JL
Jim Lux
Fri, Sep 23, 2011 5:16 PM
On 9/23/11 10:04 AM, Jose Camara wrote:
I think you are right, often the internal, free running osc will give you better results. You can use the GPS or rubidium to calibrate the internal one just before you need some more accurate absolute frequency measurements on the SA.
It will depend on what measurement you are making, and whether phase noise or frequency accuracy is more important. For day to day use, the external ref will work, except when perhaps you need to look at very close skirts, where maybe the internal alone can give you lower noise. In most cases, you don't really need either (checking a filter, EMI, radio output, etc.) but a lot of thing in this list is because we can, not because we need. :-)
Get a real clean, low phase noise 3rd signal, measure it using the internal and external osc, look at the skirts. They might even be the same, if the limit is elsewhere in the SA signal chain.
One other thing is that some spectrum analyzers aren't really designed
for low noise performance. Since the noise floor is often pretty high,
the design of the whole RF chain (e.g. spur levels and such) might have
assumed that lots of things would be hidden in the grass. If the
analyzer is of the recent "bring a band of RF down to an IF, sample and
FFT it for fine resolution" architecture, such things as the number of
bits in the ADC and the "cleanliness" of the sampling clock might have
been chosen based upon doing 1024 point transforms being displayed with
100dB dynamic range (10dB/div and 10 divisions).
(not to mention the spectrum analyzer actually generating spurious
signals. I ran across that one last year and thought I had an
interference source, but, no, went back and checked the spec sheet and
it said spurious are <-80dBc, and sure enough, there it was at -82 dBc.
And stories about the first LO coming back out through the input are
legion.)
On 9/23/11 10:04 AM, Jose Camara wrote:
> I think you are right, often the internal, free running osc will give you better results. You can use the GPS or rubidium to calibrate the internal one just before you need some more accurate absolute frequency measurements on the SA.
>
> It will depend on what measurement you are making, and whether phase noise or frequency accuracy is more important. For day to day use, the external ref will work, except when perhaps you need to look at very close skirts, where maybe the internal alone can give you lower noise. In most cases, you don't really need either (checking a filter, EMI, radio output, etc.) but a lot of thing in this list is because we can, not because we need. :-)
>
> Get a real clean, low phase noise 3rd signal, measure it using the internal and external osc, look at the skirts. They might even be the same, if the limit is elsewhere in the SA signal chain.
>
One other thing is that some spectrum analyzers aren't really designed
for low noise performance. Since the noise floor is often pretty high,
the design of the whole RF chain (e.g. spur levels and such) might have
assumed that lots of things would be hidden in the grass. If the
analyzer is of the recent "bring a band of RF down to an IF, sample and
FFT it for fine resolution" architecture, such things as the number of
bits in the ADC and the "cleanliness" of the sampling clock might have
been chosen based upon doing 1024 point transforms being displayed with
100dB dynamic range (10dB/div and 10 divisions).
(not to mention the spectrum analyzer actually generating spurious
signals. I ran across that one last year and thought I had an
interference source, but, no, went back and checked the spec sheet and
it said spurious are <-80dBc, and sure enough, there it was at -82 dBc.
And stories about the first LO coming back out through the input are
legion.)
BC
Brooke Clarke
Fri, Sep 23, 2011 5:54 PM
On 9/23/11 10:04 AM, Jose Camara wrote:
I think you are right, often the internal, free running osc will give
you better results. You can use the GPS or rubidium to calibrate the
internal one just before you need some more accurate absolute
frequency measurements on the SA.
It will depend on what measurement you are making, and whether phase
noise or frequency accuracy is more important. For day to day use,
the external ref will work, except when perhaps you need to look at
very close skirts, where maybe the internal alone can give you lower
noise. In most cases, you don't really need either (checking a
filter, EMI, radio output, etc.) but a lot of thing in this list is
because we can, not because we need. :-)
Get a real clean, low phase noise 3rd signal, measure it using the
internal and external osc, look at the skirts. They might even be the
same, if the limit is elsewhere in the SA signal chain.
One other thing is that some spectrum analyzers aren't really designed
for low noise performance. Since the noise floor is often pretty high,
the design of the whole RF chain (e.g. spur levels and such) might
have assumed that lots of things would be hidden in the grass. If the
analyzer is of the recent "bring a band of RF down to an IF, sample
and FFT it for fine resolution" architecture, such things as the
number of bits in the ADC and the "cleanliness" of the sampling clock
might have been chosen based upon doing 1024 point transforms being
displayed with 100dB dynamic range (10dB/div and 10 divisions).
(not to mention the spectrum analyzer actually generating spurious
signals. I ran across that one last year and thought I had an
interference source, but, no, went back and checked the spec sheet and
it said spurious are <-80dBc, and sure enough, there it was at -82
dBc. And stories about the first LO coming back out through the input
are legion.)
