What Spectrum Analyser

[SoundLockEng]

Hi all, been watching the videos for quite a while but thought I would join the forum to ask a few questions.

I have been servicing and repairing pro audio equipment professionally for about 10 years. So far I have been doing this work with a fairly basic set of kit because really it is very easy work and doesn't take any serious tools for fault finding.

I have been designing equipment and would like to look at purchasing a spectrum analyser suitable for audio work along with analysing noise floor measurements as i want to put more attention to detail in my high end design work.

All the spectrum analysers I have seen so far seem to start at the KHz range and I would like to be measuring as close to DC as possible.

I was also thinking of getting a new Siglent SDS2000X series osc too as they seem to give a really good feature set for their price.

Thanks in advance for the suggestions!

[branadic]

You might want to take a look at Stanford Research FFT Analyzers SR760, SR770, SR780, SR785 or Audio Analyzer SR1?

[ap]

The stanford analyzers are good if you want to measure power spectral density (nV/sqrt Hz), but they are pretty rare and often expensive. HP had equivalents, more affordable often, got mine for about 500usd.
For standard audio, and considering also its internal generator, a RS UPL is hard to beat. there is a homepage showing its abilities, but forgot the name and dont have my link list with me here, so you may wish to google arround a little.

[montemcguire]

Don't forget about Audio Precision. While their top of the line analyzer (the APx-555) costs more than a nice car, it can show you a spectrum of the distortion analyzer residual, usually as good as -155dBC for second and third harmonic, depending upon level and frequency. Older models, available used, will have a higher residual, perhaps -130dBC to -140dBC, but are far less expensive.

Devices that only show a single number for THD+N are pretty useless IMHO. An FFT of that residual is far more enlightening: noise, hum and actual distortion can all be tweezered apart. One drawback to some of the older AP analyzers is that they require an ISA card slot on a PC: back in the mid 90s when my old 2322 was new, that was standard on PCs, but it's a little challenging to get one of those together now. That machine will either be extremely expensive or free on trash day.

[Martin.M]

the very amazing UPL (Rohde Schwarz) is out of production 

the actually system of them is the UPV.
my Tip is to rent one for a time and if you like, order 

greetings
Martin

[EEVblog]

The best in the business is Audio Precision.

[Kleinstein]

At the low price end, there are also software solutions that use a PC sound card. Not high end, but can still compete with old instruments in the audio range.  You just miss harmonics above 20 kHz with most sound-cards.

[bson]

You just miss harmonics above 20 kHz with most sound-cards.

That's far from the only potential trap!  The sound card will have an acquisition LPF which may introduce phase nonlinearities/group delay, which needs to be calibrated for if in fact what you want to look at is the group delay of a DUT.  The other is it might have an input DC block in which case it can't tell you about the presence of an offset.  Next, the input impedance may not be well controlled, making it difficult to use a standard probe to poke about a circuit.  You will need a signal generator to properly calibrate the readings, assuming the software supports it.  The weakly controlled input impedance means you will need to recalibrate for each test set whenever any change is made.  Since it's PC software it's not going to play well with general GPIB setups, so say you connect a thermal probe to a bench DMM and want to correlate THD+N with the temperature inside the DUT case or some component, or the delta between multiple components, well the PC software may or may not make this correlation viable.  Finally, there's the question of what DSP processing is done on it before the PC software even gets to see the signal.  Is it going to try to reduce noise or otherwise "improve" the sound?  Keep in mind sound cards are designed for specific use cases, which isn't being used as a test instrument, and something like hiss elimination means you can't use it to measure hiss.

[nctnico]

Maybe the Digilent Analog discovery could also be a start although the dynamic range may be lacking. OTOH I'm quite sure there are USB PC sound interfaces which are designed for this purpose.

[vk6zgo]

Don't forget about Audio Precision. While their top of the line analyzer (the APx-555) costs more than a nice car, it can show you a spectrum of the distortion analyzer residual, usually as good as -155dBC for second and third harmonic, depending upon level and frequency. Older models, available used, will have a higher residual, perhaps -130dBC to -140dBC, but are far less expensive.

