Low frequencies on Spectrum Analyzers

[jgalak]

My next major test gear purchase is going to be a Spectrum Analyzer.  Probably either a Rigol DSA815-TG or a Siglent SSA3021X.  Both are priced similair, and both fit my budget (let's please not start a fight over which one to pick, that'll be a different thread   ).

My use is mostly for VHF and UHF radio use, 140 and 440 MHz, so both have adequate bandwidth.

However, I also occasionally need to look at low frequency audio signals (like, say APRS tones at 1.2 and 2.2 kHz) and I've noticed that both SpecAns have a range that only goes down to 9kHz. 

Is this normal?  Are there any SpecAns in this price bracket (sub-$2k) that go down lower?  Or is this inherent in this sort of SpecAn design?

I think I can use my Analog Discovery 2 in SpecAn mode for the low frequencies (though a quick google search didn't find a range specification), but that seems annoying - I was hoping a single device could do both.

[wn1fju]

Most of the "RF" spectrum analyzers have to worry about local oscillator feedthrough and can't get too close to DC.  If you are willing to go old, the
HP 3561A or HP 3562A are very nice 0-100 kHz analyzers.

Or in a pinch, I've always found that using the sound card of a computer and a freeware FFT program gives very reasonable results.

[DaJMasta]

While most SAs only go down to 9kHz or so, there are a number of models that have extended range, down to 100 or 20Hz.  A good example is some of the R&S FSE series analyzers, some of the current Keysight models, some Advantest models, etc.  They are out there, but they seem to be less common - at least if you're buying new, though, I don't know of too many options that go that low.  That said, if you don't need it all on one instrument, there are many options for audio band FFT measurements and many are inexpensive.

[nctnico]

For low frequency stuff I'd get a DSO with good FFT unless you need a large dynamic range. Even if a spectrum analyser supports really low frequencies the performance is not very good. For example: the DANL for my (higher end) spectrum analyser is specified at -125dBm at 10kHz and -156dBm at 1Ghz.

[jpb]

As people have said, audio solutions are relatively cheap - much cheaper than a spectrum analyser and will work a lot better, e.g.
https://quantasylum.com/products/qa401-audio-analyzer
(I don't own the above but there has been quite a lot of discussions on these forums).

If you're spending $2k on an analyser, spending <$500 on an audio solution (probably more like < $200 if you go for a usb audio interface or sound card) is only 10 -20% of the cost.

[jgalak]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

[srce]

Oscilloscopes don't have great dynamic range.

I'm using a Lynx E22 soundcard in to MatLab.

[SWR]

I recently had to make the exact same decisions as you are facing with very similar applications.

I chose the Siglent SSA3021X based on better specs and solid reviews, and use it for RF data communication (LoRa 170/433/868Mhz, WiFi/GPS (2,4GHz: easy mod to 3,2GHz) and EMC pre-compliance.

For the low frequency stuff and general debug I chose a GW Instek MDO-2204EX multi instrument with 4ch 200MHz oscilloscope, DC-500MHz spectrum analyzer, dual arbitrary waveform generator, digital multimeter and dual power supply. The SA has >80dB dynamic range which is about 40dB less than the Siglent SSA at low frequencies.

Those two instruments alone cover about 95% of my measurement needs. 

[nctnico]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

The DS1054Z has useless FFT. Get a scope with 1MPts FFT and it will be usefull. Even for closely spaced frequencies. The problem with an oscilloscope is that the FFT frequency bins depend on the samplerate. With longer FFT you'll get a decent frequency resolution without needing to be very specific about the samplerate. With short FFT you'll need to set a very low samplerate to look at low frequencies.

[SWR]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

I don't think you're using it wrong. The FFT on these units leaves a lot to be desired.
You really need a fast updating higher resolution FFT like a million points to get closer to the feel of working with a real SA. The user interface on a spectrum analyzer is much easier to work with than a FFT math function. The newer GW Instek scopes have this interface, and maybe others do as well?

[SWR]

Oscilloscopes don't have great dynamic range.

I'm using a Lynx E22 soundcard in to MatLab.

That's not a bad idea if you don't mind dragging along a stationary PC.
You get 117dB dynamic range but with inflexible attenuation, so you might loose some of it depending on how well your signal maches the line level input. You're propably not going to loose 30dB though.

[jgalak]

I do have the AD2, which has a SpecAn mode.  I've been unable to get good spec on it, but it seems to go down all the way to DC.  How good it is I can't tell.  Gotta be better than a soundcard, right?  Or not...

[technogeeky]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

I'll have to check later, but the best you can get for low-ish frequency FFTs is to start the FFT, go into the 2nd page of options, and go to longer and longer timebases until the option Trace greys out. Then, go back one setting to a shorter timebase, and select Memory. This will give you your best chance of seeing low frequency signals.

I also want to add: As exemplified by Shahriar @ The Signal Path, some (older for sure, but some newer) spectrum analyzers claim to have some lowest frequency range (for example, 9kHz). Some of them are restricted (in software) to not go below this. However, some of them (in this example the DSA1030A) will actually display tones well below that limit with the caveat that the amplitude accuracy is no longer guaranteed. This video shows tones as low as 100Hz.

[technogeeky]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

The DS1054Z has useless FFT. Get a scope with 1MPts FFT and it will be usefull. Even for closely spaced frequencies. The problem with an oscilloscope is that the FFT frequency bins depend on the samplerate. With longer FFT you'll get a decent frequency resolution without needing to be very specific about the samplerate. With short FFT you'll need to set a very low samplerate to look at low frequencies.

