Learning to use a Spectrum Analyzer

[VSV_electron]

Hello!

I've just got myself Rigol DSA815-TG. I've never worked with a SA before and I'm still on my basic electronics learning path. I have a few questions on how to get into the SA territory and do it safely.

I'd like to start from some basic things, maybe filters characterization? I'm planning to build some RF circuits and then complete receiver/transmitter units when I better understand what I'm doing.

Please advise what cables/connectors I need to start using DSA815-TG.
Should I get some...
https://aliexpress.com/item/4000337127352.html
...N-type male connectors first like the above ones and make some RG58 cables with the other end terminated with the basic SMA/BNC connectors as needed?

Also I know many recommend to always connect DUTs to the SA through a 30dB or even 40 db attenuator to protect the input from accidental overloading. That's of course a too broad generalization but I need to start somewhere.

Any advice on how to safely approach the SA learning path is appreciated.
Also what tasks would you actually recommend to undertake to appreciate the SA basic capabilities?

[shabaz]

That's a sure-fire way of ending up with many junk connectors.

Here's what I would do (not necessarily what others would do, but it's at least a reasonable option).

(1) Get a semi-decent (not AliExpress) SMA-to-N adapter and fit it onto the SA input (and another one on the TG output). Now, the good connectors on the instrument are protected from wear and tear; plus, SMA is more practical when prototyping.

(2) Get a few SMA-male-to-SMA-male cables (RG-316 coax is thin and easy to handle in a lab). There's double-shielded RG-316 options too, a bit more expensive.

(3) Get a 10dB SMA attenuator and fit onto the SA input. Some people attach DC-blocking devices, too; I don't. Depends on how careful you are with what you attach to the SA.

(4) Get a few (personally, I have about fifty because they can be used almost as consumables) SMA female connectors (normal through-hole vertical ones, or end-launch, or both; it's your choice); they are inexpensive from LCSC, and use them to wire your filter, etc, on your PCB.

(5) Connect the SMA cables and enjoy.

[VSV_electron]

...
Here's what I would do (not necessarily what others would do, but it's at least a reasonable option).
...

Thanks much! That's a very organized and comprehensive plan.

[Wallace Gasiewicz]

To measure active circuits, you can use a 10X oscilloscope probe. This can protect your SA input. You can also use the probe with the GND attached to the tip as a "sniffer" or small antenna to find the signals given off by active circuits. You can characterize a probe by just putting it into the circuit, from your SA generator to SA input, I did this and I was surprised that the trace was relatively flat in the frequency range of the probe. If you have a suitable scope probe try it on your SA using the generator on the SA.   You will need N to BNC adaptor for this.   Again, this works only in about the freq range of the probe, but can be quite helpful.
 Most of the time you probably are not concerned about the exact power of the signal, but the frequency is of most concern, you will see the frequency desired and any spurious freq also. 
Measuring the real output of active devices like amplifiers will require attenuators, possibly ones that can deal with several watts.  I do this all the time but I have a setup that puts all the power thru attenuators for radio testing. My usual attenuator handles more than 100 Watts.
Shabaz advice is good, I use some RG 58 cables and the smaller 316 ones are definitely easier to use, most of the time. The SMA to N connectors will probably stay on the SA almost all of the time, so get good ones, the SMA connector side will get a lot of use.   
Sometimes you can pick up used connectors since a lot of these adapters are used by cell tower service people. 
Your idea about starting with passive devices like filters is a good one, you will not overload your SA input. Remember that you are measuring the characteristics of hot only the filter but the connectors and if the filter is on a board the characteristics of what is on the board.     
I needed a SAW filter for a project and was surprised at how inexpensive some of these are on e bay.  Just a suggestion.

[Solder_Junkie]

I too have the same Rigol SA with tracking generator.

Firstly think in terms of dBm levels, use a Watts to dBm chart if necessary.

Secondly, the maximum input level is not where you want to be making harmonic and spurious measurements. Like any receiver, a spectrum analyser will create spurious signals itself if driven too hard. An input level of -10 dBm should be a good starting point.

