Scribd
Upload
87 views5 pages

Distortion Levels in RF Clipping

Distortion Levels in RF Clipping

Uploaded by

Blagoja Gjakovski
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
87 views5 pages

Distortion Levels in RF Clipping

Distortion Levels in RF Clipping

Uploaded by

Blagoja Gjakovski
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 5
(Wireless World, October 1976: Distortion levels in r.f. clipping Theoretical intermodulation distortion levels from clipping a two-tone s.s.b. signal by D. A. Tong, B.Sc. Ph.D Datong Electronics Ltd Speech clipping at r. is now a well known and accepted method of increasing transmitter power. So far little experimental work has been done to determine th i jon distortion gene id by rf. clipping. Thi study attempts to find the magnitude of distortion that can be expected when ideal clipping is applied. The analysis occurs with a two-equal-tone even for small amounts of clipping but, probably due to the waveform, this is not so objectionable when listening to speech. In the use of am. and £m. transmitters af. speech clipping has been widely: tused to raise the average-to-peak amplitude ratio of the speech ‘waveform. Disadvantages of this tech- nigue are a high level of distortion, and the incompatibility with ss.b. transmit ters which are now the most common kind in use. If, however, the clipping is performed, at ron ass. signal most of the distortion is eliminated and there is no incompatibility with ss.b. trans- mitters. Radio frequency clipping by 15 to 20 dB can increase effective peak power by a factor of ten ‘An add-on rf. clipper can be placed in series with the transmitter microphone lead because the device generates its own sb. clips it and demodulates it back to af. A single audio frequency sine wave, when processed in such @ ‘way, is reproduced as a sine wave with, assuming an ideal system, zero distor: tion, no matter what the degree of clipping. If the same signal is subjected todirect clipping it approaches a square wave and a set of odd harmonies is generated. Ifthe wanted frequency is at the low end of the speech band, many of the harmonics are also in the speech bandwidth and cannot be filtered out, ‘When several sine waves are simultan- ously fed into a clipper the situation is ‘more complex because a range of sum and difference products between var- fous harmonics are generated. The calculation which follows was undertaken in an attempt to find out hhow much distortion can be expected with ideal rf. clipping applied to a typical two-tone ss.b. test signal. Although a general calculation for the Jntermodulation produced by simultan- ously clipping sine waves of varying, amplitude would be complex, the spe- cial case of two equal-amplitude sine waves is more easily solved. The analysis shows tliat quite high intermo- ulation distortion occurs, even for ‘small amounts of clipping. This is certainly not a condemnation of rf. clipping for speech transmissions, but a point against in-band distortion, pro- duced from a two-tone test signal, being lused as an important parameter for characterizing a transmitter used for speech communications. Calculation A two-tone test signal can be regarded fas a double sideband suppressed carrier signal produced by muluiplying together ‘two hypothetical signals, The test signal is defined as = Aposot+Accose,, (1) ir} Fig. 1. (a) Modulating waveform, f(t) Whick, when multiplied by cos Ga; +e9) 72) gives the two-tone test sighal ft). (b) Two-tone test signal ft). Note its peak amplitude is twice that of the waveform in (a). The phase of the ‘waveform with period 40/ (sy +6.) is shifted by = radians (180°) in each ‘successive section of the envelope. In both a) and (b) the effect of amplitude limitation is shown by the dotted line ie. the sum of two signals. This can be written in the product form as flO) =2A,c0s (5 +13y/2)t. cos(Aw/2}t (2) ‘where Ma = ,—uy This second equation shows why the peak amplitude of the ‘two-tone signal is twice that of each separate tone. It also shows that the envelope of the rf. component, at ‘angular frequency «-Fs)/2 is amplitude modulated by a low frequency wave. form which is equal to half the differ. ence of the two original tones. The waveforms fit) and 2A,cos (Su/2)t are shown in Fig Because the envelope shape of f(t) is irectly related to that of the modulat- Ing waveform 2A cos Aet/2, any ampli- tude limiting of the composite signal f(t) can be regarded as due to similar limiting of the modulating waveform. ‘This is indicated by the dotted lines in Fig. 1. ‘To predict the effects of rf. clipping fon a two-tone signal, one can first calculate the Fourier components representing a clipped cosine-wave, and then consider the effect of modulating costs; +4,/2)t separately by each of these components. The results of each nrodulation process, Le. each multipli cation, can then be added together to give the final frequency spectrum. ‘Any complex modulating signal {fo(t) with a fundamental frequency of 72 can be represented as the follow: ing general Fourier series ul Paysindut/2+aysin 2Aut/2+, ete + bcos Aut/2+Bysin 2ut/24 ...0te. ‘This expansion simplifies if there is no de. component in the waveform, and the clipping is symmetrical. Under these conditions only the odd cosine terms are left fall) = bycosdunt/2 + bycos3.bw.t/2 + bycosS.Aet/+ ete @ This is why every effort should be made, in any af. or Ff. process to achieve @-sd@ | @;-26@ | atte ato 2 Fig. 2 Frequency components of the clipped waveform in Fig. (2). Note that there is no component with frequency (o; +15) / 2despite the cyclic variation at this frequency in f0) Fig. 3(), Function f(t) used in the calculation of intermodulation levels {full line), Between t, and ty t, and tit isan undistorted section of the waveform Ay C08 Sut. Elsewhere the latter has been attenuated to an extent dependent on the value of a, The full Ao Aline 4 a) a Ages ee titi _j @y428@ | areho ata @,r380 PRESEN line is the waveform expected when feeding a cosine into the circuit of Fig. 3(b) which is an equivalent circuit of a practical diode clipper. Diodes Dy and Dare assumed to have infinite reverse resistance, zero forward resistance, and zero offset voltage. The input signal comes from a source of zero internal resistance and the output is fed into an infinitely high load resistance. The parameter a is equal to Ry/(Ry+Ry. The input Aycos Aut, is ‘an imaginary signal. In @ real situation the clipper would have the same circuit except that the battery voltages would be 2A, ‘Wireless World, October 1978 symmetrical clipping. When each of these modulating components are used separately in equation (2), and the results are added to give the complete ‘expansion for f(), the result is fl) = Abyc0s wt + Abjens wt “+ Abyc08 (oy + Su)E-+ ADCO (AU) “FAbse0s (4) +2.4a)t+ AbscOs (or-25u)t Ady 29608 (6, +RM)EF AB, 526608 (r+ ndoyt ‘The lne spectrum corresponding to this expression is shown in Fig 2. and the peak amplitude of each frequency Component is shown above the corre: sponding spectral line. Before discussing the actual magni- tudes of by, by ete, iis worth making a few points about Fig. 1and 2. The higher, the subscript, i.e. the higher the harmonie of fs() responsible for the partleulor component, the weaker the resulting component. Out of the Unwanted intermodulation products, attention will be concentrated on the components at oj tAu and ode, These are the largest and also the closest in frequency to the. desired Signals and therefore the least likely to ‘be tered out. Fig. 100) is slightly misleading because the dotted lines, which represent the peak-to-peak amplcude after clipping, suggest that the waveform at frequency (0) +63)/2 ‘wll be lipped. In fact if unis waveform fs modulated by the clipped version of fad) shown in Fig. Wa), e. @ cosine Wave of original amplitude A, clipped to A pesk amplitude of, the sine shape of the high frequency waveform will be retsined. This treatment. therefore applies only to the ease where clipping is achieved by an ideal rf. compressor with threshold ata and with along time constant compared to 42/(6,+.3) But negligible compared to. 4n/(o\—u2) Nothing is said about the spectral components which will appear centred fn harmonies of (ci te2)/2, but, because these are always discarded by subsequent fitering, this is not impor- tant. One assumes that for components centred on (6\+0,)/3, there will be repligible difference between true clip: ping and the hypothetical rf. compres Sor mentioned above. The same assumption is also implicit in Kahn's treatment of hard elippers (2) “The magnitudes of coefficients by and b, can bederived using Fourier analysis for the waveform shown by the full line in Fig. S@). Aa ideat limiter or clipper (R in Fig. 36) equal to zer0) allows no increase in input waveform “amplitude after it has reached a certain threshold as shows. by A in Fig. 3a). Practical clippers approach this cond. tion and therefore, the caleulation has been carried out for the general case. The wo results are D ale 0) 1-24 Wireless World, October 1976 Lococo] ai 44, 3/2 Jat 1-29) © where x=A/A, and a is the gain of the clipper, see Fig. (a). It should be noted that in two-tone testing it is normal to ‘measure the amplitude of the strongest intermodulation product relative to the product of the desired component. The intermodulation level in decibels will therefore be equal t0 TP pac ™20 108 gb, /B © ‘where by and by are given by equations (@) and (). In the case of infinite lipping with an ideal clipper, « and ‘A/A,are both zero. Itcan be shown that by becomes 44/32 and b, becomes 4A/x. For infinite clipping therefore, by/bs is 3. The minus sign is neglected because it merely implies a phase shift of radians in the cos 3Aut/2 term relative to the cos Js¢/2 term, This result provides a check on the calculation because the waveform f(t) Js now a symmetrical square wave and the expansion for this is well known (IL Tealso gives a useful result because the worst intermodulation product ccavsed by infinite clipping of atwo-tone signal will be at a level of 9548 relative to one of the wanted signals. In this special case itis easily shown that by is 4A/ms, therefore the level of ‘every intermodulation product can be derived, se Fig. 6 ‘A second check on the results can be obtained by making A=Ayand St=0. It is found that b expected because this is the case for no

You might also like