Which Spectrum Analyzer Measures Very Low Frequencies?

[TinyMirrors]

I've been looking at spectrum analyzers, but the problem I'm having is finding equipment that measures in the range of 0.1 to 100khz. The interesting areas for me are between 0.1 and maybe 1-4khz. What device might give me better information for lower frequencies like this. The more modern equipment seem like it's looking at very high frequencies, so I'm stuck trying to find a list of old equipment that could do the job.

Thanks!

[tautech]

Welcome to the forum.

Others are in a similar pickle as most stop at ~10 KHz.
There's a recent thread that might give you some guidance:
https://www.eevblog.com/forum/testgear/looking-for-a-dynamic-signal-analyzer-with-extended-bandwidth/

[innkeeper]

What your looking for is more likely a DSA.
Dynamic Signal Analyzer

I have an older unit, hp 3652a
it is purpose-built for exactly that 64uHz -100khz range.

happy hunting

[w2aew]

I've been looking at spectrum analyzers, but the problem I'm having is finding equipment that measures in the range of 0.1 to 100khz. The interesting areas for me are between 0.1 and maybe 1-4khz. What device might give me better information for lower frequencies like this. The more modern equipment seem like it's looking at very high frequencies, so I'm stuck trying to find a list of old equipment that could do the job.

Thanks!

Most spectrum analyzers are AC coupled and start at 9kHz or higher.  There are exceptions that go down to a few Hz.  Most are pricey though.   The Tek RSA5100 series effectively goes below 1Hz, but again, pricey for the hobbyist. 

Another option is a Dynamic Signal Analyzer.  Still a bit pricey, but this a better fit for your application.

You might be better off using a digital scope with an FFT function.  With such a low frequency requirement, you can take advantages of it's hi-res (in-situ boxcar averaging) mode if it has it, to gain some resolution and dynamic range.

[innkeeper]

Tiny, could you tell us a bit more about what you are going to use it for? What kind of measurements you need, do you need tracking generation, will you be doing distortion analysis etc.

[ejeffrey]

The SRS760/770 is another "audio spectrum analyzer" that can go from sub Hz to 100 kHz.

If you can relax the 100 kHz requirement, a PC sound card may be the cheapest solution and can have quite high performance.

If you need up to 100 kHz but don't want to spend for even a used audio spectrum analyzer / DSA, then you can look for a multi-function DAQ card like the ones from national instruments.  You should be able to find one with a ~1 MS/s ADC, which will be good for up to a few hundred kHz no problem.

[rstofer]

If by 0.1 you mean 0.1 kHz (seems reasonable from context) then the Analog Discovery II might be worth considering.

Download the free software and play with the DEMO Spectrum gadget.  If you want 0.1 kHz to be a major division, the maximum scale will be 1 kHz.  If you can accept it being just 1 tick then 10 kHz.  This isn't really a resolution problem, it is a scaling problem.  If the Stop frequency is higher, ticks are worth more and more and there are only 100 ticks - by definition.

https://store.digilentinc.com/analog-discovery-2-100msps-usb-oscilloscope-logic-analyzer-and-variable-power-supply/

https://store.digilentinc.com/waveforms-previously-waveforms-2015/

[PartialDischarge]

HP 3577A, 3589A, 3588A, 3561, 3562, 35660A, 3585
Anritsu MS4630 and others im not aware

[TinyMirrors]

Hi innkeeper,

We're moving an electro-static mirror. It's really small and delicate. Simply put, it works by applying a voltage across two terminals and the mirror moves accordingly. Since the mirror is so sensitive to voltage changes we are trying to find sources of noise in our electronics that may be causing the mirror to move or vibrate. We want to be able to set a voltage and stay at that exact angle for long periods of time. But if a specific frequency of noise is present the mirror can respond and may oscillate. We've tried using a cheaper scope to look at the fft for noise, but it's just not giving us useful information since each bin is 10hz and it's difficult to see what the frequencies are in the noise. We're looking for noise from the power supply or from other sources that may show up in the signal. We have found a sawtooth wave at a bad frequency for the mirror using the scope, but we're wondering what else we could find with better equipment. We have access to a keithley voltage and current source which can measure the voltage from our power supply as well, but the stability of the voltage is great according to that device.

 
Thanks for the advice!

[TinyMirrors]

I'm terrible at using forums, but thanks everyone for your replies!

[innkeeper]

that is certainly a job suited for a DSA.
I saw MasterTech posted a nice list of DSA's but search DSA or Dynamic Signal Analyzer, and you will find many.
spend some time reading the manuals for them.
HP has a nice application guides and notes too for DSA's that can help you with your specific application, search for those as well.

[JohnnyMalaria]

Why not get an inexpensive ADC and then write some code? I use Python with the Numpy numerical library (all free) and its FFT is insanely fast. For the frequency range you want you can easily achieve real-time processing and the Matplotlib graphics library is very capable. With that combination of software and a tremendous online following, you can achieve great results with just a handful of code. You'll need an ADC with Python support. Mine is from https://www.mccdaq.com/data-acquisition/low-cost-daq - they provide a great Python library. You can capture, FFT and plot in less than a dozen lines.


