Spectrum analyzer- how much useful do you consider it?

[Jane]

As I can see  quite a lot of users here have an experience ( or use regularly) a spectrum analyzer.
As It is still one of the missing equipments in my lab I would like to ask how much useful a spectrum analyzer is for you? For what do you use it? ( From time to time I also repair a digital tuners and not sure if it would be useful or not necessary)
Thank you for your feedbacks

[Howardlong]

I manufacture some products with single chip digital tuners, but because the entire part is integrated, including VCO, PLL, divider chain, mixer, LNA and IF/ZF/baseband amps, a spectrum analyser wouldn't tell you anything at all as you can't get into the chip anyway.

What does help though is an accurate RF signal generator, so you can check things like sensitivity, phase noise and image rejection. Both of these are analysed at baseband, using the devices' internal high resolution ADC streamed to a PC over USB. On one of my products, insufficient image rejection is the biggest single failure mode. On another it happens to be excessive phase noise.

I do use spectrum analysers, but not for fully integrated digital tuners. More usually it's to check an RF PA's spectral purity or, with an appropriate probe or two, pre-compliance testing for a new product.

I don't use them to spy on people or search out for aliens or any weird stuff like that, but occasionally, if your not sure if something's working or not, sniffing out a known internal frequency like a reference oscillator on a DUT with an appropriate probe on an SA can be enlightening when fault finding.

[Solder_Junkie]

As a radio ham who builds quite a lot of my own radio gear, I find a spectrum analyser very useful to align filters, check oscillators for the correct levels and transmitters for spurious outputs.

I have two, one is the Rigol DSA815 with tracking generator and the other a battery powered hand held that's good for tracing interference such as plug top switched mode chargers.

[georges80]

For my specific use I have a DSA815-TG and a DC LISN and a Beehive probe.

The setup has paid for itself as an aid to drastically reducing EMC on several of my products. It makes it very easy to evaluate changes on a prototype prior to going to production (position of power filters and routing/ground plane topology).

I purchased my SA specifically for this task. Hopefully it'll come in handy for other projects as time goes by.

cheers,
george.

[Neganur]

filter characteristics, power mask, harmonics/spurious, IP3 of components- primarily for measurements around receiver/transmitter systems either when you develop your own circuits or verify (document) the performance of something you sell.

If you're not really doing much with higher¹ frequencies or radio stuff you can probably focus on other more useful instruments. Lately, I find myself wanting a current probe much more than an SA at home. I don't see Dave use an SA much for example. It's one of those things that can potentially be an expensive dust collector if you're not really needing one. An SA is a nice instrument though if you have everything else and have some spare cash.


¹: higher as in, much more than your oscilloscope can handle or cheat with FFT

[SArepairman]

I do not have a spectrum analyzer or dynamic signal analyzer in my lab and I feel blind. 


It depends on what you like doing.

[VK5RC]

As a Ham I actually use a SA more than an oscilloscope, I have two, one good one with a tracking generator to 6GHz, another not quite so good to 20GHz, mixers to 50GHz.

[Jane]

Thank you all for your  feedbacks
As far as I understand because  I repair DIGITAL tuners, not analogue, SA is of no value for me and would not help. Is it so or I am wrong?
@Howardlong

What does help though is an accurate RF signal generator, so you can check things like sensitivity, phase noise and image rejection. Both of these are analysed at baseband, using the devices' internal high resolution ADC streamed to a PC over USB. On one of my products, insufficient image rejection is the biggest single failure mode. On another it happens to be excessive phase noise.

Can you please provide further details how RF signal generator can be more useful?
Thank you

[Howardlong]

Thank you all for your  feedbacks
As far as I understand because  I repair DIGITAL tuners, not analogue, SA is of no value for me and would not help. Is it so or I am wrong?
@Howardlong

What does help though is an accurate RF signal generator, so you can check things like sensitivity, phase noise and image rejection. Both of these are analysed at baseband, using the devices' internal high resolution ADC streamed to a PC over USB. On one of my products, insufficient image rejection is the biggest single failure mode. On another it happens to be excessive phase noise.

Can you please provide further details how RF signal generator can be more useful?
Thank you

If the tuner is integrated into a single ship, then I'd say an SA is of limited use as you can only access the baseband or low IF signal, a scope will often be fine for that. However, if the device is made up of  separate parts, e.g. the local oscillator, mixers, LNA, etc then a spectrum analyser will certainly be useful as you can trace through the chain where a fault lies.

