[Elecraft] K3 RF gain cal questions

[Wayne Burdick]

Chuck Guenther wrote:

> Is there any problem using the Rx input jack for the RF gain cal....

No. Works equally well.

> Suppose that  I do an RF gain cal, then change the filter
> gain (in the Crystal filter configuration part of the K3 utility) of  
> the
> roofing filter I used during the gain cal?  Should I repeat the RF  
> gain cal?

Yes, if you want it to be as accurate as possible. But typical per- 
filter gain settings are in the low dB range (or should be), so a  
small change won't noticeably affect S-meter accuracy.

> What exactly is "calibrated" during an RF gain cal?

The K3 uses a low-noise JFET IF amp. Gain control of this stage --  
what we refer to as "hardware AGC" -- is achieved in two ways: by  
varying the current through the preceding PIN-diode attenuator, and by  
varying the JFET source voltage (its gate is at ground, so this varies  
the transconductance). A very wide range of IF gain control is  
achieved in this manner. But the gain curve varies a bit from one  
radio the next, because JFET pinch-off voltage varies (the voltage  
differential between gate and source at which drain-source current  
ceases to flow).

The RF gain calibration procedure injects a DC voltage at the hardware  
AGC control node, sweeping it slowly over a 0 to 3 V range using a D- 
to-A converter. The DSP measures the resulting signal level (from your  
XG2), building a table of control voltages vs. 1 dB attenuation steps.  
This table can later be used by the DSP to determine the strength of  
incoming signals that activate hardware AGC; it simply monitors the  
AGC detector voltage at the control node.

S-meter readings are a composite of the detector-voltage-to-dB table  
value and the DSP's internal representation of signals below the  
hardware AGC activation level (also in dB), so the S-meter is now  
calibrated. The RF GAIN pot, when rotated counterclockwise, simply  
adds a DC voltage at the control node (via the DAC mentioned above),  
so it is now calibrated as well, given that the DSP knows exactly what  
voltage to inject to achieve the desired attenuation.

Wayne
N6KR