Regular reports of my grabber activity and that of others, plus information on QRSS software, hardware and technique that comes my way

Monday, August 22, 2011

Measurement of 30m Noise Background at W4HBK

In my last post I described a method I use to disengage the AGC of my TS-440 so that accurate measurements of signal levels can be made.  Last Winter I tried to measure the background noise on my receiver but found that AGC compression was a major problem.  Now that I have that ironed out I've tried again to measure the background noise, this time with success.

My system consists of an Inverted V antenna into a Kenwood TS-440 receiver whose audio output goes to a computer sound card and then the Spectrum Lab software.  All signal levels to which I refer are as measured on the SL display.  I define background noise as the signal level of the noise above the noise floor of my receiver which is determined by replacing the antenna with a 50 Ohm resistor.

I used a little know feature of Spectrum Lab called "Watch Window"   to record my data. WW allows you to select a specific part of the spectrum to be recorded as a strip chart, e.g., amplitude vs time.  I described this in a previous post called "New Toy" back in March.  Several expressions are available to measure both signals and noise.  The noise expression is,

     noise(freq1, freq2)

which measures the noise level between f1 and f2 and eliminates the effect of discreet signals using the following scheme:
   1.  an array of amplitudes from the last FFT calculation is sorted into order of increasing amplitude.
   2.  The amplitude of the lower quartile value is then returned as an estimate of the mean noise level. 

With the frequency range set to cover 10140.0 to 10140.1 and numerous QRSS signals present it ignores them and measures just the magic  You can read about this in more detail in the SL help file.  The only problem I've had is with strong RTTY and JT65 signals which cover the entire QRSS spectrum but this is rare and easily recognized..

I recorded the noise background for a 24 hour period so see how it varied with time of day.  The results are shown in Figure 1.  The spikiness is due to static crashes plus occasional clicks and pops from electrical equipment.  I can hear nothing that sounds like power line noise or crt emission, etc....just the steady crunch of sferics.  Thunderstorm activity was actually mild during the measurement period with just a few small storms no closer than 25 miles, indicating to me that the sferics were propagating in from afar.

Figure 1.  Noise vs Time of Day Measured @ W4HBK
The time of maximum noise was just around sunset at about 35 to 40 dB above the NF.  Minimum noise occurred around noon at 20 to 25 dB above the NF.

For comparison I obtained the noise estimate from VOCAP, Figure 2.  The trend for time of day is close but I measured a max to min range about twice as large as VOCAP, 15 dB vs 7 dB, respectively.  My measurement is for a particular day whereas the VOCAP estimates are based on extensive observations used to determine averages.
Figure2.  VOCAP Prediction of Noise @ Receiver
I plan to repeat the measurement several times to better characterize this time of year and then do the same for other seasons.  Can't wait for a cold Winter night to see how low we can go.

Tuesday, August 16, 2011

How to turn off your receiver's AGC

Even though the AGC is a wonderful invention there are times when it's nice to be able to turn it off.  In order to make quantitative measurements of signal levels the gain needs to be constant from the antenna input terminals to the audio output terminals.  The "A" in AGC keeps this from happening.

Very few receivers today have an "off" position for the AGC but I've found it is possible to effectively turn it off by simply backing off on the RF gain control.  I'm almost ashamed to say that in all my years hamming I have just discovered this.   Here's how it works.

Figure 1 is the AGC response curve made at Clifton Laboratories for a commercial grade receiver .  With the RF gain full on there is region from the noise floor up to a point called the "AGC knee" where receiver gain is linear.  At the knee the AGC kicks in and maintains a more-or-less constant audio output for an ever increasing RF input.  But, as the RF gain is backed off the knee moves to the right so that the linear region covers a much wider range and that's the fact we take advantage of to defeat the AGC.  If the knee is chosen carefully then all the signals likely to be received will be in the linear region.  It is always possible that a really strong signal will appear inside the bandpass to activate the AGC but for QRSS band conditions this is very unlikely.
Figure 1.  Receiver AGC Response Curve

As the RF gain on my TS-440 is reduced the S-meter moves up from the "peg" at S0 to higher and higher values...the higher it goes the wider the linear range.  I have found that S9 is a good set point.  To compensate for the drop in audio output I increase the audio gain.  Now you might think that in doing this that the SNR is being degraded but that's not the case.  While doing these experiments I keep track of the noise floor as well as the signal and the difference remains the same.  There is eventually a point where the signal is too weak to stay above the noise floor but for my system this is substantially above the S9 set point.

Not all receivers have the same AGC response curve and the linear region may not actually be all that linear.  To check my TS-440 I inserted attenuators of known values at the antenna terminals and looked for the corresponding drop in output as read via Spectrum Lab.  Actually I have only one attenuator I consider calibrated, a Tektronix 011-0059-01 fixed 20 dB type with BNC connectors.  I used this to check the internal attenuator in my TS-440 which is supposed to be 20 dB and checked out within a dB.  Thus I have two accurate 20 dB steps of attenuation for a total of 40 dB.

Figure 2 is the test setup.  For a signal source I used my Palomar R-X Noise Bridge which provides a strong, stable signal for testing.  With the noise source OFF the output as read by SL defines the system noise floor.  With the noise source ON the output was adjusted for a convenient level above the noise floor.  Attenuators were added to vary the input and the output observed on SL.
Figure 2.  Test Setup

Figure 3 shows the effect of the attenuators for two cases.  The first is with the AGC turned off by reducing the RF gain and you can see that the output closely follows the input.   The second run was made with the AGC full on at maximum RF gain where you can see the output hardly follows the input, particularly for the first step.  I did this run for two noise levels corresponding to S4 and S1.  The weaker level came closer to following the attenuation as is suggested in Figure 1. You can see AGC curves for other receivers at the previously cited link to help understand what goes on around the AGC knee  My hypothesis that reducing the RF gain turns the AGC off seems to be valid and the response of my receiving system is nearly linear under these conditions .  Now I can measure signal strengths and compare them in a meaningful way.
Figure 3.  Effect of Attenuators With and Without AGC

The impetus for this study was a desire to measure background noise so I could compare Summer and Winter conditions on the various bands for use with path loss calculations.  For example,  VOACAP predicts the signal level in dBm reaching a receiver from a transmitter for known power and antennas.  The unknown factor is the veiling effect of band noise.  I have begun making background noise measurements and can see the daily cycle and the effect of sferic spikes. This is interesting in and of itself and I'll describe it in a future post.

de w4hbk