[Elecraft] K3 - PSK31 - AGC/No AGC

[Don Wilhelm]

David,

I do not understand your logic.
Let's divorce the AGC from the product detector for a moment.  The K2 
takes the input to the AGC circuit from the IF and converts it to a 
frequency that is removed from the IF signal.  The AGC circuit produces 
a DC voltage that is used to control the gain of the IF amplifier - period.
That has nothing to do with balanced or unbalanced - the output is a a 
DC voltage which is applied to the IF amplifier to control its gain.

That means that the only signals applied to the product detector are the 
incoming signal (already controlled by the AGC) and the BFO.

Yes, receivers of old did not use product detectors (mixers) but used a 
diode detector, and the BFO signal was injected along with the IF signal 
to produce the audio output.  The BFO injection level was fixed, and 
therefore the gain of the IF input to the detector had to be matched to 
the BFO level for proper demodulation to occur.  The result is that one 
had to ride the RF Gain to make the levels of the IF signal and the BFO 
work in harmony with each other.  In addition, the AGC was developed 
after the detector, so in those old receivers, it was best to turn off 
the AGC.

That is not true for receivers using a product detector.  BFO pickup is 
not a problem with a product detector as long as the output of the mixer 
(product detector) is restricted to a bandpass that contains only the 
audio frequencies (the input frequencies - BFO and IF - are rejected 
leaving only the audio content).

Those old operating habits (ride the RF Gain and set the AF Gain at max) 
are still being used by some operators, but it is not necessary for 
modern receivers which use product detectors - the K2 is included in 
that list.

73,
Don W3FPR

On 10/28/2015 7:33 PM, David Woolley wrote:
> Although I hadn't noticed it before, I would say that design in the K2 
> is more about stability and DC offsets than anything to do with pick 
> up of the BFO.  The balanced demodulator and last two crystals deal 
> with that.
>
> The old, basically AM sets, used high level, unbalanced detectors.  
> That required more total gain from aerial to detector, although that 
> might be split between RF and two IF frequencies.
>
> Balanced detectors can work with much lower level signals, which is 
> particularly helpful for a single conversion design, like the K2. 
> However, the AGC detector doesn't want to be balanced, and with a low 
> level signal, would have a large DC offset, that might exceed the 
> wanted rectified AGC, therefore additional gain is needed before the 
> AGC detector.  If this were done at the IF frequency, the gain from 
> input to output at one frequency would be much greater, and 
> proportionately less stray coupling would be needed to make the whole 
> system oscillate.  By shifting the frequency, which can be done at low 
> signal levels, the design reduces the gain at any one frequency.
>
> The risk with doing all at one frequency is not pickup of the BFO at 
> the input of the IF amplifier, but pickup of its own output.
>
> If there were leakage of the BFO it would get amplified just as much 
> as if additional gain was tapped off at the same point and amplified 
> at the original frequency.  The design seems to rely on the balance of 
> the detector and the two pole crystal filter, to stop the BFO getting 
> into the AGC path, and not the difference in frequency.
>