[hpsdr] Call for Comments and Discussion - CYCLOPS

[Phil Harman]

Bill,

Graham is spot on with his explanation. We are really limited to what is
available in terms of off-the-shelf first IF filters in the GHz range.

We could have designed an interdigital first IF filter in the GHz range
but that would not have been neither small nor low cost.

By using COTS dielectric filters we can keep the dimensions down to Atlas
bus board size and low cost.

The choice of second IF (96MHz) is to ensure we have sufficient image
rejection ( 2 x 96MHz away) to products from the first LO/Mixer.  With two
TOKO dialectic filters it looks like we an get about 100dB of image
rejection.  If we go lower in frequency, say within the standard range of
Mercury, then we either have to accept degraded image rejection or improve
the selectivity of the first IF filter(s).

There are also some very high performance and low cost SAW filters
available at 96MHz.

My favourite local FM station is on 96.1MHz so we are aware of the
potential pick up from broadcast stations and, with the gain distribution
used plus allowance for screening, it seems manageable.  Graham is also
within a stones throw of a number of FM stations.

If we can find a high performance, cost effective, first IF filter in the
GHz range then we can reduce the second IF frequency.  The TOKO filters
represent the best compromise we have found so far.

I expect that the usable bandwidth (i.e. so we don't have to apply
amplitude and phase correction) will be about 1MHz.  To display a larger
bandwidth we step the first LO in 1MHz steps and concatenate the results.

>From the literature searches that I have done so far most modern FFT based
analysers work this way.

It will be necessary to make a modification to Mercury in order for it to
be effective as a 96MHz second IF.  Either removal of the input LPF or a
link from input to output (with a matching network at 96MHz) will be
required.  Once Alex is available the LPF on the front of Mercury will be
optional so it can be permanently removed if so desired.

There is a LPF at the output the Preamp that prevents its VHF noise
entering the ADC but this has very little effect at 96MHz and the gain
distribution of Cyclops takes this into account.

In relation to the 10MHz reference for Cyclops, we can pick this up from
Atlas C16 so at minimum it can be the TCXO on Mercury all the way to a GPS
locked reference via Exchalibur.  The PLL chips used will accept a 3.3v
squarewave directly off the Atlas bus.

With any new design there are always options and compromises - it really
boils down to how acceptable the end result is.  In terms of noise figure
(~30dB) and phase noise etc if we can meet the specifications of the parts
we use then this is going to be a very useful device.

And yes, we are starting to run out of slots on Atlas - perhaps a longer
Atlas board or fewer projects  :)

I trust that answers the comments so far.

73's Phil...VK6APH





> Bill:
>
> I think the limitation on the bandwidth of the 1.6 MHz wide SAW filter
> is due to
> a limited selection of off-the-shelf components.  I think Phil's
> original plan
> was to scan the desired spectrum in 2 MHz steps by moving the High-LO
> and doing the finer processing in software.  This filter will probably
> require
> scanning in 1 MHz steps, or maybe 1.5 MHz steps.
>
> --- Graham
>
> ==
>
> Bill Tracey wrote:
>> Cyclops diagram and schems looks good to me so far.    A question - at
>> the risk of confirming myself as more a software person than a radio
>> person.... why the relatively narrow 1.6 MHz filter in front of the
>> output on Cyclops?    Would it not be better to let a wider swath into
>> Mercury  - the FPGA code can always filter out what we don't want ..
>> seems in an SDR solution one would want to allow as wide a swath in as
>> you can and let the firmware/software filter where needed.
>>
>> Regards,
>>
>> Bill  (kd5tfd)
>>
>>
>> At 04:54 PM 7/11/2009, Graham / KE9H wrote:
>>> ***** High Performance Software Defined Radio Discussion List *****
>>>
>>> All:
>>>
>>> An information package has been posted to the Wiki on the
>>> CYCLOPS page, under the "July 11 Update" heading, with
>>> block diagram, schematics and a proposed PCB layout for
>>> a prototype CYCLOPS spectrum analyzer card.
>>>
>>> The Tracking generator will likely be a separate card developed
>>> after the main SA receiver card is running.
>>>
>>> Please review the package and direct questions, comments,
>>> constructive criticism, etc to Phil, VK6APH or myself. Phil is the
>>> project leader and designer.  I have been assisting Phil with
>>> the PCB layout, bill of materials and sourcing.
>>>
>>> --- Graham / KE9H
>>
>>
>>
>>
>
>
>
>



 1247635980.0