Distortion Grapher

Circuit enables distortion error to be graphed

Introduction

In other articles, SPICE analysis was used to analyze the shape the distortion error of various amplifiers and circuits. Here I want to suggest a circuit to allow the same analysis to be done to live circuits. The idea is to drive the circuit with a signal generator, attenuate the output of the circuit back to the input level, then subtract the result from the same input signal to return the distortion error curve. The signal would then be graphed on an oscilloscope.

In mathematical terms if a circuit can be defined as having gain (A_V) + error (ε) and ε is scaled as if introduced before the gain.

(1)	V_out = A_V(V_in + ε)

Solving for error gives the equation of a device that will show that error.

(2)

V_out

A_V

–

V_in

The diagram in figure 1 below shows how such a device would be connected to allow the graphing of the distortion error.

Figure 1: Connection of test fixture with other components to graph distortion error.

Circuit Design

Of the possible topologies that in figure 2 uses the least number of operational amplifiers.

Figure 2: Schematic of Distortion grapher test fixture.

Choose operational amplifier. It is desired that the operational amplifier have noise and distortion well below the distortion levels to be graphed. Then it is desired the operational amplifier be a type that has no design peculiarities that might produce unexpected results. My first thought to use the AD797 fits the first qualification but not the second. The OPA227 fits the second specification with only a slightly higher noise level. Use the 4 circuit version: OPA4227. If the AD797 is desired the circuit may have to be modified.

Want R₃ and R₄ to dissipate only ½ Watt when driven at a voltage giving 1000W into 8Ω.

(3)

R₃ =

8Ω × 1000W

½W

= 16kΩ

Because I want a potentiometer for R₇ equal to R₃, round up to 20kΩ.

Define R₅ and R₆ to define some maximum attenuation for a worst case where 1000W is produced from a 250mV input.

(4)

V_@1000W =

2PR

2 × 1000W × 8Ω

= 126.491V

(5)

R₅ =

20kΩ × 250mV/126.491V

= 39.5285Ω

Choose 1% value: R₅ = 39.2Ω.
R₅ and R₆ might be omitted without problems.

Choose 1kΩ for R₁₁ and R₁₂ then calculate R₁₃ and R₁₄ for a voltage gain of 10.

(6)	R₁₃ = 1kΩ × 10 = 10kΩ

Choose 1% value: R₁₃ = 49.9kΩ.

Other components are chosen arbitrarily according to familiar practice. Also for now, it is presumed that the reader knows how to construct a conventional operational amplifier power supply of ±15V.

Figure 3: Parts List
U₁-U₄	OPA4227	R₇	20kΩ dual potentiometer, ½W
D₁-D₆	1N914 or 1N4148	R₉, R₁₀	100Ω
R₁, R₃, R₈	47kΩ	R₁₁, R₁₂	1kΩ
R₂, R₄	20kΩ, 1%, 1W	R₁₃, R₁₄	10kΩ
R₅, R₆	39.2Ω

Optionally small capacitors might be placed in parallel with R₁, R₃, and R₈ if you desire some rf filtering on the inputs.

Operation

Refer to figure 1.
Set text figure potentiometer R₇ to maximum attenuation.
Connect signal generator to input of device under test (DUT) and to its input to test fixture. Connect DUT to test fixture: Non-inverting single ended circuits to v_in+, inverting single ended circuits to v_in–, or bridged outputs to v_in+ and v_in–.
Connect output of text fixture to y input and signal generator to x input of oscilloscope.
Connect all grounds in common.
Set oscilloscope to x-y graph mode and make any initial and continuing adjustments necessary to properly scale the graph.
Adjust sine or triangle wave amplitude of signal generator for desired output from DUT.
Raise R₇ until outputs of DUT and signal generator appear to cancel.
Increase y-signal gain of oscilloscope to show full-screen display of distortion error and adjust R₇ again if needed for better cancellation of the test signal.

Document History
March 5, 2015 Created.