FYI, a problem with solutions using a PC soundcard is that there's usually a low frequency cutoff at 10-20Hz which may be an issue.

[2N3055]

PicoScope 4262.
You can do FFT up to 1 Mpoints in provided software, and also have API to write your stuff (Visual studio, Python, Matlab).
Top notch instrument for what you need.
Regards,

Sinisa

[Lee Leduc]

Take a look at the Digilent Analog Discovery 2.
https://store.digilentinc.com/analog-discovery-2-100msps-usb-oscilloscope-logic-analyzer-and-variable-power-supply/

Has a spectrum analyzer that may fit your needs.
https://reference.digilentinc.com/learn/instrumentation/tutorials/ad2-spectrum-analyzer/start

I use it in the lab for teaching Analog Op Amps. Very nice little tool for the price.

[innkeeper]

It really depends if he is looking to do more experimentation with stuff, and roll his own, or play with various gear that might work or be writing his own custom code and debugging that OR if he is looking for a new tool that will give him the data he needs to do the task at hand. This dosnt seem like an application where someone wants to deal in maybe's

It seems to me he is looking for the latter.  Cost wise, yea some of the options could be cheaper in capital outlay, but once you factor in your time, probably not so cheap.

In my own personal experience, I was able to acquire a nice OLD 2 channel DSA with a GPIB port, and an NI GPIB  to USB adapter pretty cheap, and frankly has paid for itself 10 times over in time savings. It has its place.

The other thing is that if you investigate other documented issues and testing of such issues, they will most likely be talking about DSA functions, analysis and results. Past research for such things was most likely done with DSA's

a DSA with 2+ channels is really the way I would go for all the above reasons, and if you need to go cheap, get one off ebay that's backed by someone reputable.

again, not that you cant get away with doing it other ways, but, you will always be fighting the equipment to get it to do what a DSA does out of the box..
 

[JohnnyMalaria]

I don't know what the budget is but if you can find an HP3582A, it will do nicely. I used one extensively many moons ago. It's 2-channel digital 0.02Hz-25.6kHz.

[tecman]

My Advantest R9211C covers 0.01 Hz to 100 KHz

paul

[Doctorandus_P]

The latest Siglent oscilloscopes seem to have a pretty impressive FFT capability, and the scope's go down to DC. FFT is much better than you would expect from an 8 bit ADC.

The FFT of my Rigol DS1052E is no more than a toy.

Alternatively you might look into seismic equipment, or PC software such as Octave or Matlab.

[innkeeper]

I have both an hp3562  DSA and a silent sds1104x-e (hacked to 200mhz) and there is absolutely no comparison, the fft function on the sds1104x-e doesn't even come close to what the HP DSA can do.. i mean not even worthy of being in the shadow of the DSA its that far different. 

I suppose if you got nothing is its something, but, frankly from a usability standpoint and doing real readings and analysis... it is a toy compared to the DSA.

[coppice]

Most spectrum analyzers are AC coupled and start at 9kHz or higher.

What is special about 9kHz that makes it the lower bound for so many spectrum analyzers?

[innkeeper]

the first thing I noticed in my hp dsa schematic is the whole front end up to the DAC is DC coupled.
there is a lot of balanced circuitry in there, and although there are a lot of opamps there are some dual fets that are internally matched and will track from a temperature standpoint.
a lot of care is taken for LF noise entering the system.
its a purpose-built system for dc-100khz

grab a schematic for a hp3562 or a 3561 and look at the front end... its interesting.

[Gerhard_dk4xp]

The applicable EMC standards require 9 KHz as a lower measurement limit.

regards, Gerhard

[coppice]

The applicable EMC standards require 9 KHz as a lower measurement limit.

regards, Gerhard

Spectrum analyzers were starting at 9kHz before things like ITU K.76 and CISPR 16 were around.

[JS]

The scope option isn't the best but you probably need an scope or already have one and might cut it for your requirements. Analyzing data in the PC is a great tool. Big memory depth would be your biggest friend for fft to be usable. Capturing a sample from the scope to the PC and running a script is quite easy and doable, I use my Rigol ds1054z and capture directly from matlab or octave via usb or network. I wouldn't even think using a pen drive, I did that in the university where I needed to take the samples hone to play around but doesn't make sense at home.

Now you are just limited by your programming skills (and the 8 bits, but that's quite a lot with 24M of memory, as long fft gets things well under 1LSB)

I think capturing data directly from the computer with an scope is a very very very useful tool for data analysis, the scope is a great tool for a lot of things as stand-alone, but post processing opens a whole different universe, given math functions of the scope are mostly toys. I think THE thing missing from scopes is the lack of user script addons as math functions! Maybe some have but I haven't came across them and I even less know the features and limitations they would have.

JS

[joeqsmith]

My old HP3589A is spec'ed down to 10Hz.   Shown with a 0.5Hz signal using the high resolution narrow band.  It is VERY slow.   Many years ago, I was using an old HP3582 that was designed for this kind of work.  It had two channels and I think it even had GPIB. 

Personally, for the 0.1 to 4KHz you are asking about, I would buy a data acquisition board and use the student version of Labview on a PC.  Then you could do what ever you want.