Receiver comparisons: Testing is ongoing and results will change / be added to.
Receivers are set to normal. I.E. Preamp and attenuator off (unless noted), AGC / IF Gain to max. All tests done in USB mode at 14.2mHz
Results have been rather startling to say the least, and very unexpected. The 4 commercial rigs (these ones, or in general, or the test method) did not perform well on this occasion. This is being looked into. The test methods seem correct and robust, but will be re-affirmed here and in the lab. A definition of the exact test setup will follow, along with discussions around the results and expectations.
Note: The HamPi has different results to the others on face value, but do not read into this at face value – this I believe is a good thing.
Test procedure / methods are note lower down this page for each column in the following table.
Preamp Off / P1 / P2
Receiver Sensitivity A.
μV at 12 dB SINAD.
Preamp Off / Pa1 / Pa2
Receiver Sensitivity B.
(10db over noise
(dBm / uV)
(on S meter)
by xx dB with
below / above signal:
3, 5, 10,
20, 50 kHz
P1 / P2
(20mV to 200mV)
(but read later
|-145 none||109 .8uV|
|-96 ~10uV||83||6 6.5 (2kHz)|
|-141 none||107 1uV|
|-98 ~10uV||70||7 7 (2kHz)|
|-90 7uV||60||Note ***|
|-130 -140 -145||-109||105|
|-114 -126 -129|
-104 -117 -120
-96 -108 -112
|Test number||Test||Test method|
|1||Noise Floor||Reduce signal on RF signal generator to the point at which signal cannot be heard on the speaker, then increase by 1dB to hear signal, and record value. Test done without preamp on and off, and in USB mode, all else set to normal.|
|2||Receiver Sensitivity A||Receiver Sensitivity has been measured using two methods. |
a) Receiver sensitivity: measured at 12 dB SINAD. Results in dBm and μV
|3||Receiver Sensitivity B||b) This the usual method in Ham Radio. Measured by first measuring the noise floor and then increasing the signal to a point 10db above the noise floor, and then subtracting the floor reading from the signal reading. (dBm / uV). Personally I prefer the SINAD method above, it gives a fuller picture.|
|Method: Increase the RF level on the input with a signal generator to a point where tie S-Meter moves away from S0. Record the signal level in dBm and uV. |
Comment. This is a bit of a subjective test, and I have not seen a receiver that performs perfectly here. Far from it. Also the AGC has to be detected at the noise floor, not always easy especially with an Audio driven AGC. On all the receivers tested (apart from the HamPi) the AGC floor needed an RF signal greater than -100dBm. This means that when displaying, say, and S1 signal, it is in reality more like a S5 signal (-97dBm). This is to say, a lot of signals will be heard when the S-Meter is not moving off the S0 stop. On the HamPi the AGC Threshold is set to -107dBm, or 1uV.
Remember, when the AGC is below the threshold, the IF is on full gain and at maximum sensitivity. Essentially the AGC is off.
Here is a tale of S-Levels in an ideal world: (from WikiPedia)
S-Meter dBm uV(RMS~50Ohms)
S9+10dB -63 160.0
S9 -73 50.2
S8 -79 25.1
S7 -85 12.6
S6 -91 6.3
S5 -97 3.1
S4 -103 1.6
(-107dBm = 1uV)
S3 -109 0.8
S2 -115 0.4
S1 -121 0.2
|5||Dynamic Range||Method: a) Noise floor measured (No signal input) i.e. 4dB Sinad. b) Signal increased to 14dB SINAD i.e. 10dB over noise floor. Interfering signal increased to a point where SINAD is reduced to noise floor measured in a) above. Result = rf signal for 14dB Sinad minus the Interfering rf signal.|
This test has been done at 3kHz away from the desired signal. (Normally this done at 2kHz away from the wanted signal and on CW. This was not possible on SSB with some receivers, as it entered the with the filter skirt of the SSB filters).
|6||Comprehensive Blocking||I have included this range of tests to check the skirt of the SSB filters with an interfering signal of -20dBm at various distances from the fundamental frequency. This can show where QRM will get through.|
The SINAD will be degraded by ?? dB with a -20dBm interfering Signal at below and above the wanted signal when set to 12dB SINAD: @2kHz @10kHz @20kHz @50kHz
(*** NOTE: Tests to be redone at 3 and 5kHx rather than 2Hx. See notes elsewhere)
|8||AGC linearity||Method: Measure the S-Meter accuracy at different levels. @-73(S9), @-63(S9+10), @-53(S9+20), @-43(S9+30) |
Older analog receivers will struggle with this as the electronic AGC will be non linear and hard to calibrate. Also the AGC has to be detected at the noise floor, not always easy especially with an Audio driven AGC.
