Info

out

to controlled Va{>£5

Fig. 5. Audio/carrier AVC hookup.

stages to get it in, we find that we must vary the gain of each stage by 114/3 or 38 db. Lets assume that the gain with no control voltage applied is 42 db; this gives us a total if gain of 126 db or approximately 2 million times.

This means our 1-microvolt input signal will come out as approximately 2 volts.

If we increase our input signal to 11.22 microvolts (21 db greater) and leave the gain unchanged it would give us an 1J &-volt output signal to the detector. However, we don't want the signal at the detector to exceed 4 volts (6 db greater than the no-AYC condition) which means the gain must be reduced somewhat If it is reduced to 35 db per stage, a reduction of 7 db, we will get an output of 2% volts.

Now lets hit the amplifier with a full volt input, To have 4 volts output, the total gain must be reduced to 12 db, which is 4 clb per stage.

At this point, let s look at some figures developed along the way. With 2 volts at the detector, our control voltage was zero. With 234 volts out, we had a control voltage reducing gain by 7 db per stage. And with 4 volts out, the control voltage reduced gain by 38 db per stage. Thus in the first Ji volt of control, we were reducing gain by 28 db per volt, while at the other end of the control range we were reducing gain only 19 clb per volt.

And tubes (or transistors either) just don't like it that way. 1 bey tend to change gain at a relatively constant rate, The 6SK7, which used to be standard for IF-amplifier service, had an average gain-change rate of about IK db per volt. At —3 volts, it had maximum gain, while to reduce this gain by 41 db required 29ft more volts of bias.

Since our original specifications allowed us only a 2-volt range from no control to full control, and practical tubes work with a 30-volt (approximately) range, something was obviously wrong with our original specification. The trouble was that we wanted performance too close to perfection.

In practice, we can expect the working range of an AVC system to be about 1/2 to 1/3, per controlled stage. That is, with a single controlled stage the variation in ouptut will range from 1/2 to 1/3 of the input variation (60 db input change results in approximately 20 db change of output; 20 db input change results in about 10 db output change).

If we have two amplifier stages on control, the output variation will be cut in half. With 60 db input change, the output level will change only about 10 db,

With three stages controlled, the variation drops to 1/3. 60 db variation of input gives only about 7 db output change. By increasing the number of stages under control, we can keep reducing the output variation, but a practical limit is reached eventually. Nobody wants to put 100 amplifier stages into a receiver simply to get output constant within 2 db for a 60-db variation in input!

\ rid it probably wouldn't work right anyway.

In case you re interested, Fig. 3 includes the single-stage control graph for the 6SK7 and also for the 6BA6, more representative of the newer tube types now in use,

AVC for overload protection

Which fairly well disposes of point 1. How about point 2—over-load protection?

As mentioned a bit back, to get wide control range the control voltage must be made fairly large, and not a1 tubes can take it. This isn't of importance particularly with a factory* built receiver because we assume the manufacturer's engineers took {his into consideration, but it does stand to warn us not to sub-stitutc IF tubes unless we know exactly what we're doing.

in particular, a sharp-cutoff tube should never be used to replace a remote-cutofF type. The S-meter may read higher on noise, and it may even read a bit higher on some signals, but the chances are rather great that the substitution will play havoc with the AVC action of the set, You may find that the S-rneter just

10 AVC IM FILTER

Fig, 6, Simple delayed AVC.

sits there hovering around 9, with almost no change between a weak signal or a strong one. You may not even hear the weak signals. What happens is that the sharp-cutoff tube amplifies set noise up to the point where the AVC takes hold, then cuts off at that amount of AVC voltage, so that stronger signals don't get by and weaker ones are highly distorted.

Even substituting another remotc-cutoff tube should be done only with care, because t!Li-transfer characteristics (linearity to you side-banders) of the different types at similar bias voltages is quite likely to differ. That 6SK7 may be there for a reason; possibly no other tube can handle the particular level of input signal at high values of AVC voltage.

Norma design procedure to prevent overload is to first select tube types which can handle large values of voltage while having high grid bias applied; for instance, a typical communications receiver may be called upon to furnish a 66-volt swing at the output of the final if stage, wrile the stage is biassed to some 20 volts negative, This is a rough requirement It it can't be met, about the only thing to do is to leave the final stage off the control line, which then restricts the range of control you have available. Already we're compromising.

Next is to estimate the control voltage produced by various input signal levels, and check to see that each stage can handle it's voltage requirement at that level of control voltage. If things are arranged so that the first controlled stage overloads before any of the others, you'! satisfy the overload-protection requirement—but this is seldom practical to do.

Incidentally, while AVC is usually omitted from rf stages to keep from hurting the signal-to-noise ratio, it's about (he only way to prevent mixer overload- A point usually overlooked is that the AVC, if properly designed, won't affect the ri stage gain until the signal is already strong enough that there's no worry about S/N ratio!

In applying the control voltage to the complete amplifier chain, it's frequently helpful to use only part of the control voltage on some stages. For instance, the rf stage may get only 1/3 of the control voltage. If control voltage ranges from 0 to 30 volts, that on the rf stage may range only from 0 to 10 volts. This frequently gives protection for the mixer, while helping make sure S/N ratio isn't reduced until a signal is particularly strong.

Fig, 4 illustrates one way to get partial control, and shows also application of similar partial control to the last if stage to protect it against overload. This can be carried on to the point of applying full control to only the first

if stage, a little less to the second, etc. However, doing it this way will reduce the control range below that indicated by Fig. 3. so the compromise should be made on the basis of what you want more.

