AGC and JT65

[VK4WDM]

I have read in a couple of places that AGC should be turned off for digital modes. I certainly get a higher signal level with it turned off with JT65 but decoding seems to suffer especially with stronger signals. I would be interested to hear what others do: AGC fast, slow, or off?

73

Wayne VK4WDM

[VK4UH]

I turn AGC off in JT65 and FSK441 where this is possible. (Not every radio allow you to do this)
otherwise the signal strength measurements are not consistent and there may be "dead" zones after loud noise peaks etc as the AGC recovers.
Better to reduce RF gain rather than use AGC

Kevin VK4UH

[VK2DVZ]

Using JT65 on 23cm EME I switch the AGC off, as my transceiver has that feature available - much more sensitive on weak signals and gives good decodes.

However on both 2m and 70cm when using digital modes, I select fast AGC due to the fact that on the transceiver used on those bands the AGC is not able to be turned off. It works OK for JT65 on EME and also for FSK on MS or on JT terrestrially.

(On 2m, with Mark VK2AMS running 120w at a distance of about 5km from me, my setup will not decode his signals at all as they are so strong. If we are both TXing at the same time it is not a problem).

[VK2OMD]

I have read in a couple of places that AGC should be turned off for digital modes. I certainly get a higher signal level with it turned off with JT65 but decoding seems to suffer especially with stronger signals. I would be interested to hear what others do: AGC fast, slow, or off?

73

Wayne VK4WDM

AGC in traditional communications receivers is an automatic measure to reduce non-linearity due to overload of later IF stages or the detector itself. Non-linearity is manifest as apparent gain compression and IMD, and is easily demonstrated using steady input signal at a range of levels.

It derives from the AVC concept applied to early broadcast receivers as a measure of producing constant sound 'volume' from an AM signal automatically compensating for the different carrier levels from one station to the next, but implicit in that, that it prevents overload of the receiver stages.

In my experience, non-linearity with AGC off escalalates just a few dB above the would be AGC threshold, evidence of a sound (pardon the pun) receiver design were the Noise Figure of the receiver is maintained at its best with increasing input signal until a good S/N ratio is achieved before gain is reduced (and NF increased) to maintain linearity. In other words, AGC should not degrade S/N on weak signals (this is known as 'delayed AGC').

Above is a plot that shows the onset of non-linearity in an IC7000. The blue line is output power scaled in terms of rx input power from a SSG, and the brown line shows the error or gain compression. The sudden death nature of this receiver is probably attibutable to the DSP detection, overloading the ADC is about as brick wall as you get. Conventional receivers (ie mixer detector) have a slighly softer knee, but the result is much the same. The role of AGC is first and foremost to protect the receiver stages from overload and the consequent non-linearity (gain compression, IMD). Of course, effective AGC operation depends on dynamics that suit the application, the big challenge in migrating from mature AM AVC to effective AGC for SSB, and good receivers provide some adjustment.

A spectrum plot of the receiver output is more revealing than our ears.

I can understand why a receiver manufacturer does not provide an AGC ON/OFF switch, and I can't understand why hams modify radios to add such a thing were it is not provided. The sanest reason for a manufacturer to add such a switch is to bow to market pressure, even if that is not rationally based.

You might argue that a non-linear receiver is not a disadvantage for some modes, but it is always a disadvantage if there is more than one signal within the detector passband.

Now above I showed the steady signal response of an Icom IC-7000. The IC-7000 has a huge AGC defect, is has an attack response that is way to short, it is activated by short impulses that would not significantly degrade receiver output, and in responding to these short impulses that we cannot 'hear' in another receiver, it pumps the AGC down and the listener loses copy of the signal to which they were listening. This anecdote to show that despite very mature knowledge about the behaviour of AGC in a good receiver, a manufacturer can still turn out a produce that would appear to have not been field tested. In this case, the fix is not to turn AGC off (I don't think you can do it without modification), but to use the Noise Blanker which blanks the impulses before they generate AGC control signal.

Noise blankers can be effective in avoiding nuisance activation of AGC, though noise blankers degrade system linearity.

Limiting the bandwidth prior to the detector input helps with the problem of AGC activation by undesired nearby signals.

You may find some situations where manual control of RF gain helps, it injects AGC control signal so it essentially limits how much gain the receiver has, but without limiting the capacity of the AGC to further reduce gain to avoid overload. In some situations where dynamics is the main issue, this can be an assistance.

In my experience, properly adjusted AGC on a good receiver is not an impediment to linearity, it protects linearity.

Owen

[VK2KRR]

Hi Wayne,

Which bands are you using JT65 on? HF or VHF ?

On HF & 6m I usually put AGC onto FAST when using digital modes (other than FSK441). I've done testing using WSPR and find that Fast AGC helps to get rid of all non wanted signals faster, such as lightning static (esp 160m), and other weird squiggly noises and pops associated with 10m mostly. And on 6m Es helps with removing meteor ping signals quicker. Provides a better average signal to noise ratio.

But as with the other comments I usually leave AGC off (or slow) on 2m and above. Though have not done much testing to compare its effect on signals on these bands, but on slow it seems to work fine.

[VK2OMD]

... Though have not done much testing to compare its effect on signals on these bands, but on slow it seems to work fine.

Slow AGC usually refers to slow decay, and it is appropriate to cases where the underlying signal is of varing amplitude... like SSB telephony or A1 CW as it avoids the compressor effect and cyclic pumping of the background noise which is annoying to the ear.

In the commercial world, this problem of SSBSC was often addressed by sending reduced carrier (eg -26dBc) and extracting it to drive a fast response AGC system independently of the energy in the sidebands from moment to moment. Bit too advanced for hams though, but it would probably be pretty easy to implement in DSP mod/demod transceivers.

For signals like nary FSK, using faster decay lets the receiver track changing signal level more closely, delivering better average S/N ratio.

These so called digital modes aren't perhaps the no brainer the WIA president makes out when he puts a case for FLs to have 'digital modes'.

Owen