I have 20 dB and 10 dB “N” attenuators, a Weinschel 6 to 60 dB 4 GHz variable attenuator, a 30 dB 3 GHz 50 Watt attenuator and a home made 40 dB resistive tap for transmitter testing up 100 Watts (use an additional 10 dB or 20 dB attenuator to bring the 50 dBm level down to 0 or -10 dBm). Details at:

http://on4khg.be/wordpress/wp-content/uploads/2015/02/Simple-RF-Power-measurement-W7ZOI-W7PUA.pdf

Beware of trying to measure levels in circuits using oscilloscope probes. A spectrum analyser is a 50 Ohm instrument, you will get misleading results with high impedance cables/probes from oscilloscopes. You can easily damage the input circuit of a spectrum analyser… there is a thread somewhere else on eevblog covering making a high impedance test probe for an SA that also limits the maximum output (sorry I cannot find the thread, but it’s there somewhere).

SJ

[VSV_electron]

...
Firstly think in terms of dBm levels, use a Watts to dBm chart if necessary.

Secondly, the maximum input level is not where you want to be making harmonic and spurious measurements. Like any receiver, a spectrum analyser will create spurious signals itself if driven too hard. An input level of -10 dBm should be a good starting point.
...

The biggest problem for myself as a beginner SA user I see - is correctly identifying the order of magnitude of the input signal and selecting the attenuators accordingly.
I've read in a few threads that even professionals at work manage to overload the SA input sometimes and the SA input circuit (if lucky) gets burned.

The experienced RF (or just all things electronics) man will identify the input signal level to the SA quite easily "at the first sight" but me... I can surely make potentially a bad mistake. Is there a rule of thumb for selecting the attenuator value when starting measurements on the "unknown" circuit?

For instance, I've seen a discussion regarding the TinySA (which of course lacks the TG). One way to characterize filters with it to use a broadband noise generator circuit. However the advice from the TinySA creator was to consider carefully the integral power input level from such a circuit over the whole noise frequency range because it can easily go far above the TinySA input capacity.

I know the input power capacity of Rigol DSA815-TG is noticeably higher then that of the TinySA but I guess that would be a dangerous practice to rely on that relative advantage and not being very careful about estimating the input power level in the first place. That comes with experience of course and again I'm lacking one.

[tggzzz]

For instance, I've seen a discussion regarding the TinySA (which of course lacks the TG). One way to characterize filters with it to use a broadband noise generator circuit. However the advice from the TinySA creator was to consider carefully the integral power input level from such a circuit over the whole noise frequency range because it can easily go far above the TinySA input capacity.

I know the input power capacity of Rigol DSA815-TG is noticeably higher then that of the TinySA but I guess that would be a dangerous practice to rely on that relative advantage and not being very careful about estimating the input power level in the first place. That comes with experience of course and again I'm lacking one.

Just so.

In addition non-linearities (both in the UUT and SA) plus noise => intermodulation distortion product, i.e "spurious signals". Add all these effects together, and the dynamic range is much more limited than with a proper tracking generator.

Also realise that high performance high frequency SAs tend to have small sensitive diodes in the mixer directly exposed to the outside world. These diodes are easily overloaded and burned out by excessive RF power or very little DC voltage. Thus a good quality DC block is a wise investment.

N.B. DC blocks are foolproof but not damnfoolproof: applying a large DC offset to a large DC block capacitor can still lead to a transient current that can be sufficient to damage the SA input.

[Solder_Junkie]

The tracking generator side of the Rigol will not damage the input when sweeping filters, etc.

A 100 Watt amateur transceiver with 50 dB of attenuation will give 0 dBm to your analyser, likewise running a 10 Watt transmitter with 50 dB of attenuation will not be a problem, it will present your analyser with -10 dBm.

Think power attenuators with transmitters.

SJ

[VSV_electron]

...
Also realise that high performance high frequency SAs tend to have small sensitive diodes in the mixer directly exposed to the outside world. These diodes are easily overloaded and burned out by excessive RF power or very little DC voltage. Thus a good quality DC block is a wise investment.

N.B. DC blocks are foolproof but not damnfoolproof: applying a large DC offset to a large DC block capacitor can still lead to a transient current that can be sufficient to damage the SA input.

First - with your reply you reminded me of one of my questions I asked myself couple weeks ago: "What is so special about the 1N5711 diodes that are said to be almost exclusively detector diodes?". I actually was thinking about them being very sensitive with their 0.015 mA current rating but still that question was "hanging in the air" for me.

