This article is reprinted from the July 1997 issue of the EDA digest.
Ground Noise
ABSTRACT/SUMMARY
Much energy is often expended to decouple power supplies to reduce power supply interference in equipment while the noise induced in the circuit grounds, which is often the major contributor to interference is usually overlooked. This article examines an example of ground induced noise.
In the October 1995 issue of this digest, I attempted to outline the scope of electro-magnetic interference and compatibility. In subsequent articles I discussed specific design issues related to reducing electro-magnetic emanations. In this article, I will discuss one of the most pervasive problems existing within electronic equipment. The problem is ground noise and it has very serious affects on equipment performance.
To demonstrate the ground noise affect, I designed a simple group of two circuits which share the same 15 Volt power supply. One circuit is an audio amplifier with a gain of 100 utilizing a 741 operational amplifier. The other circuit is a switching amplifier utilizing a MOSFET and a 10W load. A circuit diagram of the circuit is found in Figure 1.
The circuit of Figure 1 was then SPICE simulated. The switching amplifier was driven by a 10 Volt square wave at 1 KHz. While the audio amplifier was driven by a 10 mVolt sine wave a 1 KHz. Figures 2 and 3 illustrate the results of the simulation.
 Figure 1 In Figure 2, the output of the audio amplifier, remains a sine wave with a magnitude of approximately 1 Volt p-p reflecting the amplifier gain of 100.
 Figure 2 In Figure 3, the output Voltage of the switching amplifier, is a 15 Volt square wave. The peak current drawn from the power supply by that square wave is 1.5 Amperes. Since an idealized connection scheme especially related to the power source and grounds is normally utilized in a schematic and in a spice simulation, the 1.5 Ampere peak current drawn from the power supply has no affect on the output waveshape of the audio amplifier.
 Figure 3
But as we all know, idealized zero impedance ground and power supply wiring is not a part of nature. For that reason, R7, .01W, was added to the ground return of the circuit while maintaining all other circuit characteristics. The resulting circuit is found in Figure 4.
 Figure 4 The modified circuit was then simulated and the results plotted after waiting for 100 periods of the switching waveform to eliminate initialization transients. The results of the simulation are illustrated in Figure 5.
 Figure 5
Figure 5 illustrates a major departure from Figure 2. In Figure 5, we observe the original 10 KHz sine wave of approximately 1 Volt amplitude. However, added to that waveform is a 1 KHz drooping square wave. The drooping square wave is also approximately 1 Volt p-p. Therefore, the interference noise, derived from the square wave, is equal in magnitude to the desired output signal, the sine wave. This occurred because only 10mW of resistance was added to the ground lead. 10 mW is the resistance of approximately one foot of 20 gauge wire.
One might ask, "Why not use heavier wire?" If we use one foot of 10 gauge wire, its resistance will be approximately 1 m
 Figure 6 In Figure 6, the square wave is not as pronounced as in Figure 5. But it is still present. In a typical audio system, it would cause a very disturbing hum rendering that system unusable. Of course, we could use even heavier wire with better results. But remember that the connections to the wires have some resistance. Also, there might be no room for such heavy wire and its material and fabrication costs would most likely be prohibitive. So how does one design a product containing several circuits with differing requirements operating off the same power supply and exhibiting little of no interference between them? The answer is very carefully. During the initial design stage of an equipment, the noise potential problems between the constituent components must be assessed and appropriate design actions taken. Otherwise, remedial actions, which are often expensive, will be required.
One can not guarantee that all compatibility problems will be avoided by utilizing the recommended design practice. However, a well thought out EMC design rarely needs more than a fine tuning to meet a specific requirement.
Leonard Zuckerman designs electronic, electro-optical and electro-mechanical products for automotive, industrial, consumer, government and space applications. He draws on 30 years of engineering experience and holds a BEE degree from CUNY, City College. Len can be reached at (631) 673-3881.
EDA has continuously published the EDA digest, a quarterly minitechnical journal since July, 1983. EDA maintains Copyrights to all articles from the EDA digest. No part of the EDA digest can be reproduced without written approval.  
Electronic Development Associates, Inc.
All rights reserved. |