Irrespective, an RF signal generator, is always useful whether or not you have an SA as you have a way of measuring objectively the performance of the tuner. Relying on off-air signals is very hit and miss, you'd be lucky to measure sensitivity for example to within an accuracy 10dB or so.

Although it's not often thought of as a measuring and test instrument, I also frequently use an SDR in conjunction with an RF signal generator with appropriate probes to trace RF signals in receivers rather than an SA, the visual response is much quicker than an SA and if you want you can hear the signal too, a bit like a continuity test for RF. You can also hear the noise, so can make a rough aural judgement if a stage has relative gain or loss to some degree. You can locate faults at the component level with this method. SDRs are also a lot cheaper than an SA, but they lack absolute measurement capability and have limited spans. SDRs do tend to have much better sensitivity than SAs though.

[tautech]

It's not just a case of will be of use but:
Do you know how and when?

Plus the additional cost of:
Cableing
Adaptors
Waveguides
Attenuators

SA's are a significant investment, worthy of course if you have the need.

Yes they are the answer for some work like that Bud and MarkL did in this thread:
https://www.eevblog.com/forum/blog/eevblog-683-rigol-ds1000z-ds2000-oscilloscope-jitter-problems/msg563524/#msg563524

It seems the Rigol DSA815 is the current flavour, but add the TG.

Don't have a SA myself, never ventured that far into black magic territory.

[Andy2]

Like Solder Junkie, I'm a Radio Amateur with a long history of home-brew. Over the last 40 years I've built receivers, transmitters, converters and all manner of other stuff. It amazes me now that before I had a SA I would just trust to luck, tune for maximum smoke and go on the air! These days I aim for -70dBc for all my transmitter unwanteds. I use two SA's, both with TG's and I honestly don't know what I'd do without them. The Signal Hound SA44b (now with software that works...) has quite passable phase noise performance (around -120dBc at 10KHz) and can be used to set up narrow filters etc. The other is a Rigol which is a more convenient instrument but is more suited to sprog-finding etc as its poor phase noise (-85dBc at 10KHz) limits its close-in performance.
They can go unused for days or weeks at a time but when I need one, I need it!

[Neganur]

I also frequently use an SDR in conjunction with an RF signal generator with appropriate probes to trace RF signals in receivers rather than an SA, [...]

I can second that, they're incredibly useful if you just want to see if a signal is there and approximately where it is. Say, you want to see if your device switched to frequency hopping or not, or if anything else is in the air that could interfere. Most of the available devices I have seen only work in popular bands like 800/900/2.4G/5GHz, but I'm sure there are more versatile ones around since it's a growing market.

[Howardlong]

There are certainly several SDRs that cover a lot more than just a few ISM bands, from LF through to GHz.

They are primarily designed as radios though, and when used as test equipment one should be aware of that. This is not necessarioy a limitation, it's just that they aren't designed to give accurate absoulute measurements, and they are not accurately normalised in the frequency domain in the way that an SA typically is. As a radio, down at HF and below there is so much ambient noise that paying attention to getting single digit noise figure in a receiver would be nothing more than an academic exercise with no practical benefit. The same is not the case at VHF and above where single digit noise figure is expected and demanded. In an SA on the other hand more attention is given to maintaining a constant and accurate absolute reference across the frequency range, and sensitivity is very much a secondary consideration.

Having said that, SDRs can certainly give accurate relative measurements at a given frequency and within a certain limited span, and give accurate sensitivity and noise figure measurements too with the right software (and a calibrated noise source in the case of measuring noise figure). If you have an accurate RF signal generator, you can use that as the reference for your SDR to give an absolute. measurement.

In general, I'd say an RF signal generator is generally going to get more use than an SA, but that's like comparing apples and oranges. And anyway, an RF signal generator doesn't look as sexy as a Spec An on the bench.

FWIW I use an Anritsu MT8802A with the spec an option, option 7. They offer a 300kHz to 3GHz range for RF signal generation, RF power measurement and spectrum analysis in one package. They're frequently on ebay, they were bought by cell phone manufacturers in large quantities, and that's reflected in the user interface, but they are also extremely useful for basic RF analysis, production testing and fault finding. It's the RF signal generator that gets used by far the most out of its functionality.

[Bud]

Often times, parasitic oscillation in rf amplifiers can be identified with a spectrum analyzer, and for UHF amplifiers it can be in Ghz region.

[Howardlong]

Often times, parasitic oscillation in rf amplifiers can be identified with a spectrum analyzer, and for UHF amplifiers it can be in Ghz region.

