Riding the RF Gain control – part 3

This article continues on from Riding the RF Gain control – part 2 and explores the operating advice when applied to the next generation of receivers.

Conventional superheterodyne communications receiver with DSP demodulation.

The next generation of receivers was a conventional superheterodyne with a DSP based demodulation stage (initially at quite low Intermediate Frequency to suit the power of the available DSP chips).

Communications receivers were enhanced by replacement of the demodulators with a DSP performing demodulation digitally. The DSP sampled the IF signal and digitised it, and channel filtering and demodulation was performed ‘mathematically’ using the digital data stream as input.

There are two significant differences with this change:

  • receiver bandwidth can be determined by digitally synthesised passband filters in the DSP; and
  • first step in the DSP process is conversion of the IF signal to a digital stream in an analogue to digital converter (ADC).

Critically, the Analogue to Digital Converter (ADC) had an overflow point, and overflow of the ADC creates serious IMD and major degradation of received signal, overflow has to be prevented at all cost. To limit the power delivered to the ADC, a narrow ‘roofing’ filter usually preceded it, and the channel filter was digitally synthesised.

So, the challenge for the AGC system was now to prevent ADC overflow, and that was easily enough done by deriving AGC from the digital output of the ADC. In simple implementations, if the roofing filter was significantly wider than the channel filter (eg 8kHz roofing filter, and 500Hz channel filter for CW), signals within the roofing filter but outside the channel filter contributed to AGC action and so there was an undesirable outcome that signals near to but outside the channel would cause pumping of the AGC. This problem was addressed by using two AGC loops, one to prevent ADC overflow, and one to level the channel audio output.

Being digitally derived, the AGC could have near zero attack time, which meant that it desensitised the receiver on impulse noise that might be of such character as to not be a problem if it was ignored as in early generation receivers. Receivers that suffered this problem (eg IC-7000) were just poor designs, not a failure of the concept of AGC. In the IC-7000, a circumvention is to run the Noise Blanker whenever there is impulse noise present, it effectively removes the impulses for AGC detection in that case.

Delayed AGC

Lets look at the Delayed AGC response of the TS2000 with preamp off on 7MHz, noise floor is -126dBm in the receiver being measured (Duffy 2012).

Linearity of PowerOut/PowerIn is very good up to -104dBm, above which the PowerOut is levelled quite effectively.

So, the Delayed AGC system allows the output signal to grow proportionately to input signal building better S/N ratio until the AGC threshold at -104dBm is reached and S/N has built to 22dB.

If you manually reduce RF gain for signals below the AGC threshold, you increase receiver Noise Figure and decrease S/N ratio.

AGC disabled

It is often held that disabling AGC improves ‘sensitivity’.

Lets again use the TS2000 as an example of this type of receiver, and explore linearity of PowerOut/PowerIn with AGC disabled.

Linearity of PowerOut/PowerIn is very good up to -104dBm, above which the PowerOut no longer increases proportionately to input, and is effectively levelled at about -99dBm, levelled by overflow of the ADC at just 5dB above the AGC threshold determined above.

So, the Delayed AGC system prevents the signal causing overflow of the ADC which would occur in this receiver within 5dB of the AGC threshold. Note that the S/N ratio in the gap between AGC threshold and ADC overflow is 22-27dB, these are not ‘weak signals’ in receiver terms.

In this receiver, there is no warning of ADC overflow, just degradation of S/N caused by distortion.

If you disable the AGC, you prevent the automatic control loop from preventing ADC overflow and resultant distortion and S/N degradation. You may also lose S Meter reading, and warning of the overflow condition.

Nevertheless, you will see claims such as:

I run my K3 with the AGC off, and the RF gain WAY down most of the time… Drages signals out of the noise rather well that way. 80 and 40 meters most of the time.

I have not used an Elecraft K3, but I think it unlikely that it is such a poor transceiver that it needs to be operated in this way, nor that there is any real benefit except for extremely strong signals to do this.

Is it any wonder that many if not most good receivers of this type do not incorporate an AGC OFF switch, though that has not stopped hams modifying them to ‘improve’ them with such a ‘feature’.

Next part

In the next part, we jump ahead to the direct sampling SDR receiver.