Regarding the DC block and the attenuators: should I always have two of them connected in series on the SA input when not of course working with the TG?
And then again I need to consider the DC block and the attenuator itself power rating according to the input power level first?
Oh, no I'm not trying to emphasize "this is so complicated!" - why wouldn't it have to be? - but now I need to think what attenuators I need to get, that is - their power rating. And for that I of course need to know what power level my active circuits would output. Otherwise I need to limit myself to low power circuits in the beginning to be on the safe side. Maybe, get some 5 to 10 W DC Block and 30 dB attenuator and start from there? Too much attenuation? - Go down to 20 dB attenuator. Well, I think if the 30 dB or 20 dB attenuator goes first into the test circuit then the DC block doesn't have to be of high power capacity, right?

[VSV_electron]

The tracking generator side of the Rigol will not damage the input when sweeping filters, etc.

A 100 Watt amateur transceiver with 50 dB of attenuation will give 0 dBm to your analyser, likewise running a 10 Watt transmitter with 50 dB of attenuation will not be a problem, it will present your analyser with -10 dBm.

Think power attenuators with transmitters.

SJ

Thanks SJ, good points!

[tggzzz]

Regarding the DC block and the attenuators: should I always have two of them connected in series on the SA input when not of course working with the TG?
And then again I need to consider the DC block and the attenuator itself power rating according to the input power level first?
Oh, no I'm not trying to emphasize "this is so complicated!" - why wouldn't it have to be? - but now I need to think what attenuators I need to get, that is - their power rating. And for that I of course need to know what power level my active circuits would output. Otherwise I need to limit myself to low power circuits in the beginning to be on the safe side. Maybe, get some 5 to 10 W DC Block and 30 dB attenuator and start from there? Too much attenuation? - Go down to 20 dB attenuator. Well, I think if the 30 dB or 20 dB attenuator goes first into the test circuit then the DC block doesn't have to be of high power capacity, right?

How much do you trust yourself, that the UUTs are functioning correctly, the general environment w.r.t. static and power?

Personally I will always have a DC block, since I don't trust myself not to forget and faulty equipment might have a (small-ish) DC offset. On one of my SAs I even put the DC block inside the case (easily reversible if/when I sell it).

Power always needs to be considered; I like a sanity check measurement before attaching the SA. I don't have a power meter per se: I sold my (common) HP432 plus (rare working fragile) thermistor mount/sensor. I do still have some Ge diode "envelope detectors" suitable for a sanity check.

If there is so much power in the environment that it might easily be hazardous to me or the SA, I would consider either a directional coupler to "sniff" a fraction of the power or have a permanently installed attenuator.

One point to consider is that (just like USB connectors) SMA and similar connectors can't have things connected/disconnected indefinitely. That implies using a "sacrificial connector" and/or leaving blocks/attenuators/etc more or less permanently attached. Not a problem for production/field equipment, but it is irritating for lab bench equipment.

[VSV_electron]

...
I do still have some Ge diode "envelope detectors" suitable for a sanity check.
...

Thanks - it all makes sense but could you please elaborate a little on this thing you mentioned about (RF) power "sanity check" using Ge diode "envelope detectors" ?
Asking you for the complete breakdown of the technique is too much of course but if you briefly give me some basic idea on the subject that would be much appreciated.
By "Ge diode" you of course mean Germanium diodes?
We have full unopened boxes of Ge diodes from the times of the Former SU "Germanium power" still floating around among other things. It's hard to say when these huge supplies with cease to exist. They seem to be endless.
I'd like to learn what I can do with the Ge diode to make a power "sanity check".

[tggzzz]

...
I do still have some Ge diode "envelope detectors" suitable for a sanity check.
...

Thanks - it all makes sense but could you please elaborate a little on this thing you mentioned about (RF) power "sanity check" using Ge diode "envelope detectors" ?
Asking you for the complete breakdown of the technique is too much of course but if you briefly give me some basic idea on the subject that would be much appreciated.
By "Ge diode" you of course mean Germanium diodes?
We have full unopened boxes of Ge diodes from the times of the Former SU "Germanium power" still floating around among other things. It's hard to say when these huge supplies with cease to exist. They seem to be endless.
I'd like to learn what I can do with the Ge diode to make a power "sanity check".

Horribly/wonderfully simple This kind of thing
https://en.m.wikipedia.org/wiki/Envelope_detector
http://hparchive.com/Manuals/HP-11096A-Manual-HF-Probe.pdf
https://www.grwiki.org/grwiki/images/6/67/GR-0874-0213-D_1963-11_874-VR-Manual.pdf

But for your purposes, any power measurement tool is useful, since not much accuracy required.