While that is certainly true, the OP is dealing with digital tuner repair, I am not sure that parasitic oscillation is going to be a common enough failure mode to justify an SA purchase on its own.

Actually I'm not sure that my understanding of "digital tuner repair" matches the OP's yet. In my world a digital tuner these days is a single chip direct conversion or low IF device primarily designed for satellite, cable and terrestrial TV reseption. If I have misunderstood, please correct me!

[Jane]

Thank you for  replies. As I need to work/repair DIGITAL satellite tuners I do not need spectrum analyzer but only RF generator and SDR
But not sure what models of  RF generator and SDR.
Can you please suggest?
Thank you.

[Jane]

Here is a picture of the tuner I deal with. I think it is a digital tuner. So, now if I have RF generator
 what will be a procedure to find out which part( in tuner) is faulty?
I 'd say, I must start satellite receiver, connect RF generator to tuner( instead of the antenna) and use SDR to measure signal on the path starting from  antenna connector through tuner chip path  to demodulator's data pins
Is is so or there is a different ( better) procedure?
Thank you for advice.

[Howardlong]

The right hand side is the RF section, and that feeds into the middle section which looks like either a low IF downconverter or zero IF (ie, direct conversion) with quadrature differential outputs. You can see that from the four ceramic 0603 caps adjacent to each other, they indicate a quadrature differential output. Either way it's a single chip tuner, with no external LO parts to look at.

So you can put a reasonably large carier signal on the tuner input, perhaps -50dBm or so, and assuming it's tuned to the right frequency, you should see the frequency shifted signal on each of the four lines with a basic oscilloscope.

Make sure that there is a sufficuent DC block on the RF signal generator, the signalling up to the LNB iSTR is about 18v.

Remember that the signal at the F connectors is going to be between about 900MHz and 2GHz, there is already a first downconversion in the LNB itself.

Unless you have some service documentation, I would get some measurements from a known working unit first and use those as the basis for a test procedure.

[Jane]

Thank you for your reply.
I found  a datasheet of the tuner chipset( CX24118a)
( http://media.digikey.com/pdf/Data%20Sheets/NXP%20PDFs/CX24118A.pdf)
 and those pins with capacitors, numbered from 15 to 18, are marked as
Output The negative differential I channel output to demodulator. Zout = 1 k? // 10 pF.
Output The positive differential I channel output to demodulator. Zout = 1 k? // 10 pF.
 Output The positive differential Q channel output to demodulator. Zout = 1 k? // 10 pF.
 Output The negative differential Q channel output to demodulator. Zout = 1 k? // 10 pF

Can you please let me know what shift and what frequency I should see on those caps? And is a common 200 Mhz oscilloscope good enough for such task?
Thanks again for you explanation and time

[Neganur]

Not sure what frequencies it uses (at the base band I mean) but you should see two sine waves I/Q at 90 degrees and a frequency at the difference between the input sine and the frequency the receiver is tuned to (some 100's kHz I guess, again no idea what the filters are) A 200 MHz scope is sufficient, I think the document said there's a baseband filter that has 100 MHz as highest band limit.

[Howardlong]

Looks like it's a zero IF tuner from the datasheet. Your average transponder bandwidth is 36MHz, so you'd need to be able to see +/-18MHz as it's a zero IF tuner, so the baseband is viewable on all but the most useless scopes.

Continuing on from what Naganur said, if the LO of the tuner is 1GHz, and your signal generator pumps in 999MHz, then you should see the difference, ie 1MHz, on each of the four outputs. There will, however, be 90 degrees between each (or near enough that a casual viewing on the scope wouldn't be able to tell any difference).

A spectrum analyser or SDR would help to determine if any of the front end parts are broken in the RF section. Typically there are often sacrificial parts here, such as diode clamps. By direct probing with an RF signal connected you can find shorted or open parts. You are often looking for ball park figures in direct RF probing, within a dB or two is often OK, 10dB usually isn't.

[EEVblog]

Depends entirely what you are into.
Some design engineers can go their entire career without ever needing to use a SA. Others might use it every day.
In a previous job we used a Dynamic Signal Anayser more than a scope or SA.

[Jane]

Thank you all for your feedbacks.
As Howardlong said,  SDR   in conjunction with an RF signal generator with appropriate probes to trace RF signals would be good enough for my needs( or at least as the beginning).
But can anyone suggest what kind of SDR and the appropriate probes I should  use?
Thank you.