Leif,
I would like to thank you for the possible explanation you have given
about the compression tests made by QST reviewers on Perseus.
That could be the reason for the test results, even if the ARRL
Test Procedure Manual report establishes that the BDR test should be
made with the AGC off whenever possible, which is a case appliable to
Perseus since it has this option :-D
Amazingly what on the ARRL Test Procedure Manual is called BDR,
on QST appears to be called Blocking Compression Gain, or something
like that, which is not defined on the manual itself at all.
Even more amazingly, the BDR test as specified on the ARRL manual
does not give an exact idea of the true "blocking" dynamic range of a
receiver.
Here is an example.
If you browse the net for the QST report on the K3, for instance, you will
see that its BDR is reported as something like 140 dB.
This is a true value, the K3 has really 140 dB BDR, as per the ARRL manual.
But... does the K3 performs better than, for instance, Perseus - which is
rated 125 dB
at large spacings, when a large carrier interferer is placed a couple of
kHz apart the desired signal?
Surprisingly the reply is: NO.
The BDR definition as given by the ARRL Test Procedures Manual says (see
pag. 48):
5.7.1
"...The blocking dynamic range is the difference between the level of the
noise floor from
the level of undesired signal that produces a 1 dB decrease in a weak
desired signal..."
Ok, but how much weak the desired signal should be? Should it be at the MDS
level
or somewhere else? The ARRL manual does not specify exactely but it looks
like that
it is in the range of - 80 dBm or so.
(I understand that for big DXers a S 8 is a weak signal not a strong one, of
course)
For such a weak signal, which nevertheless is usually 50 dB above an HF
receiver MDS, the effects
of the phase noise of the receiver 1st conversion LO can be usually
neglected.
In the case ot the K3, for instance, the reciprocal mixing at 2 kHz spacing
is something
like 95 dBc (as measured on the QST report). This means that the noise added
to the
interfered signal even by a +7 dBm interfering carrier is at - 88 dBm and
still lies below the
weak - 80 dBm, 1 dB compressed, desired signal.
What about the SNR of the weak signal in this case? It is 8 dB or so
(-80-(-88) = 8).
Now take a SDR like Perseus (sorry if I take this biased example) which has
a 114 dBc
phase noise performance at the same 2 kHz offset and feed it with the same
+7 dBm intefering
carrier.
Ouch! But Perseus is well saturated at + 7dBm (clipping level at - 3 dBm)!
Ok, it is. Let insert its 10 dB attenuator so that it isn't anymore.
The MDS of the receiver has gone to say - 115 dBm but its noise floor with
the 10 dB attenuator
inserted is still lower than the LO phase noise, which is now at a + 7 - 114
= - 107 dBm.
Say that the total noise (phase + floor) is at - 105 dBm.
What about the SNR of the same - 80 dBm weak signal now?
- 80 - (-105) = 25 dB. 17 dB better than the K3!!!
How can Perseus produce a signal quality wich is 17 dB better than the K3
even if the input signals
are the same and the K3 is rated 140 dB BDR while the Perseus a more
moderate value?
The reply is that a true BDR test should check the dynamic range when the
weak desired
signal is at the MDS level, not 20 or 50 dB above it! If you do the test
with a desired signal which is much
higher than the MDS you are missing the true dynamic capability of a
receiver at small spacings
(as in most cases you're simply neglecting the receiver LO phase noise).
The K3 and Perseus are just two examples, anyway. Substitute the K3 with
another (good) analog rig and
Perseus with a QS1R, repeat the calculations and you will obtain almost the
same result.
When the interferer is within +/- 50/100 kHz good digital platforms
outperform any analog receiver, even
those who are rated with a 140 dB BDR.
But even at larger offsets, no receiver has a *true* 140 dB BDR in a 500 Hz
bandwidth unless its LO
phase noise is less than -167 dBc/Hz, a very small value which sometimes
even crystal oscillators fail to reach.
Sorry for the lenght of this reply, maybe a little bit off topic, but ok,
now Alberto can give a look to
Winrad to check for the AGC bug if he wishes it so :-D
73
Nico, IV3NWV
2008/11/28 Leif Asbrink <[EMAIL PROTECTED]>
> Hi All,
>
> The latest QST, Dec 08, has a test of the Perseus HF receiver.
> The dynamic range data suggests 1 dB compression for a signal
> 99 dB above MDS which evaluates to -27 dBm. The mechanism is
> that sideband noise from the test oscillator activates the AGC
> and reduces the gain. (With AGC off there is no gain compression
> until above A/D saturation.) perseus.exe has a very low
> threshold for the AGC by default. I do not know how to
> change it (if possible.)
>
> I tried to do the measurement with Winrad where I can set the
> AGC threshold. (ARRL Lab should have no problem to use Winrad.
> Using Linrad would require far more time....)
>
> Things behave as expected with one exception:
> When I switch off the AGC and increase the signal level, there is
> a point where Winrad overflows. The program crashes and the exit
> button does not work. I have to use Ctrl Alt Del to be able to
> kill Winrad. The crasch happens at S9+20 dB when RF gain ("Lev")
> is set to maximum and it does not happen at all when RF gain is
> set to minimum. The bug is 100% reproducible so it should be easy to
> find:-)
>
> 73
>
> Leif / SM5BSZ
>
>
>
