agc  AGC Circuit Design & Performance  

The difference between a proper AGC system and a poorly designed one is dramatic. A good AGC will not cause operator fatigue. Nor will it cause the operator to miss weak stations. There will be no need to ride any gain controls. The whole idea of Automatic Gain Control is to keep the receiver AF output constant, regardless of RF input levels. This should include very weak as well as very strong signals. Unfortunately, this is not always the case!

I recently saw a YouTube video where the poster compared a friend's Kenwood TS590S to his own ICOM IC-7600. I happen to own both of these fine transceivers. Needless to say, I was very interested in his test results. The comparison was done on a quiet band. It was 6m if I recall correctly.  However...

He erroneously concluded the TS-590 had a noisier receiver. In fact, it actually had a lower AGC threshold. He had to increase the AF GAIN of his ICOM to hear weak signals on that quiet band. The sudden appearance of a signal above that AGC threshold level could have caused him some discomfort, not to mention awakening anyone fast asleep in the house!

I compared my own radios and confirmed the TS-590S did indeed have a lower AGC threshold than the prettier and more costly IC-7600.

I then began to wonder if this new ham knew what the ATTENUATOR and RF GAIN controls were meant to do. ... but I digress!

 Desirable AGC Attributes

Let's consider the AGC characteristics of a good SSB/CW communication grade receiver. (product detector enabled)

  • Low Threshold    The AGC system should be effective on both strong and weak signals, as mentioned above. There should be no reason whatsoever to ride the AF or RF gain controls whenever the AGC system is enabled. These adjustments should be a set and forget affair. A low AGC threshold becomes even more important when using narrow bandwidths on quiet bands that hear less atmospheric noise.
  • Fast Attack    The gain of the IF amplifier must respond quickly to avoid being blasted by the sudden appearance of a strong signal. This means the AGC timing capacitor should be charged by a low impedance source and the AGC loop should be compensated so these fast changes are applied to controlled stages without delay.
  • But Not Too Fast    You want only legitimate signals to act on the AGC system. If the designer properly implements the previous step, the AGC response will be too fast and respond to non-repetitive noise, such as a spike caused by someone turning on a lamp or a thermostat kicking in. This is a good problem with an easy fix! ...a judiciously selected resistor between the Low-Z charge source and the timing cap to increase the charge time to about one millisecond.
  • Headroom    For that one millisecond duration, signals can become quite large and the receiver's IF and AF amplifiers need to remain linear and not clip. The human ear can detect fatiguing distortion products long before clean overshoot. I suspect some radio designers test AGC circuit performance by using only the steady carrier from a signal generator.
  • Decay Time    The AGC level on AM is determined by the carrier strength of a received signal. With SSB or CW there is no carrier. So the average AGC level must be determined by momentary peak level of the received signal. In order to avoid an unpleasant inrush of noise between syllables or when the key is up, the AGC circuit must maintain gain levels during these no-signal moments.

 How I Set The AGC Attack Time On Homebrew Gear

You will need to increase the attack time if the S-meter flies upscale and the receiver goes quiet until the AGC recovers. This fast attack condition must exist to some extent before it can be reduced to the proper level. You will have the right timing when a moderate noise spike causes AGC action which makes the S-meter kick upscale and the background noise to diminish but remain barely audible. You don't want to miss a weak signal when a noise spike appears! This will correspond to approximately a one millisecond attack time.

You can set the attack time by inserting a resistor between the Low-Z charge source and the AGC decay timing capacitor. Over the years I've learned the best way to optimize the resistance value is to experiment, using a variable resistance and a spike source.

So how does one create the noise spike? I use a Weller soldering gun. It produces a strong noise spike whose level can be easily controlled. To generate the spike, simply pull the trigger. Then release so the tip stays cool! The noise spike can be coupled to the RX/IF amp via proximity to a short length of wire connected to it's input. The strength of the spike can be adjusted by changing the distance between the wire and soldering gun.  Easy Peasy!

This "high tech" methodology was learned when I was young with little disposable income. The only instrumentation I had was a Heathkit VTVM along with a surprisingly versatile and useful Barker & Williamson grid dip meter. Later, I acquired a military surplus CRT to build a simple scope for monitoring the output of my homebrew transmitters. This was when I learned about the joy of retrace blanking. ...once again I digress!

The last item to be discussed on this topic is about how much coupling should exist between the test spike source and the AGC controlled UUT. Simple. How strong is the spiky noise at your QTH? Use that much coupling and no more! I once had a homebrew receiver for 80m with a nicely optimized AGC attack time. Later, I installed central air conditioning that produced a very substantial spike on 80m when starting up. I was forced to lengthen the receiver's AGC attack time from its previously optimized value. Remember, faster is usually better!