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hi Jim:
Do you know how the HP/Agilent 4395A stacks up as a SA? I really like
the true RMS power detection and the 1 Hz RBW (not video).
<http://www.home.agilent.com/agilent/product.jspx?id=1000000864:epsg:pro&pageMode=OV&pid=1000000864:epsg:pro&lc=eng&ct=PRODUCT&cc=US&pselect=SR.PM-Search%20Results.Overview>
Have Fun,
Brooke Clarke
http://www.PRC68.com
http://www.End2PartyGovernment.com/
Jim Lux wrote:
> On 9/23/11 10:04 AM, Jose Camara wrote:
>> I think you are right, often the internal, free running osc will give
>> you better results. You can use the GPS or rubidium to calibrate the
>> internal one just before you need some more accurate absolute
>> frequency measurements on the SA.
>>
>> It will depend on what measurement you are making, and whether phase
>> noise or frequency accuracy is more important. For day to day use,
>> the external ref will work, except when perhaps you need to look at
>> very close skirts, where maybe the internal alone can give you lower
>> noise. In most cases, you don't really need either (checking a
>> filter, EMI, radio output, etc.) but a lot of thing in this list is
>> because we can, not because we need. :-)
>>
>> Get a real clean, low phase noise 3rd signal, measure it using the
>> internal and external osc, look at the skirts. They might even be the
>> same, if the limit is elsewhere in the SA signal chain.
>>
>
> One other thing is that some spectrum analyzers aren't really designed
> for low noise performance. Since the noise floor is often pretty high,
> the design of the whole RF chain (e.g. spur levels and such) might
> have assumed that lots of things would be hidden in the grass. If the
> analyzer is of the recent "bring a band of RF down to an IF, sample
> and FFT it for fine resolution" architecture, such things as the
> number of bits in the ADC and the "cleanliness" of the sampling clock
> might have been chosen based upon doing 1024 point transforms being
> displayed with 100dB dynamic range (10dB/div and 10 divisions).
>
> (not to mention the spectrum analyzer actually generating spurious
> signals. I ran across that one last year and thought I had an
> interference source, but, no, went back and checked the spec sheet and
> it said spurious are <-80dBc, and sure enough, there it was at -82
> dBc. And stories about the first LO coming back out through the input
> are legion.)
>
> _______________________________________________
> time-nuts mailing list -- time-nuts@febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>
>
JL
Jim Lux
Fri, Sep 23, 2011 7:42 PM
On 9/23/11 10:54 AM, Brooke Clarke wrote:
Not off hand.. I'm usually working at microwave frequencies and haven't
used that particular one.
All of those spectrum analyzers with 1 Hz RBW are almost certainly doing
some sort of FFT, and I assume the sample clocks are controlled such
that you can hook up to some sort of external reference with 1E-9 kinds
of stability.
The question might be whether the actual clock performance has good
enough phase noise at 1 Hz offset to make the 1 Hz RBW meaningful. A
-100dBc/Hz @ 1 Hz offset would be mighty impressive, no?
In fact, the specsheet shows of -95dBc/Hz at 1-100kHz offset. The
typical performance is better, of course, but still, it's about
-90dBc/Hz at 100Hz offset.
On 9/23/11 10:54 AM, Brooke Clarke wrote:
> Hi Jim:
>
> Do you know how the HP/Agilent 4395A stacks up as a SA? I really like
> the true RMS power detection and the 1 Hz RBW (not video).
>
> <http://www.home.agilent.com/agilent/product.jspx?id=1000000864:epsg:pro&pageMode=OV&pid=1000000864:epsg:pro&lc=eng&ct=PRODUCT&cc=US&pselect=SR.PM-Search%20Results.Overview>
>
>
Not off hand.. I'm usually working at microwave frequencies and haven't
used that particular one.
All of those spectrum analyzers with 1 Hz RBW are almost certainly doing
some sort of FFT, and I assume the sample clocks are controlled such
that you can hook up to some sort of external reference with 1E-9 kinds
of stability.
The question might be whether the actual clock performance has good
enough phase noise at 1 Hz offset to make the 1 Hz RBW meaningful. A
-100dBc/Hz @ 1 Hz offset would be mighty impressive, no?
In fact, the specsheet shows of -95dBc/Hz at 1-100kHz offset. The
typical performance is better, of course, but still, it's about
-90dBc/Hz at 100Hz offset.
RD
Robert Deliën
Sun, Sep 25, 2011 9:32 AM
One other thing is that some spectrum analyzers aren't really designed
for low noise performance. Since the noise floor is often pretty high,
the design of the whole RF chain (e.g. spur levels and such) might have
assumed that lots of things would be hidden in the grass.
True, it's one of the many selection criterions for selecting the instrument that meets your needs.
I've been looking a the luggable HP series 859x and 856x, preferring the latter because they have a PLL YIG whereas the fist uses a free-running oscillator. But these machines are old, 80's and 90's, pricey, and not really THAT good. Add decent range (up to 9GHz to see recent 5.8GHz devices) and a tracking generator and before you know it, you'll be paying $6k or more for a 20 year old instrument.
