Does professional spectrum analyzer such as keysight,RS has `aliasing` problem?

[vk6zgo]

Aliasing is more often an oscilloscope problem.  A spectrum analyzer calls it the way it sees it.

The problem is "how it sees it!"

A transmitter on around 433MHz (in an ISM application, so it wasn't radiated to air) would usually work OK for several hours, then there would be sharp drop in output power from the transmitter.

Looking at a monitor point on the output, with the spectrum analyser (An early 2000s IFR), we saw that, when the fault happened, the single carrier at 433MHz would be replaced by multiple carriers spaced over a 40MHz range up to 477MHz.
Checking back to the carrier source, which was a PLL, gave a similar result.

The fault was annoyingly intermittent, we could get several hours of normal operation, then an episode.

The fault could adversely affect the process it was used for, so it was also important that the operator could tell when it happened, & call us, but leaving the IFR connected meant we couldn't use it in the workshop for other work, so we bought a little "El Cheapo" analog SA for the dedicated monitoring spot.

Luckily, one of the Techs was present the next time the fault occurred,.
The carrier smoothly slid up from 434MHz to around 477MHz----no spurious carrier or anything, just a PLL which had become unlocked!

The Tx fault was found to be that it required the data commanding the required frequency to be constantly available, & there was an intermittent fault in the header connecting this.
Data loss caused the PLL to go to its default frequency.


The  IFR SA problem was that it didn't sweep continuously like the analog one, but, instead "swept" by taking samples over time.
This was perfectly good for most purposes, but a carrier shiftng fast in one direction was displayed as multiple carriers.

I guess you could call the IFRs problem "aliasing."

[vk6zgo]

When you think about it, what is a SaH doing when it is sampling a 1 GHz tone with a 700 MHz clock, but mixing it down to a 300 MHz tone?

Well, sort of... The difference being is that instead of multiplying (mixing) with a single sine wave we multiply with an impulse train. This is represented by a convolution with an impulse train as well in the frequency domain which makes the spectrum periodic. Aliasing happens when this periodicity is small thus spectral content of different periods overlap. It's just a different phenomenon that doesn't happen in non-sampled systems because the impulse train multiplication never occurs.

Real world analog mixers don't just mix single sinewaves.
Analog TV signals are 5MHz wide, so that is lot of different frequencies---& the spectrum is periodic.

[tggzzz]

Thus I have a Tek1502 which measures 140ps risetimes while sampling at ~40kS/s. Without aliasing of the sampling signal, it wouldn't work.

I worked with analyzers having a... I don't really know what's the proper name, sampled superhet? architecture. In these the first stage mixer downcoverts to an IF that is set to be in the middle of the 2nd or 3rd Nyquist band of a following digitizer.

Look at Tayloe mixers in SDR receiver dongles.

E.g. 50MHz wide band was donwconverted from RF to between 100MHz-150MHz and then sampled with an ADC clocked at 100MHz. In this case the concept what you guys are arguing is even more apparent because the 2nd stage downconversion to baseband is clearly performed by undersampling. In my mind that still is sort of a nice side effect of process but again I understand that might not be the universal truth.

A standard question I used to ask interviewees is "There is an audio signal on a 10MHz carrier. What is the minimum sampling frequency". The answer isn't 20MS/s. The point is to be careful of the definition of "signal".

[gf]

uua
The  IFR SA problem was that it didn't sweep continuously like the analog one, but, instead "swept" by taking samples over time.
This was perfectly good for most purposes, but a carrier shiftng fast in one direction was displayed as multiple carriers.

In fact, a sweeping SA can capture every frequency in the span only occasionally (i.e. once per sweep, and only for a short duration), while a sampling-based real-time SA captures the full real-time bandwidth continuously without time gaps. Therefore a real-time SA is not supposed to miss anything that happens just intermittently, while a sweeping SA can well miss intermittent frequencies. Or was your sampling SA not a true real-time SA?

EDIT: I just noticed "434MHz to around 477MHz". That's quite much, and even if it is a real-time SA, it might excceed the real-time bandwith supported by the analyzer.

[2N3055]

uua
The  IFR SA problem was that it didn't sweep continuously like the analog one, but, instead "swept" by taking samples over time.
This was perfectly good for most purposes, but a carrier shiftng fast in one direction was displayed as multiple carriers.

In fact, a sweeping SA can capture every frequency in the span only occasionally (i.e. once per sweep, and only for a short duration), while a sampling-based real-time SA captures the full real-time bandwidth continuously without time gaps. Therefore a real-time SA is not supposed to miss anything that happens just intermittently, while a sweeping SA can well miss intermittent frequencies. Or was your sampling SA not a true real-time SA?

EDIT: I just noticed "434MHz to around 477MHz". That's quite much, and even if it is a real-time SA, it might excceed the real-time bandwith supported by the analyzer.

Real time SA has a parameter called instantaneous bandwidth.  That is what is sampled directly, and within that BW it will have 100% POI or close to that....For instance Signal Hound SM200B has I.B. of 160 MHz and SH BB60C has "only" 27MHz.  So SM200B would catch it directly and BB60C would not. But, BB60C would probably show signal sweeping  to 477MHz and you would be able to see what is going on, because despite I.B of "only" 27MHz it has 24 GHz/sec sweep speed in sweep mode.....

[TheUnnamedNewbie]

A standard question I used to ask interviewees is "There is an audio signal on a 10MHz carrier. What is the minimum sampling frequency". The answer isn't 20MS/s. The point is to be careful of the definition of "signal".

Lots of research going into undersampling ADCs, that have S/H that can go to a few times the sample frequency for digital RF/SDR type applications.

[Joel_Dunsmore]

The term of art used for the vernacular "aliasing" is "image-response" when you refer to spectrum analyzers.  And you want your SA to reject images, thus the figure of merit is often called "image rejection".  High performance SAs use two methods of image rejection: 1) up-convert, filter, down-convert (used from DC to mid 3.6-8 GHz range depending on SA).  2) Filter and direct down-converter, usually with a swept tuned YIG filter.  The up/down has the advantage of being very stable, but exposes the front end mixer to all signals in the input range and thus the input converter can be saturated by a nearby signal that is not being measured (e.g. a 900 MHz cell phone transmitter could saturate the front end when looking at a tiny 2.4 GHz wifi signal).  The swept-tuned mode used in microwave bands has the disadvantage that it can be more unstable and has limited bandwidth for real-time acquisitions (typically 40 MHz).  But typically you can bypass the preselector to get wider bandwidths at the expense of being sensitive to images. 

Images come about because the first converter is tuned to an LO freq (say 20 GHz) to measure a signal offset (typically above) to the first converter, mixing it to the IF (e.g. 322 MHz IF).  So when you set the SA center frequency at 20.322 GHz, the LO is at 20 GHz, and IF comes out of the front end mixer at 322 MHz and is digitized.  But if you ad a signal at 19.678 MHz, it would also covert to 322 MHz and show up as a signal and we call this an image; the process is quite similar to alias signals in an ADC.  So we rely on the YIG filter to remove the image.

Or, some versions of SA don't have any image protection, and use digital image rejection to move the LO about, process what shows up in the IF, and sorts out real from image signals.  Turns out, if you do this well you can get very good results; we do it well in the Keysight PNA (option S9309xx).

P.S. This is all discussed, in excruciating details, in Chapter 8 of second edition of my book ( www.tinyurl.com/JoelsMicrowaveBook )