[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linrad] Re: [Moon-net] SDR-IQ and Linrad

Hello Fletch,

> I fired up the latest version of Linrad-for-Windows with the SDR-IQ this
> morning and met with limited success in that Linrad recognised the SDR-IQ.
> When setting up the parameters for the SDR however, I am presented with a
> number of SDR-IQ-specific parameter questions that I do not really
> understand for the most part.  I was able to get the displays properly
> operating after a fashion by entering values for the parameters, although I
> really am unsure what those parameters represent.  Has anyone been
> successful with the SDR-IQ and Linrad, and if so, can you give me an idea of
> what the SDR parameters represent and how to properly apply them?

The SDR-IQ (as well as the SDR-14) have an A/D converter sampling at
66.666 MHz giving a bandwidth of 33 MHz. The USB link to the PC limits
the bandwidth to something between 160 and 230 kHz depending on 
operating system and hardware - and of course also depending
on whether you have other softwares running simultaneously.

The data rate, samples per second therefore has to be reduced by a
large factor and that is accomplished by an AD6620 (Analog Devices)

How to run the AD6620 depends on what you want to achieve. This
chip does not quite have the computing power needed and the USB
link only has 16 bits while 24 would be needed to not loose any
information. To make a setup that is good in all respects is not
possible so one has to make compromises.

You may look upon the AD6620 as three frequency mixers followed by
IF filters. The parameters determine how well undesired responses
are suppressed in the filters.

The first decision to take is to decide what bandwidth you want to
send into the PC. If you decide something like 150 kHz, you will 
have a dynamic range for signals that show up within the passband that
is smaller than the dynamic range for signals outside the visible
window. By only sending say 10 kHz to the PC, this problem is minimized.
On the other hand, the dynamic range within the 10 kHz will be
even smaller, one would loose 2 bits of dynamic range when sampling
16 times slower.

Since we are on MOON-NET I assume you want to use the SDR-IQ for
EME so I suggest you try a decimation factor of 432 to start with.
The total bandwidth will then be 154.3 kHz, but you will loose a 
little at both sides because of the filter responses.

The first decimation step will cause false responses separated by
66.666/M where M is the CIC2 parameter, the first one Linrad wants
as an input. It is good to make M as large as possible if you
have selectivity in front of the SDR-IQ that makes the false 
responses harmless. By setting CIC2=12 you create false responses
every 5.555 MHz away from the desired signal. In case you have no 
filter ahead of the SDR-IQ, when using it on HF for example, the
level of these spurs depend strongly on their separation from
the center frequency. When asking for a bandwidth of 154 kHz, 
we have to worry about spurs up to about 70 kHz away from the 
center. That is 0.21% bandwidth and therefore as we can read in
table III of the AD6620 documentation from Analog devices, these
spurs will be suppressed with about -63 dB. Only the spurs within
0.026% bandwidth (+/- 8.6 kHz) will be suppressed by 100 dB or more.
Presumably any converter from VHF or microwaves will attenuate
signals as far out as 5.5 MHz with something like 40 dB so these
spurs should be of no concern in EME.

The second step is governed by CIC5 while the third one is governed
by RCF. With CIC2=12 the product of CIC5 and RCF has to be 36 for
a total decimation of 432. RCF means ram coefficient filter, a 
conventional FIR filter that preceeds the last rate conversion.
Linrad will automatically compute the coefficients for a FIR with
adequate spur suppression for the spurs (aliases) that are generated
in the last rate conversion. The size of the flat center region
depends on the amount of resampling needed. If one makes CIC5=18
and RCF=2 one gets an extremely good RCF filter but the spurs
generated in the second rate conversion will become large.
With a bandwidth of 150 kHz at a sampling rate of 5.555 MHz (2.7%)
the spur suppression would be something like 30 dB only. This might
be OK for microwaves where there is only noise.

On the other hand, with CIC5=3 one gets 100 dB suppression, but then
RCF has to be set to 12 which means that only 1/12 of the total RCF
response falls within the passband that will be sent to the PC and
therefore the central flat region will be only about 80 kHz wide.

For EME you might set CIC2=16, CIC5=13, RCF=2 to get 160.25 kHz 
output sampling rateand a very wide flat response giving a usable 
bandwidth of 151 kHz. For HF something like CIC12=8, CIC5=7 and
RCF=7 might be more appropriate.

Finally you have to set the output shift. The output from the 
last step inside the AD6620 is 24 bit. The USB will only allow
16 bits so you will have to shift the output to accommodate
very strong signals in case you have any inside the passband. 
Try output shift=0 to start with. If you see too strong signals 
within the passband, then change for a larger shift. With too 
much output shift, your noise floor will be degraded....

How it all works is well described in the AD6620 documentation.

Note that digital receivers like SDR-xx should be fed with as much
signal as possible for best dynamic range. Signals that may
occur on the frequencies where IM3 is expected are caused by other
mechanisms than intermodulation and they are typically at the
same level, some 30 dB above the noise floor, regardless of the
level of the input signals:


  Leif / SM5BSZ

This message is sent to you because you are subscribed to
  the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to <linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>