Note To Self: For a future receiver project, consider a front panel control labelled AGC ATTACK TIME. The idea is to use the fastest attack time permitted by conditions existing at any moment. Start with an adjustment range of 400 microseconds to 2.5 milliseconds.

  •   Does anyone know if this has ever been done before?

 AGC Decay Time

The AGC decay time of a good SSB/CW communications receiver can be set to vary between 1 ms for signals with very rapid fade to four seconds or more for slowly fading signals. For example, fluttery multi-hop over the pole DX could benefit from a fast decay. A leisurely 75m roundtable SSB QSO or slow fade 10m signal would call for a slow decay. Ideally, the decay rate should be adjusted to match the rate of QSB. This will keep the recovered audio constant while suppressing fatiguing background noise as much as possible.

For receivers with only one decay time option, 200-500 ms has always worked reasonably well for me, even on multi-hop DX signals.

Observation: The continuously variable AGC decay time used on Kenwood's TS-870S is utterly brilliant!

 Audio Derived AGC    May 12, 2016

Don't bother! There's a lot of effort required to achieve second rate performance. Let me explain...

I once built a 5 watt 20m CW transceiver whose receiver had an audio derived AGC system. I actually built the transceiver to try the newly introduced VN88 fast switching FET in single-ended class C RF amplifier service. (There was no such thing as an RF power FET at the time.) They worked well as long as I kept the drive down to avoid the eventual disaster caused by turning on the gate protection diodes fourteen million times per second! ...yet again, I digress!

As long as I was experimenting, I decided to try my hand at audio-derived AGC. I was able to see "almost acceptable" performance by using full wave rectification of the AF signal followed by a "hang" timer, which was then followed by a traditional slow discharge type of AGC recovery. The idea was to avoid charging the timing cap from a full gain, no signal condition whenever possible!

Increasing the low-end AF rolloff from 300Hz to 500Hz helped. This change reduced the low frequency thud when a strong signal appeared. Using a small speaker also helped out a bit.

There's yet another serious downside to audio derived AGC. It's important to move the BFO frequency so there is no zero-beat signal coming from the IF. An audio derived AGC system will not respond to a strong carrier tuned to zero beat. This can result in saturating the IF amplifier. You need to ensure zero beat is well down the IF filter skirt, resulting in the characteristic lack of low-mid frequencies in the recovered audio of AF AGC receivers.

The fastest attainable attack time was still a bit too slow. The attack time worsens for those like me who prefer a lower pitched CW note. There were so many design tradeoffs, I concluded it wasn't worth the effort. That was my first and last attempt at 100% audio derived AGC.

I worked 37 countries that summer with the VN88 rig using a ground mounted vertical with 33 radials. BTW, the antenna was made from half of the driven element of K9QVB's TA-33 destroyed by a tornado and later found a block from John's QTH. That autumn, I stripped the rig down for its Collins mechanical filter & some parts for another project. The tri-band vertical remained in use for years!

 Hybrid AGC Possibilities    May 5, 2016

Often, the best performance of a superhet communication receiver is attained using a single conversion scheme with local oscillator pre-mixing or a DDS LO. This scheme has one drawback, however. The amount of usable IF gain must be limited in order to avoid the inevitable BFO leakage into the IF amplifier being amplified to the AGC threshold. Good shielding helps a lot but will not usually take you 100% there on quiet bands.

This was the case when I used the Drake TR7/R7 combo. There was plenty of IF gain in the TR7, but it couldn't be used without the BFO exceeding the AGC threshold. By contrast, the R7 was a dream because it had copious amounts of IF gain at 5.645MHz with the BFO signal at 50KHz. I could use the 300Hz CW filter on 10m with a dipole and the atmospheric noise would be above the AGC threshold. (RF preamp engaged)  Sweet!

However, the Drake R7 was a triple conversion receiver. So how does one get this kind of performance with a single conversion receiver on the higher HF bands where the atmospheric noise is low?  Here's a thought...
This Norwegian-American had an idea pop into his head:     

A single conversion receiver using both AF and RF derived AGC. The main AGC would be the superior RF AGC, but with a somewhat higher than normal threshold. The first 10db or so of gain reduction would come from an audio derived AGC having both a very desirable low threshold and immunity to any BFO leakage!

  •    Does anyone know if this has been tried before?

 Return to Ham Radio Page
 Return to George's Place
    May 12, 2016

 ☚  Let's talk about Ham Radio!