Reaction time

And now we're down to point 3, reaction time. This is more frequently referred to as "time constant"- Like control range, it involves many factors.

On one hand, we never want reaction so fast as to wipe any audio off an incoming signal At the other extreme, we don't want reaction so slow that the receiver takes hours to adjust itself. Our desired reaction time is somewhere in between.

To find out what its practical limits are, let's look first at the area of wiping-out-audio. we want to preserve all audio frequencies above 100 cps, the fastest our system can be allowed to react is 1/100 second (l/f)5 or 10 milliseconds. If we don't care about anything below 300 cps, we can reduce reaction time to 3,3 milliseconds. For '^broadcast quality" response to 30 cps, however, reaction time minimum is 33 milliseconds.

What this means is that if our AVC system can reach the desired control voltage in 33 milliseconds when signal increases, or drop back in the same time if signal drops, it will just cancel out a 30-cps tone on the incoming signal. For actual broadcast-receiver use, a time constant of }i to JA second is more commonly used.

Rut we're not so much interested in BC as in communication. Let's see how fast the reaction time must ¡»e for CW, If we intend to operate at speeds up to 50 wpm, this means the spaces between dits will be only about 40 milliseconds wide, if we intend for the receiver to have usable response during this time, the AVC system must react in less than 1/ iO of this time, which gives us a reaction time of 4 milliseconds for high-speed CW,

Since this will also suffice for 300-cps response to audio, it might appear that a 4-ms reaction time for both initial response to increased signal and for drop-off with reduced signal would be good.

But how about SSB^ to this mode, we want to hold average gain constant throughout a word at least—and 4-rns reaction wouldn't do this. To fill this need, circuits known as "hang" AVC circuits have been devised, which charge rapidly but hold the control voltage in place for a while after the signal drops again,

Which brings us to the Slow-Fast-Oil * knot on the modern receiver* Ideally, the complete system for today's use should have a uniformly fast (4-ms or so) charge time, with a choice of either equally rapid release time (fast) or a much slower release, on the order of Ji to 1 second, in practice, many receivers consider 100 milliseconds fast. These can usually be improved by changing resistors or capacitors as necessary to reduce the charging time constant by a factor of 25, but in doing so care must be taken that nothing else is changed. For instance, reducing a series resistor by 25 times might drop voltages throughout the system so much as to make it stop working, and reducing the filter capacitor by 25 times will usually shorten the "slow " time constant as much as it will the "fast/'

However, il the "fast" and £<slow" positions use two different capacitors, the ' fast capacitor can be made smaller as desired. We'll get into this a bit mure down the road a spell, with schematics, but first we have some more points in design to dispose of.

Delayed AVC

Next on the list is that AVC! should not reduce sensitivity. In a simple AVC system such as that of Fig, 2, any input signal at all I no matter how weak ) would result in some control voltage, which would in turn reduce receiver gain. This is not good.

To overcome it, the AVC system is so arranged that no voltage is developed until the input signal exceeds a specified value, This is called "delayed AVC" but contrary to (he implications there is no time delay. The delay is a matter of voltage. A wide variety of circuits has bceit developed to introduce the "offset" action; well look at some a bit farther on*

Distention from AVC

The final point to be considered in designing the perfect AVC system is (hat the AVC should introduce no distortion of its own.

The point was partially considered when examining control range, but there's more to it still. When the AVC runs the amplifiers down near cutoff, a phenomenon known as ^modulation rise" takes place which introduces second-harmonic audio distortion il a modulated signal is being received. While the signal is still quite audible, it sounds quite "mushy/" It your receiver sounds perfect on weak signals but all signals above S9 are a hi I distorted, modulation rise is quite possibly present.

When voltage delay is introduced to satish point 4. then a type of distortion known as "differential distortion' may come along with it This type of distortion results from the AVC coming into action part way up the waveform on a weak AM signal, but in practice lias not been found to be appreciable unless modulation at the 100-percent level is being received. If J00-percent modulation is being detected, similar distortion will also be generated in the detector, so differentail distortion can be considered merely a fine point of AVC engineering.

Sources of AVC voltage

You may note that so far, except for Fig. 2, we haven't mentioned just where in ihe receiver the control voltage is derived. This is because (lie control can be developed from either of two sources, and when all factors are taken into consideration either works equally well (or equally poorly if you prefer the negative connotation).

Most conventional, of course, is the practice of using rectified carrier \oltage as in Fig, 2 for the control, This approach lends itself well to AM use but is a bit stickv tor SSB and CW since the incoming "carrier" is intermittent at best and may not necessarih bear any relation to the actual signal strength- For example, the sentence "his sis is JT consists entirely of dits and spaces, li a stead\ carrier would be coming in at 10 microvolts, the average voltage for the duration of this sentence would be less than 5 microvolts, and the AVC would treat it as a 5-microvolt carrier. On the other hand, the sentence \ om to 0" is all dahs and spaces, so would be closer to 7Ji microvolts average.

To overcome these problems, many designs have taken AVC control voltage from the audio, after detection but before the gain control. With SSB and CW, this gives equally gt x )d results, and in many cases proves superior, since there's no worry about bio energy getting into the AVC line and hurting receiver sensitivity.

However, with AM its again sticky. If all signals were modulated 90 percent or so, an

0 0

Post a comment