FarhanSi5351 has much higher non-harmonic spurious than si570, and maybe not
good for the LO"s.
Sent from my Galaxy Tab4 10.1 4G LTE Tablet
-------- Original message --------
From: Ashhar Farhan <farhanbox@xxxxxxxxx>
Date: 8/30/2015 12:16 AM (GMT-05:00)
To: minima@xxxxxxxxxxxxx
Subject: [minima] Re: A homebrew spectrum analyzer
allison,
you could be right about using the CMOS version of Si570. Let me try plugging
in a CMOS version and look at the spurs. probably, today. the CMOS version with
it's higher output can easily drive the ADE-1 without needing the MAV-11.
Infact, it might do to add an attenuator in the path.
there is another mod to consider : replace the post mixer MAV-11 with a 6db
pad. This will bring up the noise figure to 25db or so, but i guess it would be
an easier and cheaper build to do away with both the MAV-11s. I find myself
using an 20db attenuator all the time in front of the specan. so, i guess
permanent pads on all sides of the first mixer can only be good.
I found that the microprocessor lines bring in a lot of spurs into the radio
blocks. I shielded the I2C lines coming into the Si570 with RG174. Instead, I'd
like to try mounting the Si570 inside the digital enclosure and bringing out
just the Si570 output.
I suppose the Si5351 is the best bet for the tracking generator. i am going to
throw one in right way and check out how it works. What would be a good level.
a number of people have been campaigning for Si5351 in place of the Si570. My
argument is 8 dollars is not worth degrading your specan for. On the other
hand, Si570 is easily hand soldered. the Si5351 is impossible for me to handle.
- f
On Sun, Aug 30, 2015 at 1:46 AM, allison <ajp166@xxxxxxxxxxx> wrote:
On 08/29/2015 02:56 PM, Ashhar Farhan
wrote:
allison,
i tried two different tracking generators based on the
hetrodyne principle. they both leaked into the IF badly,
raising the noise floor by 15 db.
I used power splitters so there is more isolation between the source
and the two destinations.
Given a chance the RF doing to mixer one can easily find a path to
the tracking generator
mixer if not stopped in its tracks.
Also even though the 570 and MAV11 are driving the mixer a low pass
filter at the mixer at IF
plus 80mhz or so can help as higher order harmonics from the source
are an issue. Also
try to not over drive the MMIC they can do weird things, even
oscillate at selected frequencies.
Seems like a painful loss but a input attenuator of 6db is advised
at the mixer. It protects the mixer
from an oops (transmit into it) and also it guarantees the input
filter has a good termination.
I've measured a few Spec analyzers and typical front end noise
figures are high in the 22db
range. If you need lower noise figure (improved input sensitivity)
the solution is a clean
(more than quiet) preamp for time when you looking down below -90 to
-100 dbm or so.
Hint if it slides around the bench easily likely not enough
shielding. Each major stage lives
in it s own box. For spec analyzers its a must. Soldered up boxes
with lids made of copper
clad work very well for this.
Now, instead, I am going to try an Si5351 directly at the
signal frequency.
I'd think that would work. But adjacent channel coupling is likely.
I did try the CMOS version of Si570. but i noticed more
spurs than with the LVDS version. I am not sure if I must
attribute it to the LVDS version alone. I had substantially
increased the shielding between changing the Si570 type.
Played with CMOS and LVDS and noted no real difference save for
interface. Shielding is
everything. Generally speaking the NCOs (Si570) are close to spur
free if harmonics are
ignored. They unfortunately due to the near perfect square wave are
strong on harmonics.
Spec analyzers are a good exercise as they allow one to look deep
(small signals) over wide
bandwidth and a consequence is that you start seeing things you
might have not noticed in
a narrow band system or thoroughly filtered systems. All the good
ones I use even the
portable Rhode&Schwarz FSH6 are remarkably heavy and its all
shielding.
FYI take care with microprocessor clock and data lines especially
the SCLOCK. They
can make a racket.
Allison
- f
On Sat, Aug 29, 2015 at 11:59 PM,
allison <ajp166@xxxxxxxxxxx>
wrote:
Farhan,
One suggestion:
Add tracking generator (difference of first osc and second,
filter and
pump up with amp)
as then it adds sweeperino capability another useful
function in one
instrument. It adds
little to the complexity but adds huge functionality.
One comment. Shielded modules, dead bug or PC are a must to
get the
best result.
Allison
On 08/28/2015 01:01 AM, Ashhar Farhan wrote:
> While working on the Minima project, I needed a
spectrum analyzer to
> measure the performance. The new specan is as
complicated as a simple
> SSB rig, probably even simpler.
>
> This is an Si570/Arduino controlled spectrum
analyzer that can be
> built for less than hundred dollars. It is simple
to build and it can
> be used to measure IMDR, spurs, harmonics, etc. It
doesn't need an
> oscilloscope.
>
> I like you all to read about
> it http://hfsignals.blogspot.in/p/specan-reboot-of-w7zoi.html
>
> - f
