Dynamic Signal Analysers/FFT Analyser Recommendations - Money no object

[David]

Hi all,

I am looking to purchase some new Dynamic Signal Analysers for our lab at work. We have been using HP 3562A's (and similar) for years but they are really showing their age and keeping them going is becoming harder.

HP don't have any replacement DSAs and finding new models seems to be harder than I had hoped.

Money is essntially no object.

Key requirements:
* Must be a portable self-contained unit (these are used in both development and production environments and having a PC based unit really would be a last resort)
* Must have a signal generator option to allow frequency response measurements
* Must work down as close to DC as possible (<100uHz) - upper end around 5kHz
* Noise performance must be equivalent or better than the HP 3562A

We design low noise amplifiers with ~1nV/rtHz @1Hz noise floors and very low 1/f noise. We can use an external LNA where we need to.

I have seen some ONO SOKKI FFT analysers which may do the job. The Stanford units look OK but have a very outdated interface. I'd like something without a CRT!

Cheers,

Dave

[Sighound36]

Hi Dave

The most accurate SA I've used so far by a country mile is the Keysight N9042B  ultra low phase noise of -135dBc big real time bandwidths up to 4Ghz lol not used the new R&S version

https://www.keysight.com/gb/en/product/N9042B/uxa-signal-analyzer-2-hz-50-ghz.html

Although 2Hz is the bottom limit and possibly not low enough for you

A left field option is the blended multi bit ADC SA Marku pro


https://www.liquidinstruments.com/

[Electrole]

Take a look at https://www.bksv.com/en/instruments/daq-data-acquisition/lan-xi-daq-system
They are well-known within sound and vibration, and some modules should go down to DC.
Perhaps a configuration would suit your requirements?
I have no affiliation with the company, BTW, but I have used their products.

[hpw]

Key requirements:
* Must be a portable self-contained unit (these are used in both development and production environments and having a PC based unit really would be a last resort)
* Must have a signal generator option to allow frequency response measurements
* Must work down as close to DC as possible (<100uHz) - upper end around 5kHz
* Noise performance must be equivalent or better than the HP 3562A

We design low noise amplifiers with ~1nV/rtHz @1Hz noise floors and very low 1/f noise. We can use an external LNA where we need to.

As you like to have data below 0.000xxx Hz ... as your HP3562A and others are FFT based.

This means

a) sample rate dependent
b) sample size dependent to get the required bin freq. step.
c) a good ADC with good input stage and ADC reference stage and in addition LDO power.

Just look picture below (log scaled from lowest  FFT bin) , as SR 48kHz and 32 Mio FFT points is not good enough for you requirements.

So lowering the sample rate is one option (as the ADC supports).

Also to know/understand, that the sample time will be anyway long.

my 2 cents

Hp

[Someone]

R&S discontinued their analysers, the R&S UPV is probably what you were imagining.

Computer based is the way to go (since thats basically what they became) and you'd be surprised at the performance of recent scopes and signal generators, but if the signals are well constrained in amplitudes then audio interfaces are a cheaper option.

[CorporateReference]

Hi all,

I am looking to purchase some new Dynamic Signal Analysers for our lab at work. We have been using HP 3562A's (and similar) for years but they are really showing their age and keeping them going is becoming harder.

HP don't have any replacement DSAs and finding new models seems to be harder than I had hoped.

Money is essntially no object.

Key requirements:
* Must be a portable self-contained unit (these are used in both development and production environments and having a PC based unit really would be a last resort)
* Must have a signal generator option to allow frequency response measurements
* Must work down as close to DC as possible (<100uHz) - upper end around 5kHz
* Noise performance must be equivalent or better than the HP 3562A

We design low noise amplifiers with ~1nV/rtHz @1Hz noise floors and very low 1/f noise. We can use an external LNA where we need to.

I have seen some ONO SOKKI FFT analysers which may do the job. The Stanford units look OK but have a very outdated interface. I'd like something without a CRT!

Cheers,

Dave

We have several Stanford Research SR770s and SR780s.  The latest models have USB adapters and LCD screens. The units are dinosaurs but there are not many units out there that can do DC-100Khz like they do with low noise.  Their support engineers are extremely knowledgeable (some even came from NIST), and best of all they service and support their units. We just had a SR770 calibrated and serviced earlier this year. Reach out to aberger@thinksrs.com.  We looked into the Picoscope 4262 for its 16bit FFT but its slow and clunky compared to the SR units. They just reliably work, even the big semiconductor mfgs swear by them.

[DaJMasta]

Have you considered audio analyzers?  Audio Precision is sort of the industry standard on that side, and something like the APx series probably meets your requirements - though the generator only goes down to 100mHz.  SRS also has newer equipment as an audio analyzer - the SR1, and another search term to consider is Frequency Response Analyzers.  There are also low frequency noise analyzers available (at least as a system) like the E4727B.

I don't know if the software suite with an audio analyzer would suit your measurements, but I think it's likely the hardware involved would be capable, and it's likely a software option or custom software running on the unit's OS would probably do the job.


There's also low frequency network analyzers (AP310, Bode 100, etc., though these are usually PC based) which are generally DC coupled but advertise a higher minimum frequency (10s of mHz) because the sig gen only goes to that for waveforms.  There may be fully integrated models too, but for whatever reason, the ones I've heard of are PC based.

[james_s]

Is there really no direct replacement for a DSA these days? What are they using now for analyzing vibrations in mechanical devices?

[coppercone2]

I think you missed the memo

1) make software that is expensive with tons of options that are expensive, offer basic unappealing garbage for free with the instrument. Fix all the annoying shit in the $10000/year subscription
2) deny other way to use instrument in a productive manner. Tell the customer they need to gather gigantic amounts of data for no reason at all (hey, it makes jobs!), and that the requirement to earthmover data no one wants is not compatible with the traditional instrument that you actually want. Tell them all your customers are using the software and that no one uses instruments anymore (hmm... I wonder how true this is, if you don't take the word of your sales representative)
3) profit
4) ignore complaints call about computer compatibility, software problems, network issues, security issues, etc that storm your telephones
5) keep making stuff obsolete at Iphone pace despite the fact the AFE did not change since 1985 and the underlying data analysis math has been solidly unchanged since the 1950s,
6) forget to mention that you would need a department per instrument to make use of the dragnet data sweep pouring out of the high speed interface
7) dream about how to make the instrument use some kind of hard to acquire expensive secret consumable on a regular basis, your end goal is to turn the device into a printer. MMmmm... some kind of nano material noise lowering liquid that lasts for 15 days a tube.

as a customer, enjoy dealing with IT (and I am scared to see what happens in sensitive installations with this garbage!).

The stand alone DSA solved a problem. Someone saw a way to make people buy a solution no one needs and is not optimized for real world use.

[DaJMasta]

Another potential option if you were willing to go the custom route: an NI PXIe chassis and acquisition cards.  They have some audio/vibration analysis dedicated cards, but some of their DMM modules offer very high sampling rates at top speed (1.8MS/s), and may be good choices for very high resolution waveform capture at those lower frequencies.  You can also have a PC controller integrated into the chassis, making it at least close to a standalone device (still needs monitor and input methods).