Audio Amplifier Amplitude Response Measurement

Amplitude Response Audio Amplifier

The report shows the measuring technique of the amplitude response for an amplifier. A software supported measurement in connection with precision circuit analyzers and a sine level generator is presented.

What means amplitude response?

Short Introduction

An amplifier has the task the voltage applied at the input in such a way to amplify that the voltage at the output has the accurate same process. The user has in the hand whether the voltage applied at the input is to appear multiplied by a factor at the output. This adjustable factor is the volume control. It is of elementary importance that this multiplying factor remains always with all input voltages, frequencies and loads the same.
Amplitude Response
The amplitude response shows, the multiplying factor is how constant over the frequency? There are many representation possibilities, the fantasy is set no borders. A function representation is usual in the XY coordinate system, whose axles are however often also linear shown logarithmically to the basis 10. The professional among the readers, asks now immediately for the phase response. Correctly, both are two inseparable brothers and sisters. It is forgiven me that the phase response is measured with another measuring system and treated therein separately.
Bandwidth what is that?
Is in principle a simple affair, which is mathematically fast avowed. It is with an amplifier a useful characteristic, but something misleading, in particular for a layman.
The range is completely primitively said a guideline for the speed of an amplifier,: the more highly the range the faster. The range is recognizable from an amplitude response. It is a size, which is often wrongly regarded in purchase to an amplifier. The range shows the point in the amplitude response, at which the output voltage dropped to approximately 70.7% (1/square root 2) to target the output voltage. Now the large matter of price: would you like to use an amplifier with a frequency, with which the amplifier causes already 100%-70.7% = 29.3% errors? Assumed the range amounts to only 20 kHz, such an amplifier is how good?
- Bad - you can save your time to read a test report. To power on mesurement devices for something like that - no - only you want to see the proof, it's always the same lyre. Therefore all measuring instruments to start is pure waste of time, it is one wants exactly negatives to confirm, what one can already foresee. - Yes, - the defender of its noble pieces comes now with this popular argument: the human hearing can hear only about 16 kHz -20 kHz. There say I nevertheless easily against it 29.3% error with 30 kHz. With 20 kHz there is then already too much (is too idle me to calculate because of critics for 20 kHz and a low-pass of certain order). A lameness in the amplifier draws a tremendous chain of other negative effects behind itself ago. Which correctly decent begins with approximately 100 kHz range and at the same time high open loop under full load conditions. Not only fast by high range, but also precisely by the high open loop gain.

The pictures show the results of the system from two HP3457A (here 6.5 places) and a HP3336B frequency Sythesizer. The measurement happened in a friendly calibration laboratory, which put the devices to me at the disposal. In addition I wrote an automatic program sequence under utilization of the GPIB interfaces.

Measurement shows the ganging of both meters connected to the same signal source.

Graph shows the very constant signal source amplitude response measured with the input voltmeter

Graph shows the very constant signal source amplitude response measured with the output voltmeter

The stability of the signal source is fine, but not essential as long as the meters doing their job satisfactory. The 3336 together with the option 05 (precision amplitude attenuator) has an amazing constant amplitude. The decreased amplitude at higher frequencies has his origin in the amplitude response of the meter itself. And of course may be also from the generator (I don't think so much) or the cables or a marginal wrong termination. It's not cleary why, but for audio applications inconsiderable.

The picture left shows the measurement consists of two HP3457A and a generator HP3336B. The cables are connected like a mirror and terminated with 75 ohm. For best results the resistor should be on the end of the cable, it's difficult here to do it without an special soldered shared construction, so I missed to do it, sorry. Note the frequency is adjusted to 1 MHz.

How to get a higher accuracy?

In particular to settle the claim of "Ultra" there is some additional work to do.

a fresh calibrated instruments (constant 0.003 dB could be reduced)
same temperature during calibration and measurement, as possible measure direct after calibratiion
use for both meter same cable types and length and manufactorer
as possible ideal refexion free termination of the generator impedance
calibrate over many points as possible, put the values in a computer, do a fit function and use this function to correct future measured values
set instrument resolution to 7 1/2 digits (leds to a long measurement time)
do many measurements at same frequency and calculate mean value

More handling ways for a better accuracy:

if you have the change use a HP3458A meter (up to 8 1/2 digits). The 3457A is already one of the best meter HP builts in the past, but the 3458A that's really taking "the biscuit of accuracy". Both meters having a high accuracy.
Ask the calibration laboratory particularly conscientiously to work. The laboratory is to measure as much as possible points, in order to produce a family of characteristics, which can be out-corrected computationally. Thus the accuracy is close at the standard.

Does it make sense to do this stress of an improved accuracy because of audio measurements?

no
the measurement time should be not too long, the device under test (audio amplifier) will drift with his amplitude response vs. time and room temperature certainly more than the accuracy of the measurement system.
for audio amplifier (also the best ones) is this system already very good with enough room for accuracy.

The graphs showing details of some measurement ranges. Most meters having their highest precision under a maximum utilisation of their range. But these instruments still have a good accuracy even under a low range utilisation.

Nice to see the meter works fine in any range.	These are the voltages applied to the input meter	Voltages applied on the output meter

Measurement of the 3 volts range. Good result under any applied voltage	Voltages applied to the input meter	Voltages applied to the output meter

Shows the results of the 30mV measurement range	Voltages applied to the input meter	Voltages applied to the output meter

What will happens using the 30V range together with the 300mV range?

can be foreseen - very similar results. So far however yet tested, does not need a frequency-compensated precision voltage divider plus GPIB capable generator for higher constant tensions plus much time. Since I can predict the behavior for such a combination of measured the so far, I would like to do without such a measurement, even if such a combination for audio is meaningful applications. The accuracy reserves are still very well enough for Hifi applications.