeters, is what is the maximum DC voltage the input will stand? This is important if one is working on valve amplifiers and want to measure an AC signal level on, say, a valve anode that could be standing at several hundred volts. Unlike oscilloscopes, audio millivoltmeters rarely go down to DC, so it’s not possible to select AC or DC coupled, they’re always AC coupled. This is fine as measuring DC is better done with a multimeter, but if you’re working on a valve amp it’s as well to be sure that the meter input won’t be damaged by the HT.
Distortion Factor meters
Distortion Factor is measured by putting a tone of a certain fundamental frequency through the Device Under Test, measuring the output then removing the fundamental tone and measuring the amplitude of everything left. This ‘everything left’ more properly called the Residual, will be the Total Harmonic Distortion of the fundamental plus noise. The amplitude of the residual as a percentage of the amplitude of the fundamental is the Distortion Factor. Distortion Meters, therefore, consist of a millivoltmeter to measure the amplitude of the fundamental, a selectable filter which is used to remove the fundamental tone, and switching to move the millivoltmeter to after the filter to measure the amplitude of the residual.
Nulling of the fundamental can be manual or automatic, and the ranging of the millivoltmeter can be manual or automatic, but the principle remains the same. Clearly, what one is measuring is not just the Total Harmonic Distortion, but the THD+N, and how low one can go with the distortion measurement depends on how low the noise is, and how many harmonics are captured, i.e. what the bandwidth is of the millivoltmeter doing the measurement. For a 1KHz tone and below, a 20kHz bandwidth will capture up to the 10th harmonic, which will be pretty much as good as it needs to be, but a 10kHz tone, a 20kHz bandwidth will only capture the second harmonic at best. It could be argued that perhaps we don’t need to know about higher harmonics of high frequency tones, as we can’t hear them, but that argument doesn’t take into account intermodulation distortion, and so most Distortion Factor Meters will operate over a 100kHz bandwidth to capture the higher harmonics of the highest audible frequency. This incidentally is also why using a sound card and software for distortion measurements is limited, and I’ll refer to this in a future post.
If you’re looking to buy a Distortion Factor Meter, you’ll need to check at what frequencies the tuneable filter works at, whether continuous, or just a few switched frequencies. Also important is what the incorporated millivoltmeter goes down to in terms of level, as this will determine what the minimum measurable distortion will be.
Distortion Factor Meters sum together all the harmonics, so it’s not possible from just the % number to know whether the distortion is a relatively harmless 2nd harmonic or a nasty 5th. However, my view is that provided the sum of all harmonics is low enough, then even if the residual is made up of the nastier higher harmonics, it’s not audibly important. The debatable point is, though, how low is low enough? Conventional wisdom used to say that we can’t hear even 1% distortion on music and speech provided the harmonic content isn’t extreme, even 3% is OK, given that Reel-Reel tape recorders run at around 3% THD at peak level...in fact, that’s how Peak Level was defined, as the 3% point. So if 1% is OK, an amplifier with 0.1% THD at all frequencies and power levels will be completely transparent, at least as far as distortion is concerned. I have not seen any evidence to suggest these numbers are wrong even today.
This leads me to mention that one needs to understand for what purpose one is buying a Distortion Factor Meter. If for checking that an amplifier is working ‘well enough’ or for working mostly on valve amplifiers, then a meter that will go down to 0.3% at Full Scale, means that one can check whether distortion stays below 0.1%. For SS amps, or for confirming specifications, then a meter that can go down to 0.1% Full Scale or lower, becomes more useful. My meter indicates 0.1% full scale, and with the generator I have, I can measure down to about 0.02%. My view is that if the distortion is sufficiently low that I can’t measure it, then it’s nothing to worry about.
If one wants to do an analysis of the harmonic structure of the distortion residual, then Spectrum Analyser is needed rather than a nulling filter, an altogether different instrument, which again, I’ll refer to in a future post.
S.
Last edited by a moderator: