Ago Control Line

Fig. 3. Block diagram of two stage if strip (typical).

The amplifier blocks in Fig. 3 may be occupied by tubes, transistors, or any other active device (such as field-effect transistors, etc,} which can provide gain. In general, the amplifiers provide the gain while the transformers provide the selectivity.

Normally, most receivers incorporate automatic gain control for AM signals at least, Tliis is a feedback action which employs a dc voltage developed in the detector circuit to control the gain of the if strip v and frequently the gain of the rf stages as well. The result is that the receiver operates at maximum gain with weak signals (which develop little AGC voltage), and gain is automatically reduced by stronger signals.

ACC circuits are many and varied; we'll look at the means by which the AGC voltage is developed a little later, and that's where most of the variety occurs. In the if sti ip, the AGC acts only to reduce amplifier gain.

With tube-type circuits this is most often accomplished by use of "variable-mu" tubes, A variable-mu tubes grid is built with a graded grid-wire spacing. Some of the grid wires are close together while others are farther apart from each other. This permits a wider range of gain-control bias voltage to be applied, since the close-spaced parts of the grid reach Cutoff with low bias values but a large bias voltage is necessary to cut off the widely spaced portions. Where a "sharp cutoff" tube may go from full conduction to full cutoff with a 10-volt swing of grid voltage, a "variable-mu" (sometimes called "remote cutoff") tube may require a 50-volt or greater swing. The variable-mu tube also acts as a linear amplifier over a greater range of control voltages. Thus, this tube acts as an electronically controlled gain controller.

Although Fis, 3 shows the AGC voltage applied to both amplifiers, in many receivers only a portion (or even none at all) of the AGC voltage is applied to the final if amplifier stage. The purpose of this is to concentrate the control in the earlier stages and thus prevent any possible signal overload within the if strip. Having the final if tube operate at maximum gain at all times assures that even signals which are only moderate in strength will develop adequate control voltage in the AGC circuit.

When AGC is applied to the rf stages, it is often left off of the first rf or "antenna" stage (or only partially applied, through a voltage divider network). The reason for this is a bit different. For maximum sensitivity the first rf stage should always operate at maximum gain. Any reduction of gain in this stage reduces overall receiver sensitivity. Sometimes, though, partial app" cation of AGC to the first stage is necessary in order to provide adequate control range.

While Fig, 3 shows transformers as the coupling elements between input, amplifier stages, and output, a number of other devices may also be encountered. Transformers are, however, i:ie most frequently used. They serve two purposes. The most obvious is that of coupling two stages together. The more important, though, from a performance standpoint, is to provide the major part of the receivers selectivity. In fact, some designs have used cascaded transformers, with 2 to 4 transformers between each amplifier sLage, in order to increase selectivity; The number and quality of transformers used is affected by many factors. If, for instance, an auxiliary filter is used, fewer transformers are required since the filter provides selectivity enough, alone.

Why Filter? In examining the previous question, we noted that the transformers in the if strip (Fig. 3) provide the major part of a receiver s selectivity. If this is the case, why should any receiver require a filter?

In the first place, a receiver actually is a filter of sorts if it selects only the desired signal from the mass of signals available to it.

And the transformers in an ordinary if strip permit some frequencies to pass while rejecting others, which is the function of any filter. This is accomplished by means of the selectivity inherent in ¿my tuned circuit. Each transformer contains from I to 3 tuned circuits (depending upon the transformer design) and most contain 2.

The total number of tuned circuits in the if strip thus depends upon the number of transformers. But the effect of these tuned circuits increases exponentially. That is, if a single transformer has a certain amount of selectivity, adding another transformer to double the number of tuned circuits will roughly double the selectivity. Adding just one more transformer will approximately double it again, so that 3 transformers provide 4 times the selectivity of 1. This is the reason that receivers using several transformers between each // amplifier stage have been designed; the transformers improve selectivity, but the gain obtainable by adding more tubes just wasn't needed.

With any practical number of transformers, though, you eventually reach a limit of selectivity. And unfortunately, at the if frequencies most often used, this limit is still a bit broad for todays crowded spectrum—and impossible for SSB,

The selectivity provided by transformers depends upon the frequency-rejection characteristics of their tuned circuits. These, in turn, are composed of inductors and capacitors, It happens that inductors and capacitors can be combined in other types of circuits—and these circuits are of the kind generllv known as filters.

In the transformer, normally only two tuned circuits are involved, and each of these affects the operation of the other, Since the transformer also performs the function of coupling two stages together, the circuits are also affected by the characteristics of the stage supplying the signal and the loading of the stage to which the signal s supplied.

In the filter, however, the only function to be performed is that of frequency selection. T le design of a filter circuit involves choice ol inductor and capacitor values so that no one component has unwanted effects upon the total filter operation. Since the functions of frequency selection and oi stage coupling are separated, the selectivity of a filter can be made much greater than that of a simple transformer.

However, when a filter is composed of only inductors and capacitors, a limit is still reached; to achieve still greater selectivity, perfect resistors and capacitors would be necessary, and these don't exist.

TYPICAL IF XFMft

TYPICAL IF XFMft

Fig, L Effect of filter on improving if selectivity,

This is the point at which crystal and "mechanical filters enter the picture. Actually all non-electrical filters which are used »*

to provide selectivity are mechanical in then-operation, but the term "mechanical filter" has come to mean a specific t>pe of filter which depends upon characteristics of a machined rod.

A crystal filter achieves performance superior to that of an L-C filter for precisely the same reason that a crystal oscillator is more stable than a YFO—its characteristics can be more precisely controlled.

A quartz crystal, if ground to some rather critical dimensions and in the proper shape, is mechanically resonant at several specific frequencies, and is also mechanicHy "anti-resonant" at at least one other frequency which is usually very close to one ol the resonant frequencies, At the resonant frequencies, any vibration applied to one side (if the crystal can pass through almost unchecked. At the an ti-resonant frequency, the crystal oilers a very high impedance to any signal transfer through it. This characteristic is similar to the series and parallel resonances of a tuned circuit, but since it is mechanical rather than electrical in origin the effect is much more pronounced—in radio terms, the Q is much higher.

A single quartz crystal can serve as a filter for any of its resonant frequencies if electrodes are held in contact with its surface so that the vibration can be induced electrically and can in turn reconvert the signal to electrical form. Such a filter has been a key part of CW receivers for many years, but its selectivity is far too great for most phone use. A good single-crystal filter can have a passband as narrow as 50 cycles at the halfpower points, and only a few hundred cycles wide at the skirts of the response curve.

today's crystal filters used in SSB generators and selectable sideband receivers, though, use several crystals and match the resonant frequencies of some with the anti-resonant frequenceis of others to produce selectivity curves which are like that shown jV

in Fig. 4. The passband (top) is essentially fiat, so that the desired signal all comes through, while the rejection cuts in almost instantly at the edge of the passband to provide extreme rejection of all except the desired signal

By contrast, the typical transformer response is many tunes as wide at the skirt

(base) of the response curve as it is at the top.

The performance of any filter is measured by its "shape factor". Shape factor is the ratio of the bandwidth which the filter will pass at two specified power levels, Unless both power levels are specified, the shape factor has no meaning. General usage appears to be that t;ie 6-db and 60-db passband widths are compared, but some authorities prefer the 3-db points for the upper level.

What this means is simply that the bandwidth a filter will pass depends upon how much rejection you require. If you consider any rejection enough, then even a 10% drop in signal level would be "rejection" of the signal. This, though, would permit an unwanted signal which was only 10% stronger than the desired signal to come through with the same strength.

The "passband" we usually think of at the top of a filter's curve is that from the 6-db points. These are the two points, one at either either side of the filters center frequency, at which power transfer through the filter has dropped to Va of its maximum level. A 3-db reduction in power is the smallest we can detect by ear; a 6-db reduction amounts to cutting the signal voltage in half.

However, a 6-db rejection of a signal which is producing twice as much signal voltage as our desired signal will only make the un-desired signal the same strength as the one we want* This shows that, in order to accomplish effective rejection" of undesired signals, we must have much more than 6 db rejection*

We must, in fact, have whatever rejection it takes to first cut the undesired signal down to the same level as the desired one, and then push that undesired signal on down to a level however far we like below the desired one

In practice, a thousand-to-one ratio is usually considered good enough. That is, an undesired signal is effectively rejected if it can be reduced to one-one-thousandth the level of the desired one. And it's seldom that an undesired signal will be more than 1000 times as strong as the signal we want in the first place.

If the undesired signal is 1000 times as strong as the one we want, and we want to cut it down to 1M000 the strength of our desired signal, then the total reduction we must make in its strength is one million times. That's 1000 times to get it down to the same level, and another 1000 times to get it only a thousandth as strong as that levelIn terms of decibels> that's 60 db of power loss. W hile in many cases far less than 60 db of rejection is necessary, this represents a severe case which is still quite possible to meet in practice, and so the general usuage calls for use of the 60-db point as the wider passband in calculation of shape factor,

A typical shape factor for a receiver without a filter (except for its transformers) might be 50. This would mean that the 60-db band with is 50 times as wide as the 6-db figure. If "bandwidth" is quoted as 10 kHz at 6 db, then a band 500 kHz wide would have less than 60 db rejection, lie 60-db rejection would not be present within this much larger passband; you would naturally consider such selectivity poor since strong signals several hundred kHz away from the weak one you Ye trying to get would appear stronger than the weak one.

Extremely good filters, however, such as the mechanical types, may have shape factors as good as 1,8. If the 6-db passband of such a filter is 10 kHz, then the passband at the 60-db rejection points would be only lcS kHz. This means that an interferring signal would have to be within 8 kHz of the desired signa! in order to escape the full refection, While the 6-db passbands in both cases were the same, the effect in operation is quite different. This is why filter effectiveness is measured by "shape factor". Most good filters have a shape factor smaller than 3; no device can have a shape factor smaller than I, and even 1 itself is impossible in practice,

How Does The Detector Section Work? Ilie purpose of the entire receiver up to the detector section is merely to select a single signal out of all that are floating round and amplify it up to usuable strength for the detector. The detector itself does the major job, of converting rf to audio energy.

As we've already seen^ in simple receivers only the detector section is present.

The detector itself operates upon a mixing principle, A product detector requires signals from a beat-frequency oscillator to mix with the if strip's output, while a diode detector for AM uses the carrier portion of the incoming signal to mix with the side bands. We've already examined this process in other parts of this series. What we'll examine now, then, are the other functions accomplished in this part of a typical super-het

WeVe already met AGC in the if strip and learned that a control voltage determines the gain of the amplifiers. This control voltage is developed in the detector section.

Communications receivers must frequently operate in the presence of strong impluse "noise" such as that produced by automobile ignition systems. This noise can be eliminated or at least greatly reduced, and ibis action too is done in the detector section.

Let's look at AGC first. And to keep things simple, let's restrict the operation to AM with a diode detector. After we see how the svstem works, then we'll examine AGC and SSB/CW.

When were receiving AM the incoming signal has a carrier which was essentially of constant strength when it left the transmitter. Fading and other transmission effects may cause its strength to change en route to the receiver, but we know that it was originally constant over a period of several tenths of a second.

It* the process of mixing the carrier and its sidebands to recover the audio information, the detector produces a dc voltage which is proportional to the strength of the carrier. We can pass this dc voltage (which has the audio superimposed upon it) through a low-pass lilter to eliminate the audio, and we have a voltage which indicates the signal strength.

We could apply this voltage directly to the if and rf amplifiers as AGC control voltage, and some receiver designs do, "he disadvantage of this approach is that any signal will then cut back the gain of the amplifier stages, even if the signal is almost unread-ably weak.

To avoid this disadvantage, most designs used "delayed AGC. The delay is not in time, but in voltage; until the voltage produced by the detector is greater than some "threshold" level, no AGC control voltage is applied and the amplifier stages run wide open. After a signal is strong enough to produce a detector voltage above the threshold point, AGC is applied to reduce gain if you recall the exploration of "feedback" we made a couple of installments back, you may recognize our old friend feedback at work again here. An AGC system is. in effect, an amplifier with negative feedback put into it in such a manner that the gain of the amplifier is reduced by strong signals, but signal voltages themselves are kept out of the feedback path,

Notice that the only function of the dc voltage coming out of the AM detector is to give us a feedback signal which is proportional to the actual rf signal strength. If we can get such a feedback signal from any other source, the AM detector isn't necessary. When we're receiving CW or SSB, neither of which uses the AM detector, we obtain the feedback signal by taking a part of the audio itself and recti vying it. After rectifying it, we filter oil the remaining audio hash, and presto! we have our desired AGC control signal,

When receiving SSB, or CW, we add a few extras to the system. For instance, if signal strength increases suddenly we want the AGC system to react rapidly. This happens at the beginning of each "dit^ or "dah during CW reception, or with each syllable when listening to SSB,

When no signal is coming in, we want gain to be maximum. However we don't want the gain to go up rapidly between "dits" or "dahs" in CW, or between syllables in SSB? because if it did we would be listening to a signal apparently buried in noise. The noise would be the normal background, but the receiver would have much more gain then than when signals were present.

For this reason, we want the gain to decrease rapidly when a signal appears, but to increase relatively slowly when the signal goes away. This will hold down background mud while preventing unpleasant "thumps" and bangs".

AGC systems for use with CW and SSB include this "fast attack slow decay" chrac-teristic as a part of their design. It is normally switched out when AM operation is chosen. Fig, 5 shows one way of achieving this kind of action.

What about noise limiting?

Like AGC, noise limiters operate with a control signal which is usually derived from the source as the AGC control signal. This signal indicates average level at any instant.

Any modulation or audio on the signal is in the form of variations above and below the average level, but these are limited in

Was this article helpful?

0 0

Post a comment