[VSV_electron]

>> Thanks - it all makes sense but could you please elaborate a little on this thing you mentioned about (RF) power
>> "sanity check" using Ge diode "envelope detectors" ?

Horribly/wonderfully simple This kind of thing
https://en.m.wikipedia.org/wiki/Envelope_detector
https://www.grwiki.org/grwiki/images/6/67/GR-0874-0213-D_1963-11_874-VR-Manual.pdf

But for your purposes, any power measurement tool is useful, since not much accuracy required.

Oh, of course I knew what a basic diode envelope detector was I just was intrigued by your mentioning of the Ge diode detector specifically for "power sanity check". Wait.. does it simply mean we are ready to waste a Ge diode when the input power level is above the diode's current capacity? Do I get it wrong? Why does it have to be a Ge diode?

[tggzzz]

>> Thanks - it all makes sense but could you please elaborate a little on this thing you mentioned about (RF) power
>> "sanity check" using Ge diode "envelope detectors" ?

Horribly/wonderfully simple This kind of thing
https://en.m.wikipedia.org/wiki/Envelope_detector
https://www.grwiki.org/grwiki/images/6/67/GR-0874-0213-D_1963-11_874-VR-Manual.pdf

But for your purposes, any power measurement tool is useful, since not much accuracy required.

Oh, of course I knew what a basic diode envelope detector was I just was intrigued by your mentioning of the Ge diode detector specifically for "power sanity check". Wait.. does it simply mean we are ready to waste a Ge diode when the input power level is above the diode's current capacity? Do I get it wrong? Why does it have to be a Ge diode?

(Always difficult to know what someone here does and doesn't know )

It doesn't need to be Ge, of course, but they are common and cheap and have a low forward voltage for low power use. I imagine any low capacitance non-linear device would work.

The HP probe I added allows up to 30V RF.

[VSV_electron]

...
It doesn't need to be Ge, of course, but they are common and cheap ...

And I though about them just the opposite
Those "NOS warehouse diggers" who sell them don't think they are cheap regardless of their quantities.
And you have no way to persuade them they are wrong

[temperance]

I can recommend this channel:



There is much more on the same channel about the use of SA's. But unfortunately not playlist containing all of them in one place.

And of course R&S:

[tggzzz]

...
It doesn't need to be Ge, of course, but they are common and cheap ...

And I though about them just the opposite
Those "NOS warehouse diggers" who sell them don't think they are cheap regardless of their quantities.
And you have no way to persuade them they are wrong

Yeah. I should have written relatively cheap compared with HP thermistor sensors

I saw one of those at a hamfest for £70 (with the meter being another £20). Having sold one for £200, I was tempted - but couldn't see it working.

[RAPo]

I always use a DC-block.
For my SA(ultra) I use a RF power snitch,giving me an indication of the power
And then I have a set of attenuators in order to bring it to safe levels.

Regarding the DC block and the attenuators: should I always have two of them connected in series on the SA input when not of course working with the TG?
And then again I need to consider the DC block and the attenuator itself power rating according to the input power level first?
Oh, no I'm not trying to emphasize "this is so complicated!" - why wouldn't it have to be? - but now I need to think what attenuators I need to get, that is - their power rating. And for that I of course need to know what power level my active circuits would output. Otherwise I need to limit myself to low power circuits in the beginning to be on the safe side. Maybe, get some 5 to 10 W DC Block and 30 dB attenuator and start from there? Too much attenuation? - Go down to 20 dB attenuator. Well, I think if the 30 dB or 20 dB attenuator goes first into the test circuit then the DC block doesn't have to be of high power capacity, right?

How much do you trust yourself, that the UUTs are functioning correctly, the general environment w.r.t. static and power?

Personally I will always have a DC block, since I don't trust myself not to forget and faulty equipment might have a (small-ish) DC offset. On one of my SAs I even put the DC block inside the case (easily reversible if/when I sell it).

Power always needs to be considered; I like a sanity check measurement before attaching the SA. I don't have a power meter per se: I sold my (common) HP432 plus (rare working fragile) thermistor mount/sensor. I do still have some Ge diode "envelope detectors" suitable for a sanity check.

If there is so much power in the environment that it might easily be hazardous to me or the SA, I would consider either a directional coupler to "sniff" a fraction of the power or have a permanently installed attenuator.

