Low-frequency spectrum analyzer - time for a new product!

[maxwell3e10]

I recently got another SR780 100 kHz spectrum analyzer - can't believe it is still the best available and currently made product. Stanford Research must be working hard to stuff it with 56 lbs of expensive outdated components (including a CRT) and sells them new for $10k (luckily on ebay it costs 1/10th the price).

All it has is two 16-bit ADCs running at 256 kHz, decent front end amplifiers (5nV/sqrt(Hz) input noise) and FFT analysis with various options.  One can easily do better now with resolution and sampling rate (for example LTC2378-20) on a single PCB.

Maybe the market for low-frequency spectrum analyzers is not so big, but it seems ripe for a new product.

[Daruosha]

For low frequency stuff (up to 100KHz ), I use an external sound card (24 bits / 192KHz) plus some DIY impedance matching/attenuation and the whole system works unbelievably nice and smooth.  The tricky part to take care of the calibration (well if you can even call it calibration), but works great and I'm happy.

I was considering the SR780 earlier, but eventually decided to invest in the sound card and analysis software.

[bitseeker]

Cool. Which audio interface are you using for this, Darousha?

[Daruosha]

I bought Behringer U-PHORIA UMC404. It has 4 input channels (XLR combo inputs) and is fully compatible with ASIO drivers.

[maxwell3e10]

I am thinking more of a professional stand-alone instrument here. There are probably some USB-based scopes that have a decent FFT software, but I haven't looked into those.

Audio interface is a good low-cost option. But they use Sigma-delta instead of SAR ADC. Also many chips have an internal high pass filter at a few Hz, its hard to find this information. If the high-pass filter is external, it can be bypassed.

[exe]

I am thinking more of a professional stand-alone instrument here. There are probably some USB-based scopes that have a decent FFT software, but I haven't looked into those.

Audio interface is a good low-cost option. But they use Sigma-delta instead of SAR ADC. Also many chips have an internal high pass filter at a few Hz, its hard to find this information. If the high-pass filter is external, it can be bypassed.

I'd love to have a stand-alone USB instrument. Preferably in a form of an open-source kit . Although, I have analog discovery 2, so far it suits my needs.

[awallin]

All it has is two 16-bit ADCs running at 256 kHz, decent front end amplifiers (5nV/sqrt(Hz) input noise) and FFT analysis with various options.  One can easily do better now with resolution and sampling rate (for example LTC2378-20) on a single PCB.

Do you know if the commercial instruments use a cross-spectrum technique with 2 ADCs per channel (4 ADCs in total) to suppress ADC-noise?

We have some old HP FFT-analyzers at work and I think they go to 10 MHz (?) - so maybe there's an analog down-converter before ADC in those boxes. Keysight probably doesn't make them anymore?

If there would be a good modular SDR platform that provided the digital parts (fpga, cpu, display etc.) then the slow-speed high-resolution FFT-analyzer-board could be an interesting front-end on such a platform.. There's probably N+1 attempts out there, Zynq might be the fashionable platform atm...

[merlinb]

I am thinking more of a professional stand-alone instrument here.

I've long waited for someone to produce a lost-cost product like this for doing audio specrum analysis. 

[exe]

I am thinking more of a professional stand-alone instrument here.

I've long waited for someone to produce a lost-cost product like this for doing audio specrum analysis. 

Well, depends on what you call low-cost. There are some not too expensive, like this one: https://quantasylum.com/products/qa401-audio-analyzer . Don't know if it's good or not. I've seen a few second-hand for a good price.

[b_force]

For low frequency stuff (up to 100KHz ), I use an external sound card (24 bits / 192KHz) plus some DIY impedance matching/attenuation and the whole system works unbelievably nice and smooth.  The tricky part to take care of the calibration (well if you can even call it calibration), but works great and I'm happy.

I was considering the SR780 earlier, but eventually decided to invest in the sound card and analysis software.

I also use something similar.
By paralleling the inputs you can even bring down the noise a bit.
With 4 channels about 6dB, with a little averaging even a bit more.

SNR of 120dB can easily be done, with averaging even a lot more.

The QA401 is a bargain compared to Audio Precision btw

[maxwell3e10]

I don't think SR780 does anything fancy with x-correlation. Each ADC has about 90dB SNR, which is what you would expect for a good 16-bit DAC. It can calculate cross-correlation between two channels and has many other features, such as swept-sine measurements with a build-in source.

I just looked at QA401, its noise level is about 10 nV/sqrt(Hz) at high frequency, going up to 30 nV/sqrt(Hz) below 100 Hz. Also the bandwidth is limited on high and low end. So it maybe fine for audio work, but for me the spectrum analyzer is useful for general low-frequency low-noise analog electronics development, so it needs to have at least DC-100 kHz (ideally DC-1 MHz) response and as low input 1/f noise as possible.

[bitseeker]

I am thinking more of a professional stand-alone instrument here.

I've long waited for someone to produce a lost-cost product like this for doing audio specrum analysis. 

Well, depends on what you call low-cost. There are some not too expensive, like this one: https://quantasylum.com/products/qa401-audio-analyzer . Don't know if it's good or not. I've seen a few second-hand for a good price.

I've seen the earlier QA400 recommended here and there. However, it's intended for audio, so it may not be the general-purpose interface you're after.

I just looked at QA401, its noise level is about 10 nV/sqrt(Hz) at high frequency, going up to 30 nV/sqrt(Hz) below 100 Hz. Also the bandwidth is limited on high and low end. So it maybe fine for audio work, but for me the spectrum analyzer is useful for general low-frequency low-noise analog electronics development, so it needs to have at least DC-100 kHz (ideally DC-1 MHz) response and as low input 1/f noise as possible.

Ah, yes. I didn't see this prior to my reply. So, you already checked it out.

[b_force]

I am thinking more of a professional stand-alone instrument here.

I've long waited for someone to produce a lost-cost product like this for doing audio specrum analysis. 

Well, depends on what you call low-cost. There are some not too expensive, like this one: https://quantasylum.com/products/qa401-audio-analyzer . Don't know if it's good or not. I've seen a few second-hand for a good price.

I've seen the earlier QA400 recommended here and there. However, it's intended for audio, so it may not be the general-purpose interface you're after.

I just looked at QA401, its noise level is about 10 nV/sqrt(Hz) at high frequency, going up to 30 nV/sqrt(Hz) below 100 Hz. Also the bandwidth is limited on high and low end. So it maybe fine for audio work, but for me the spectrum analyzer is useful for general low-frequency low-noise analog electronics development, so it needs to have at least DC-100 kHz (ideally DC-1 MHz) response and as low input 1/f noise as possible.

Ah, yes. I didn't see this prior to my reply. So, you already checked it out.

There are 768kHz ADC/DACs available nowadays, that should be plenty.

[maxwell3e10]

Maybe there is already an existing high-end audio ADC that would be pretty good. If one could only disable the few-Hz high pass filter, adjust the low pass filter and have high quality mic pre-amps.