Devices that only show a single number for THD+N are pretty useless IMHO. An FFT of that residual is far more enlightening: noise, hum and actual distortion can all be tweezered apart. One drawback to some of the older AP analyzers is that they require an ISA card slot on a PC: back in the mid 90s when my old 2322 was new, that was standard on PCs, but it's a little challenging to get one of those together now. That machine will either be extremely expensive or free on trash day.

If you take measurements for THD + Noise with an input signal ,& Noise alone with the DUT input terminated,you can get a pretty good handle on which is which.
This is the way it was done in Broadcasting for many years.
Interestingly in TV & FM Broadcasting (in Oz),both were measured in dB for convenience in calculating the difference, (THD alone.)
AM Broadcasting stuck with THD in %,making the calculation more difficult.
Noise was in many cases low enough to ignore.

Any reasonable N&D Set or Selective Millivoltmeter will allow you to determine the  frequency content of the Noise.

[Keicar]

I went with an old HP 3562a Dynamic Signal Analyzer for this purpose - and I'd say that the dynamic range (80dB) is only just sufficient to do the job - if you want to analyze really low-levels of distortion you'll need to use a notch filter to remove the fundamental first, otherwise the harmonics of interest will be lost in the noise floor of the instrument.

If you're after a PC-based alternative I'd look at the QA401 from QuantAsylum (https://www.quantasylum.com/content/Products/QA401.aspx) which looks like a pretty attractive proposition for the money. The upside is that the hardware is designed for use as a test instrument, while the downside is that you're locked into using their software, as the instrument doesn't appear as an audio device to the PC (which helps in terms of avoiding unhelpful 'enhancement', of course).

[ElectricGuy]

Taking nctnico proposal, see this thread.
https://www.eevblog.com/forum/projects/analog-disovery-as-audio-analyzer/

[diyaudio]

Hi all, been watching the videos for quite a while but thought I would join the forum to ask a few questions.

I have been servicing and repairing pro audio equipment professionally for about 10 years. So far I have been doing this work with a fairly basic set of kit because really it is very easy work and doesn't take any serious tools for fault finding.

I have been designing equipment and would like to look at purchasing a spectrum analyser suitable for audio work along with analysing noise floor measurements as i want to put more attention to detail in my high end design work.

All the spectrum analysers I have seen so far seem to start at the KHz range and I would like to be measuring as close to DC as possible.

I was also thinking of getting a new Siglent SDS2000X series osc too as they seem to give a really good feature set for their price.

Thanks in advance for the suggestions!

Audio Precision Gear cost big bucks $$$ not normally found in conventional repair departments.

Try this.

https://www.quantasylum.com/content/Products/QA400/QA400Specs.aspx
https://www.quantasylum.com/content/Products/QA400.aspx

[billfernandez]

I recently got a QuantAsylum 401 (their newest audio analyzer) and so far it's working quite well.  Its performance envelope is wider than I need for my audio work at a very affordable price, and seems to have all the features I'll need for testing my audio designs.

[diyaudio]

I recently got a QuantAsylum 401 (their newest audio analyzer) and so far it's working quite well.  Its performance envelope is wider than I need for my audio work at a very affordable price, and seems to have all the features I'll need for testing my audio designs.

I need to get me one as well, haven't been doing audio for a while, I have a up and coming 8 channel DAC project, that may call for one where did you buy yours ?

[billfernandez]

I bought my QuantAsylum 401 directly from quantasylum.com, which as far as I know is the only place to get it.

[exk]

To make audio related measurements (THD, SINAD, IMD etc.) I have used Keysight U8903B. But it's expensive

[bklein]

Wake up old thread...
Not exactly on topic but close.
What would you use to validate a codec's specs?  The U8903B has a serial data interface port on the back - will it work for this?  Does the Audio Precision have support for this too?  What if you have an AQ-401?  (no serial data interface)...