One point to consider is that (just like USB connectors) SMA and similar connectors can't have things connected/disconnected indefinitely. That implies using a "sacrificial connector" and/or leaving blocks/attenuators/etc more or less permanently attached. Not a problem for production/field equipment, but it is irritating for lab bench equipment.

[VSV_electron]

I always use a DC-block.
For my SA(ultra) I use a RF power snitch,giving me an indication of the power
And then I have a set of attenuators in order to bring it to safe levels.
...

I wonder if a simple DIY tool like the well known "By Wes Hayward, W7ZOI, and Bob Larkin, W7PUA", Simple RF power meter would work fine? The file is everywhere over the WWW but I 'm attaching it for reference here.
There are also separate building hints from WH available because many did it wrong using a perfboard.

[shabaz]

That device doesn't seem to operate beyond 1 GHz.
If there's even just a slim chance that the power level will exceed your SA, then you should not rely on such a tool, you should simply not attach the EUT to the SA, and rethink what you're testing and how.

You could even use a DIY H-Field probe and a dummy load, if you are not concerned with seeing accurate power values.

The SA has huge dynamic range, and even if you attenuate the signal more than necessary, it will still likely be visible. Better to initially have lots of attenuation (and use a directional coupler and a load if you're working with many tens of W or more). 2W MiniCircuit (or similar) attenuators at least are cheap. They get a lot more expensive approaching 50W or greater (don't rely on an AliExpress one! that's a recipe for destroying the SA).

Personally I used a cheap (ancient) HP SA first, before using a decent SA. I have not destroyed any SA so far (including the ancient one). If you're really unsure of yourself, then all I can suggest is that it may be worth gaining some experience with a TinySA (I have not used one, so I'm speculating).

[Wallace Gasiewicz]

I do not know what you are trying to measure. If it the output of a radio then I am with shabaz.  you can use a dummy load and just put a few wraps of wire around the outside of the coax feeding the load.. You will get a spectrum of the radio output.  Of course it will not be calibrated for any sort of watt output of the radio but it will give you a nice spectrum of the output and a good frequency reading.     
There are all sorts of other configurations that can give you a sample of transmit power, but just a few wraps of wire o the outside of your transmission coax will give you a nice signal. Some loads even have a connection for a much attenuated sample of their input.  Some of these things are linear enough to give you reasonable watt output readings.     

If you need to know the output of a radio, you really need a RF in line power meter with a 50 ohm load. Lots safer.

[pdenisowski]

And of course R&S:

Thanks!  One of my most popular videos: 

[pdenisowski]

The biggest problem for myself as a beginner SA user I see - is correctly identifying the order of magnitude of the input signal and selecting the attenuators accordingly.
I've read in a few threads that even professionals at work manage to overload the SA input sometimes and the SA input circuit (if lucky) gets burned.

The experienced RF (or just all things electronics) man will identify the input signal level to the SA quite easily "at the first sight" but me... I can surely make potentially a bad mistake. Is there a rule of thumb for selecting the attenuator value when starting measurements on the "unknown" circuit?

As someone who trains "professionals" on "professional" spec ans:  our service department doesn't get a lot of spec ans with blown out front ends.  That may be because our customers are more careful than hobbyists, but frankly, I've seen a lot of "professionals" (myself very much included) do some ... unadvisable ... things with test equipment.

Most professionals have a pretty good idea of what powers they are going to be measuring and simply add an appropriate (external) pad where needed.  I also know quite a few customers who will use an RF power sensor when in doubt, since blowing up one of those is less costly (but again, it's still pretty rare).

If you're just interested in learning how to operate a spec an, an easy and very safe way to get test signals is just to get a cheap antenna and look at the FM broadcast band or some cellular or ISM band (e.g. 2.4 GHz, etc.).  Lots of signals to practice on, no expensive equipment needed, and essentially zero chance of damaging anything. 

(When I do internal training, I normally use a vector signal generator and ARB files to provide the signal to the spec an, but clearly this would be cost-prohibitive for most hobbyists).

[pdenisowski]

We also have a spec an teaching kit for basic spectrum analyzer training, but it's also probably priced outside of most hobbyist's budgets. 

https://www.rohde-schwarz.com/us/manual/r-s-fpc-z10-teaching-kit-getting-started-manuals_78701-584833.html