Measure an amplifier's THD without external filters
Adding a secondary amplifier boosts the THD to measurable levels.
Ken Mendez, Maxim Integrated Products, Sunnyvale, CA; Edited by Paul Rako and Fran Granville -- EDN, January 5, 2012
Measuring an op amp’s THD (total harmonic distortion) at frequencies below 100 kHz is problematic because modern amplifiers have a THD rating that is much lower than that of most test equipment. The circuit in this Design Idea uses active noise amplification to eliminate the need for external filters. This approach allows you to measure THD at several orders of magnitude lower than the resolution of the test fixture you are using.
Many THD test circuits use passive-
amplification techniques to force
the DUT (device under test) to correct
itself. This approach can generate
considerable error by terminating the
distortion signal with resistive loads.
A preferred way to generate accurate
data is to isolate the DUT and measure
it using a high-impedance buffer.
This method requires active amplification
of the distortion at the amplifier’s
output. You can use a secondary
operational amplifier for active amplification
(Figure 1). You should use a
signal source with THD better than
−70 dB. Feed this signal through the
DUT, which distorts it. The output
of the DUT is a combination of the
input signal and the distortion from the
DUT multiplied by the gain of the DUT
stage. You can set the gain and resistive
loading of the DUT according to its specification in the data sheet. This
signal at the DUT’s negative terminal is
the input signal plus the distortion but
without any gain. Connect this node
to the positive input of the secondary
amplifier. You can calculate the output
voltage based on the circuit parameters,
according to the following equations:
To detect small distortions, set up
the secondary amp for a gain of 1000
and 
connect the amplifier’s gain-setting
resistor to the input signal. This step
allows amplification of the distortion
but nulls out secondary amplification
of the input signal. Consequently, when
referenced to ground, the circuit’s output
is the original input signal plus the
secondary gain times the output distortion
of the DUT.
connect the amplifier’s gain-setting
resistor to the input signal. This step
allows amplification of the distortion
but nulls out secondary amplification
of the input signal. Consequently, when
referenced to ground, the circuit’s output
is the original input signal plus the
secondary gain times the output distortion
of the DUT.
If you want to remove the input
signal from the circuit’s output, you
can set up the secondary amplifier in
a differential configuration (Figure
2). Some amount of the input signal
still reaches the output because of
the buffer amplifier circuit’s CMRR.
However, the signal decreases so that
it does not affect the measurement of
the amplified distortion. Exact matching
of the input and feedback resistors
of the secondary amplifier reduces the
CMRR effects. Calculate the output
of this differential circuit using superposition,
according to the following
equations:
Talkback
-
As one of the authors of 3 international patents for this idea and the article, I should tell that implementations shown on Fig.1, Fig.2 and Fig.3 probably could not work well for THD measurements.
The reasons are:
1.The article does not included the main equation for THD
2.The article does not explained what will be used for measurement RMS voltmeter or some digitizing approach.
3.The differential amplifier could not reject well the first harmonic of signal, not only because CMRR is not enough, but because DUT bring phase shift in several degree. This is the main killer of this method. We should also remember that a phase shift could be change with an amplitude of output signal.
4.If I remember right, a 1 degree phase angel shift between a signal from a generator and output of DUT equals to 1.7% main harmonic level after rejection. It means if we have 1V output signal and would like to measure 0.1% THD, we need to reject the first harmonic at least 3000X or 70 DB. But because we could not reject better than 40 DB, it is clear that 1000X amplification is not correct and MAX9632 will be overloaded.
To make this idea work needs to do following:
1.Reduce gain of the differential amplifier,
2.Connect non-inverting input of differential amplier to the output of DUT
3.Use phase shifters and amplitude tuning circuitry.
The instruments based on filters have advantage compare with this method because could be used for oscillators THD measurements also.
For precission THD measurements I did use multi-stage substraction technique.
Vladimir Doubovis - 2012-22-1 23:03:21 PST -
Main problem with the circuit is that the CMRR of OpAmps drops very rapidly with frequency which might pose a problem if we need to measure an amplifier with very low distortion at unity gain.
Henry Chan - 2012-15-1 18:23:56 PST -
If it will be that easy to measure any amp distortion, why the major instrumentation companies were giving them the trouble of implementing output waveform fundamental rejection mechanisms in measuring instruments costing several thousand dollars ?
I believe that Vladimir is right: it CANNOT be that easy, and anyone that once played with an HP 333/334 in "manual" mode knows !
Jacques Fortin - 2012-13-1 11:32:59 PST -
It does not matter that the method was implemented in 80’s for power amplifiers integrated circuits THD measurements, including several international patents and articles.
The matter is that this idea still alive and some discussion about this method required.
The main issue is to increase rejection of the main harmonic of signal and as well to increase signal to noise ratio. With this purpose is better to add attenuator at the input of DUT and connect non inverting input resistor divider to DUT output. So you can work with higher amplitude of harmonics and noise floor level influence will be less.
There are a few factors that defined accuracy of measurements: CMRR of secondary amplifier plus tolerance of matching resistors, and as well a phase shift in DUT and the measurement Op Amp.
It is not a great idea to connect a non-inverting input of the second amplifier to inverting input of DUT due to an additional phase shift and DUT gain change.
The schematics Fig.3 usually used for non-linearity measurements not for THD.
Vladimir Doubovis - 2012-10-1 12:35:00 PST -
Congratulations Steve. You're an old timer. Just because you used that technique doesn't mean others have seen it.
DI lurker - 2012-9-1 13:41:58 PST