**** Note: More to be tested at S3, S5, S7
|1||On this IC-706-II, when adjusting the interfering signal (-30 to 160_dBm) to a point very close to the fundamental signal, the noise floor increases to a point where the fundamental signal can not be heard. Swamped at a 3dB decrease in SINAD. (Fundamental = -xxx dBm set to point -6dB SINAD, Interfering = -30dBm (to -60dBm) decrease 13dB SINAD but within pass-band of SSB filter). This will effect close in adjacent signals, QRM, in a pile-up etc by desensitising the receiver, especially to weak wanted signals.|
|2||On the SINAD testing comment ‘Clips filter pass-band’ means it was not possible to test with this -20dBm interfering signal this close (2kHz) to the filter as the signal was within the pass-band skirt, completely obliterating the signal. This is due to all testing being done on SSB, whereas the test normally calls for for CW where a narrow-band filter is employed. The HamPi does not employ a narrow-band CW filter (it has a narrow-band Audio filter).|
|3||Sensitivity vs Robustness. Jeremy Clarkson comes to mind on Top Gear. If you are going onto a battlefield what sort of vehicle would you take. Well Jeremy Clarkson might choose a fast nimble sports car, or a 4X4, but really, why don’t the army go with that choice? On the busy ham bands to cope with QRM what is needed is a something more like a tank! With all the interference these days from poor electronic equipment hitting the whole band, what is needed is ‘at hand’ reduction of interference – An IF gain, and other easily adjustable controls to filter or tune out the offending rubbish.|
It can be seen that the HamPi is not as sensitive as the other rigs below -107dBm. This is a known and well worth the trade-off, as many commercial rigs boast sensitivity down to -140 (The HamPi is good to -127 with the preamp off) at the cost of poor blocking figures where other close-in stations interfere or wipe out the wanted signal. This boasted sensitivity is wasted, and well below the S-Meter (AGC) being activated, i.e. Below S0. The HamPi AGC kicks in earlier however, at -107dBm, whereas the others were all below -100dBm. At all other S-Meter readings in the ‘Usable range’ s1 and above, the sensitivity was comparable than the others. Observation: Why are these sets so sensitive and is it really necessary to have sensitivity down to -140dBm range? Do we ever work stations when the S meter is not even moving? Yes the other sets are sensitive below S0, but the blocking and SINAD performance with an interfering signal is worse (bad in some cases!). This is the usual area where the best quality performance is needed on a busy band, i.e. when signals are above S1, and the HamPi seems to perform extremely well in this area compared to the others. This needs confirmation and further testing under lab conditions. See https://en.wikipedia.org/wiki/S_meter for S-Meter calibribration numbers.
|4||Why SINAD? And what is SINAD?. Signal-to-noise and distortion ratio (SINAD) is a measure of the quality of a signal from a communications device.|
SINAD is not often a measurement method on SSB in HAM radio circles, but widely used for FM. I have adopted SINAD as a measurement method as it accurately represents real usable signal compared to the noise of a signal, something other measurement methods does not do. Measurements have been done at varying distances away from the received signal with a huge interfering signal, and the degradation (increase of noise) recorded. This represents a ‘busy band’ much better than traditional measurements, and shows how the wide band interference (outside of the Band-Pass-Filter) and the close in filtering (Inside the Band_Pass_Filter pass-band, but outside of the pass-band of the main Intermediate Frequency (IF) SSB filter.
This is why SINAD is so widely adopted, it indicated accurately what is really happening.
It is not too hard to get the head around once put in simple terms. So, here goes:
20db SINAD = a ‘fully quieted signal, i.e. no noise content.
12dB SINAD = 75% signal and 25% noise. (intelligible speech can be detected 12 dB above the receiver’s noise floor (noise and distortion))
6dB SINAD = 50% signal and 50% noise.
3dB SINAD = 25% signal and 75% noise. Etc…
a) The reason 12dB SINAD is the all important target number is that for the human ear at 25% noise a conversation can be understood, and much more noise than this it cannot. In FM receivers the squelch is often set at 12dB for this reason, so that the receiver is muted when there is too much noise for the signal to be understood.
b) Possibly the reason it is not used for SSB is than on FM a 1kHz tone is 1kHz on the audio, even if the frequency is off. When measuring on SSB the frequency has to be adjusted to 1kHz with the tuning dial. In practice this is easy, and a SINAD meter does this for you, by just peaking the meter. A small inconvenience for an improved measurement.
For further reading & viewing:
– Video by Rohde Schwarz https://youtu.be/vQh-ZwUXERU
– Article by Electronics-Notes https://www.electronics-notes.com/articles/radio/radio-receiver-sensitivity/what-is-sinad-signal-to-noise-and-distortion.php
– Article by Repeater Builder www.repeater-builder.com/test-equipment/helper/helper-index.html