EMRFD Message Archive 6439

Message Date From Subject
6439 2011-07-10 06:20:32 Ashhar Farhan Pondering over spectrum analyzers
I am building a fairly primitive spectrum analyzer. I have a few
questions related to the architecture. I have never used a spectrum
analyzer, so bear with my naïve questions -
1. Why have a variable sweep rate? Why can't we keep it fixed at,
let's say, 5Hz?
2. How's the sweep rate related to the bandwidth of the filter? Does
the choice of IF (higher IF providing more samples) affect the maximum
sweep rate?
3. I would assume that frequency stability is not a major concern,
instead oscillator noise is. So we can use higher energy oscillators?
Can the second oscillator be variable too? I could use it for finer
sweeps.
4. What happens if my vco swings below the IF? Will I see th zero spur move up?

- farhan
4.

--
Sent from my mobile device
6440 2011-07-10 10:27:11 Thomas S. Knutsen Re: Pondering over spectrum analyzers
In order to compensate for different bandwith IF filters you need the sweep
rate variable.
With sharp filters you would need slow sweeps. I guess this is because the
filter ringing?
I can reccomend the HP AN-150 series application notes on spectrum analysis
for more information on this.

Stability is somewhat an consern, I find that I prefear to have the 2. LO in
my homebrew analyzer locked to an high stability reference. I did experience
some drifting of the analyzer before that. (the first IF is at 2GHz.)

Yes, your zero spur would move upwards. You should always be able to move
the zero spur across the screen with tune and span control.

I'm sorry not beeing able to explain the ratio between the filter bandwith
and sweep time better, I guess someone like Wes would be able to explain it
better.

73 de Thomas LA3PNA / AE5YS.

2011/7/10 Ashhar Farhan <farhanbox@gmail.com>

> **
>
>
> I am building a fairly primitive spectrum analyzer. I have a few
> questions related to the architecture. I have never used a spectrum
> analyzer, so bear with my na�ve questions -
> 1. Why have a variable sweep rate? Why can't we keep it fixed at,
> let's say, 5Hz?
> 2. How's the sweep rate related to the bandwidth of the filter? Does
> the choice of IF (higher IF providing more samples) affect the maximum
> sweep rate?
> 3. I would assume that frequency stability is not a major concern,
> instead oscillator noise is. So we can use higher energy oscillators?
> Can the second oscillator be variable too? I could use it for finer
> sweeps.
> 4. What happens if my vco swings below the IF? Will I see th zero spur move
> up?
>
> - farhan
> 4.
>
> --
> Sent from my mobile device
>
>



--

Please avoid sending me Word or PowerPoint attachments.
See <http://www.gnu.org/philosophy/no-word-attachments.html>


[Non-text portions of this message have been removed]
6441 2011-07-10 12:56:31 Lasse Re: Pondering over spectrum analyzers
Ashhar Farhan skrev 2011-07-10 15:20:
> I am building a fairly primitive spectrum analyzer. I have a few
> questions related to the architecture. I have never used a spectrum
> analyzer, so bear with my naïve questions -
> 1. Why have a variable sweep rate? Why can't we keep it fixed at,
> let's say, 5Hz?
Yes you can have it as slow as 5 Hz but usually you want as fast sweep
as possible to catch signals that varies over time. If you are using an
analogue oscilloscope you will have a hard time see the trace if you
sweep slow...
> 2. How's the sweep rate related to the bandwidth of the filter? Does
> the choice of IF (higher IF providing more samples) affect the maximum
> sweep rate?
If you go as fast as 1/filter-bw you will see a distorted signal, and
sweeping even faster you will smear out signal and ultimately not see
any signals!
> 3. I would assume that frequency stability is not a major concern,
> instead oscillator noise is. So we can use higher energy oscillators?
> Can the second oscillator be variable too? I could use it for finer
> sweeps.
"Correct" as in stability needs to be stable enough not to distort your
output at slower sweep speeds.
Having first LO fixed/locked and tune only second LO routinely done!
> 4. What happens if my vco swings below the IF? Will I see th zero spur move up?
Going below IF with input, signals will wrap around "zero", Position of
the "zero spur" is related to the sweep voltage, hence if tuning down in
frequency, (and not inverting the x-deflection) yes you will sweep from
"highest to lowest" frequency!
> - farhan
73 Lasse SM5GLC
6442 2011-07-10 21:23:11 Ashhar Farhan Re: Pondering over spectrum analyzers
tom, lasse,

thanks for your advice and explanations. i am reading the hp's application
note.
i also ran across http://lea.hamradio.si/~s53mv/spectana/sa.html (who
provided some solid break down of the design process).

at the moment, i am playing with various vco configurations to get a
low-noise vco than can span an octave. i am grappling two issues : getting
sufficient gain from my existing devices and being able to span an octave. i
am not sure if the vco output is falling beyond 150MHz or is it my
tektronics that's showing low pass characteristics. maybe it is time to
build an rf probe to measure the outputs.

i have built a direct conversion receiver at 100.24 MHz (14.318 on the 7th
overtone) to measure the phase noise.

- farhan VU2ESE

6443 2011-07-11 10:40:28 Brooke Clarke Re: Pondering over spectrum analyzers
Hi farhan:

The rise time of a signal is inversely proportional to the bandwidth of
the RBW filter in an analog type spectrum analyzer. So as the RBW is
narrowed the sweep time must be longer to avoid distorting the signal.
This is not true in digital spectrum analyzers, like the HP/Agilent
4395A, 4396A/B, etc, where the IF is I&Q digitized and then the
filtering and detection are done digitally. In this case the sweep
times can be considerably faster.
http://www.prc68.com/I/4395A.shtml

The phase noise of the local oscillators can be thought of as an FM
modulation of their frequency. This can have deteriorate the displayed
signal in some cases.
Early spectrum analyzers used a number of independent LOs in addition to
a sweep generator and so their frequency calibration was not the best.
Then the lock and roll type SA came out and was much better. The
ultimate is where all the LOs are digitally generated. For example the
HP/Agilent 4395A has provision for an external 10 MHz reference input
and when that's done you can measure the frequency of a signal as well
as using a frequency counter. For an example see:
http://www.prc68.com/I/BlkBox.shtml#E
http://www.prc68.com/I/DZAAS_Rx.shtml
Where the very narrow true RBW of 1 Hz and the use of an external
reference frequency allows measuring the LO of receivers using a short
length of wire as the input to the SA.

Have Fun,

Brooke Clarke
http://www.PRC68.com


> 1a. Pondering over spectrum analyzers
> Posted by: "Ashhar Farhan" farhanbox@gmail.com ashhar_farhan
> Date: Sun Jul 10, 2011 6:20 am ((PDT))
>
> I am building a fairly primitive spectrum analyzer. I have a few
> questions related to the architecture. I have never used a spectrum
> analyzer, so bear with my naïve questions -
> 1. Why have a variable sweep rate? Why can't we keep it fixed at,
> let's say, 5Hz?
> 2. How's the sweep rate related to the bandwidth of the filter? Does
> the choice of IF (higher IF providing more samples) affect the maximum
> sweep rate?
> 3. I would assume that frequency stability is not a major concern,
> instead oscillator noise is. So we can use higher energy oscillators?
> Can the second oscillator be variable too? I could use it for finer
> sweeps.
> 4. What happens if my vco swings below the IF? Will I see th zero spur move up?
>
> - farhan
> 4.
>
>