FFT Spectrum Analyzer: software-based or standalone?

[John Walker]

Hello. I work in a physics research lab, and we are planning to purchase a
spectrum analyzer. We need to cover the range from DC (actually, 0.1 Hz
would be enough) to 300-400 kHz to look at the electronic error signals we
get in a laser stabilization experiment (these are just the electronic
representation of frequency fluctuations of the laser, most of them due to
mechanical vibrations in this range of frequencies). This is the "essential
minimum" we need, but since we are making the investment it'd also be nice
to be able to go to higher frequencies, we'd have plenty of additional uses
for an instrument reaching up to 100 MHz. Ideally the instrument would be
of the FFT flavor.

We have been looking at a number of models and there aren't many which
cover the low frequencies down to tenths of a Hertz, this seems to be the
main limitation. The 89400 series of Agilent seems to be all we need and
more, but it costs a kidney, so one option we are considering is to
sacrifice a little bit of range in the low frequencies and go with a 4395A
(from 10 Hz) or even better, a 4396B (from 2 Hz). These models are more
affordable and have the added advantage of incorporating network and
impedance analyzer funcionality, which is not essential in our case but
certainly nice to have in a lab like ours, we have to build and test plenty
of home-made circuits and little gizmos.

The second possibility we are considering is a PC-based solution like the
89600 series, we would only need to purchase a suitable laptop, the
software (the 89640 seems adequate) and a hardware front-end like a VXI
mainframe. The final price is probably quite a bit lower than that of a
standalone instrument.

What I would like is to get some opinions of people who have working
experience with these two types of spectrum analyzer, standalone and
software-based. Which one would you recommend for lab use? What are the
advantages and disadvantages of each one? In terms of functionality, is any
of the options clearly superior to the other one? I have only mentioned HP
(Agilent) products, but please feel free to recommend any other
brands/models that you consider adequate for this task.

Thanks a lot!

 

[Robert Lacoste]

Another low cost solution will be to be a used HP3585A or similar.
Exceptionally good performances, DC to 40MHz, 3Hz resolution filter, and
cheap... They're not FFT based, but I just bough one on eBay for less than
1000$. And the advantage of non-FFT based equipments is the usually higher
dynamic range(80dB in that case).

Friendly,
Robert Lacoste - ALCIOM
http://www.alciom.com

 

[Aidan Grey]

Another low cost solution will be to be a used HP3585A or similar.
Exceptionally good performances, DC to 40MHz, 3Hz resolution filter, and
cheap... They're not FFT based, but I just bough one on eBay for less than
1000$. And the advantage of non-FFT based equipments is the usually higher
dynamic range(80dB in that case).

Friendly,
Robert Lacoste - ALCIOM
http://www.alciom.com

( Snip )

One thing you should consider is if you should get two seperate instruments
instead of one.

Buy an FFT spectrum analyzer for DC to the low KHz range. Buy a different
unit for work above one 1 MHz.

Two specialized tools will probably do a better job than one single one.
In
addition, it actually might be cheaper to buy two different units.


Aidan Grey

 

[John Walker]

One thing you should consider is if you should get two seperate instruments
instead of one.

Buy an FFT spectrum analyzer for DC to the low KHz range. Buy a different
unit for work above one 1 MHz.

Two specialized tools will probably do a better job than one single one.
In addition, it actually might be cheaper to buy two different units.

Yeah, that would we a good solution. In fact, if I were able to find a
suitable instrument for the DC-kHz range I'd be happy enough because that
is the range where we need one badly, RF is very nice to have access to but
not such a pressing need for us. However, I haven't been able to find a
suitable one. The closest thing I have found is one of the models offered
by SRS (Stanford Research Systems): FFT instruments operating from DC to
100 kHz. Very nice and affordable *but* 100 kHz is too low for us, we need
something that goes up to perhaps 300 kHz at least. So back to square
one...

I have been looking into the possibility -suggested by another poster- of
getting an older, non-FFT model like the HP3585A. It looks tempting enough
because for something like $1000 is basically a present... the problem is
that I don't really know how convenient non-FFT spectrum analyzers are when
you have to make a "noise" measurement, that is, measuring a signal that is
jumping around in frequency and amplitude all the time instead of being at
a single frequency all the time... What I need to get is really a
measurement of noise density (signal density if you prefer to think of it
like that, since noise is my signal here) in the frequency range of
interest. I suppose it can be done with a non-FFT (sweep) instrument, but
won't it take forever? Does anybody have experience with this?

Thanks for your suggestions!

 

[Ian Molton]

Yeah, that would we a good solution. In fact, if I were able to find a
suitable instrument for the DC-kHz range I'd be happy enough because
that is the range where we need one badly

A software one on a high spec PC can cope with that. I dont know what
the jitter is like on a modern soundcard, but if you have a spare PC
around, running linux, you can try the FFT program from my website.

 

[Ian Molton]

A software one on a high spec PC can cope with that. I dont know what
the jitter is like on a modern soundcard, but if you have a spare PC
around, running linux, you can try the FFT program from my website.

*arg*.

Didnt see the 100kHz is too low bit. sorry.

 

[ctsbillc]

Sound cards are capacitor coupled, but they do go down to a few Hz.

Spectralab is a good software package, but there are some free ones as well
which do quite a lot.

Cheap sound cards often have spurii associated with them.

Bill