It's still not great, but I think with Memory mode things aren't as useless as they once were. However, searching for the depth just now reveals this might only be 16k points which is pretty atrocious. In any case, whenever I use the Memory option on my 1054z I am able to get results -- even when down at power line frequencies like 60Hz (buried in other audio frequency content).

You can, of course, save and transfer waveforms to a computer and perform FFT which will give you arbitrarily good results up to the poor resolution of the 8-bit ADC.

I do agree that other oscilloscopes (perhaps the newer Siglents, but surely the Instek) provide pretty damn good FFT options for a oscilloscope.

I am not nearly as expert as you are, so please feel free to correct anything I say.

[G0HZU]

All of my big old classic (boatanchor) lab RF (GHz+) spectrum analysers go down close to DC. I think they all go down to about 50-100Hz but because they are swept analysers with analogue RBW filters the sweep time will be slow on something like a 10kHz span with a low RBW.

They also maintain a low noise floor down to a few kHz (especially the HP8568B) but I still wouldn't recommend using them like this for audio work because of the limited distortion performance of the analyser (compared to a soundcard) and the slow sweep speed. It quickly becomes tedious to use the analyser like this.

Probably the best RF spectrum analyser I have here for LF work is my old 8GHz Tek RSA3408A as this has a 14bit ADC and can cover from a few Hz up to about 40MHz in its 'direct to ADC' mode. So it's a bit like having a 14 bit soundcard that can cover up to 40MHz rather than just 20kHz. But it does have a few low level spurious signals in this range because of the limitations of the ADC etc.

[llkiwi2006]

For low frequency stuff I'd get a DSO with good FFT

I have a DSO (Rigol DS1054Z) but I've been very unhappy with the FFT on it for this sort of thing - I can't seem to zoom in far enough to be useful.  The minimum scale seems to be in the multiple MHz range, so seeing things in the sub-10kHz range is impossible.  Unless I'm using it wrong, of course...

The FFT on the 1054z sucks, but you can see down into the kHz range. You just need to fiddle around with the memory depth and timebase to get a low enough sample rate (maybe around a few hundred kSPS). Also set FFT to process memory instead of trace. But ultimately the dynamic range is terrible, so you might get better results just using your pc soundcard.

[egonotto]

I do have the AD2, which has a SpecAn mode.  I've been unable to get good spec on it, but it seems to go down all the way to DC.  How good it is I can't tell.  Gotta be better than a soundcard, right?  Or not...

Hello,

I don't understand this. Analog discovery has a good SA mode. What have you and what expect you?

Best regards
 

[CalMachine]

You want an SR780!



http://www.thinksrs.com/downloads/PDFs/Manuals/SR780m.pdf

[jgalak]

I do have the AD2, which has a SpecAn mode.  I've been unable to get good spec on it, but it seems to go down all the way to DC.  How good it is I can't tell.  Gotta be better than a soundcard, right?  Or not...

Hello,

I don't understand this. Analog discovery has a good SA mode. What have you and what expect you?

Best regards

What I mean is that I can't find any detailed specifications on the SA mode of the AD2.  I am assuming it's better than a sound card, but don't know that for sure.

[tecman]

I have an Advantest R9211C.  You can occasionally find one for a good price on ebay.  Nearly identical specs to the SRS or HP 35770.  Only complaint is that the interface is not as intuitive as I would like.  Same problem on the RF spectrum analyzers.  But performance to price ratio (on used) is very good.

paul

[thm_w]

What I mean is that I can't find any detailed specifications on the SA mode of the AD2.  I am assuming it's better than a sound card, but don't know that for sure.

Its better in terms of sample rate/bandwidth/input range, but worse in terms of SNR and THD.

AD: Two-channel oscilloscope (1M?, ±25V, differential, 14 bit, 100Msample/sec, 5MHz bandwidth). SNR = 75dBfs
Decent sound card: 24-bit 192kHz, probably >100dB SNR.

https://reference.digilentinc.com/_media/analog_discovery:analog_discovery_rm.pdf

[jpb]

You want an SR780!



http://www.thinksrs.com/downloads/PDFs/Manuals/SR780m.pdf

There is a SR760 for sale on ebay UK, in fact it has been for sale for quite a while which isn't surprising as they want a lot of money for it:
https://www.ebay.co.uk/itm/Stanford-Research-Systems-SR760-FFT-Spectrum-Analyzer-100-khz/272851949264?hash=item3f873e56d0:g:xrwAAOSwgjFZukt~

[Co6aka]

As people have said, audio solutions are relatively cheap - much cheaper than a spectrum analyser and will work a lot better, e.g.
https://quantasylum.com/products/qa401-audio-analyzer

FWIW, I have a QA401 and am quite pleased with it. 

[David Hess]

Is this normal?  Are there any SpecAns in this price bracket (sub-$2k) that go down lower?  Or is this inherent in this sort of SpecAn design?

There are two big reasons:

1. The first mixer is usually AC coupled through the transformer.  Some spectrum analyzers have an option to swap the mixer inputs for operation down to DC but this risks damage to the mixer if DC is applied.

2. Close in phase noise from the first local oscillator limits dynamic range at low frequencies anyway.  Early DDS shortwave receivers had lower frequency limits much higher than 9kHz because this lowered their sensitivity so much.

[joeqsmith]

Seems like you should use this as justification to buy more test gear later on. 

For low frequency, I have an old HP3589A spectrum/network analyzer that will is spec'ed for 10Hz to 150MHz.   It's big, weighs a lot and is nothing fancy.  I would think these things are fairly cheap now days.