If the
analyzer is of the recent "bring a band of RF down to an IF, sample and
FFT it for fine resolution" architecture, such things as the number of
bits in the ADC and the "cleanliness" of the sampling clock might have
been chosen based upon doing 1024 point transforms being displayed with
100dB dynamic range (10dB/div and 10 divisions).
Most modern instruments do that, at least to some degree. My R&S goes down to a RBW of 10Hz by just mixing. Additionally RBWs of 5, 3, 2 and 1Hz are achieve by additional FFT. This instrument dates from 2001, but I don't think more recent instruments can achieve a mixing-only RBW of 5Hz or below.
(not to mention the spectrum analyzer actually generating spurious
signals. I ran across that one last year and thought I had an
interference source, but, no, went back and checked the spec sheet and
it said spurious are <-80dBc, and sure enough, there it was at -82 dBc.
And stories about the first LO coming back out through the input are
legion.)
Gee, I wish I had consulted this group BEFORE buying my instrument. I'm happy with it and I don't regret anything, but you could have added a lot more arguments in favor or against…
> One other thing is that some spectrum analyzers aren't really designed
> for low noise performance. Since the noise floor is often pretty high,
> the design of the whole RF chain (e.g. spur levels and such) might have
> assumed that lots of things would be hidden in the grass.
True, it's one of the many selection criterions for selecting the instrument that meets your needs.
I've been looking a the luggable HP series 859x and 856x, preferring the latter because they have a PLL YIG whereas the fist uses a free-running oscillator. But these machines are old, 80's and 90's, pricey, and not really THAT good. Add decent range (up to 9GHz to see recent 5.8GHz devices) and a tracking generator and before you know it, you'll be paying $6k or more for a 20 year old instrument.
> If the
> analyzer is of the recent "bring a band of RF down to an IF, sample and
> FFT it for fine resolution" architecture, such things as the number of
> bits in the ADC and the "cleanliness" of the sampling clock might have
> been chosen based upon doing 1024 point transforms being displayed with
> 100dB dynamic range (10dB/div and 10 divisions).
Most modern instruments do that, at least to some degree. My R&S goes down to a RBW of 10Hz by just mixing. Additionally RBWs of 5, 3, 2 and 1Hz are achieve by additional FFT. This instrument dates from 2001, but I don't think more recent instruments can achieve a mixing-only RBW of 5Hz or below.
> (not to mention the spectrum analyzer actually generating spurious
> signals. I ran across that one last year and thought I had an
> interference source, but, no, went back and checked the spec sheet and
> it said spurious are <-80dBc, and sure enough, there it was at -82 dBc.
> And stories about the first LO coming back out through the input are
> legion.)
Gee, I wish I had consulted this group BEFORE buying my instrument. I'm happy with it and I don't regret anything, but you could have added a lot more arguments in favor or against…
RD
Robert Deliën
Sun, Sep 25, 2011 9:42 AM
Vector Network Analysers like this one often don't offer full Spectrum Analyzer functionality. They're difficult to operate too. If you don't really need the specialized functionality, I would look for something else. But if one is offered to you at a reasonable price…
Take a look at the Rohde & Schwarz FSEB20, FSEB30, FSEA20 and FSEA30. Those offer both true RMS power detection (even though a good power meter is always a better instrument to measure power) and a 1 Hz RBW, where the latter is achieved combining a 10Hz bandwidth with FFT.
I just warmed up my FSIQ (which is basically an FSEA with IQ tracking generator and a couple of other options) and made a 1Hz RBW plot of the calibration signal, to attach it to this post, but I don't have a floppy drive in my PC :-( Talking about trailing-edge technology.
> Do you know how the HP/Agilent 4395A stacks up as a SA? I really like
> the true RMS power detection and the 1 Hz RBW (not video).
>
> <http://www.home.agilent.com/agilent/product.jspx?id=1000000864:epsg:pro&pageMode=OV&pid=1000000864:epsg:pro&lc=eng&ct=PRODUCT&cc=US&pselect=SR.PM-Search%20Results.Overview>
Vector Network Analysers like this one often don't offer full Spectrum Analyzer functionality. They're difficult to operate too. If you don't really need the specialized functionality, I would look for something else. But if one is offered to you at a reasonable price…
Take a look at the Rohde & Schwarz FSEB20, FSEB30, FSEA20 and FSEA30. Those offer both true RMS power detection (even though a good power meter is always a better instrument to measure power) and a 1 Hz RBW, where the latter is achieved combining a 10Hz bandwidth with FFT.
I just warmed up my FSIQ (which is basically an FSEA with IQ tracking generator and a couple of other options) and made a 1Hz RBW plot of the calibration signal, to attach it to this post, but I don't have a floppy drive in my PC :-( Talking about trailing-edge technology.
