Why do most spectrum analyzerrs and RF signal generators start at 9kHz?

[ivan747]

Why 9kHz specifically? Why not 10?

I know they have to start at some point but this seems like an industry convention. Where does it come from?

[ataradov]

Why 9kHz specifically? Why not 10?

Probably for the same reason why prices are $9.99 and not $10 Just seems less and in practice does not matter.

[KE5FX]

Why 9kHz specifically? Why not 10?

I know they have to start at some point but this seems like an industry convention. Where does it come from?

9 kHz is a common specification limit for attenuator chips and the like. 

It's a lot more expensive to buy front-end chips that work well all the way down to DC.  (Lots of data sheets rate their components at "DC-to-XX GHz" but the reality is very different.) 

[Shock]

Local Oscillator feedthrough, pages 19 to 23.

http://www.keysight.com/upload/cmc_upload/All/02-18-03-B2B-RF-SpectrumAnalysis-Thomas-Holmes-Hightower-839.pdf

[T3sl4co1l]

Well, LO feedthrough is just however much RBW you have.  You necessarily* get a spike at 0Hz offset.

*Actually, I suppose one could make a spec with a very carefully balanced mixer, that rejects the LO signal, and perhaps also which detects signals orthogonal to it.  Then 0Hz would be very attenuated or entirely invisible, and signals down to 0Hz (within uncertainty of the RBW) would be resolvable.

As for standards, I believe some CISPR and MIL standards start at 9kHz, or something similar to it, thus demanding instruments start there.  (Most domestic standards, i.e. FCC Part 15 and CISPR 22, cut off at 150kHz.)

I've seen at least one R&S analyzer that claimed something like 25Hz to 10s of GHz.  It was safely mounted in a rack, and I'm pretty sure that, if I had merely touched it, that fingerprint alone would've depreciated it by at least my wages that day.  (Didn't see it in operation though.)

I believe, some of the more stringent MIL standards (and probably going into SIGINT territory as well) place limits on conducted and radiated emissions over an extremely broad range, i.e., anything from power supply ripple in the 20Hz range, up to radiated noise in the GHz.  The LF limits might be quite modest for most equipment (easily met without special precautions), but I don't really know how tight the limits go (especially for security roles).

Tim

[hendorog]

Times are changing.

https://signalhound.com/products/usb-sa44b/

1Hz -> 4.4 GHz

[hendorog]

I have the SA124 - and I don't have OCD so maybe I am a poor judge of an SA.

[HighVoltage]

I have been wondering about this myself.

And my guess is, that they wanted to cover 10 KHz and since you can not start at 10 KHz on a SA to cover it, they just went to a start of 9 KHz.

But I am really interested, if there is a real science answer to this.
 

[hendorog]

I have the SA124 - and I don't have OCD so maybe I am a poor judge of an SA.

Not having OCD is a gift.

The best way to drive an OCD crazy is not to finish a sentence, such

Sorry for that, and happy to help albeit in a small way.

[PE1RKI]

makes me wonder why my 26.5ghz SA goes all the way down to 1 KHz
i found another possible answer on reddit:
https://www.reddit.com/r/AskElectronics/comments/1sb6w3/why_do_most_spectrum_analyzers_start_at_9khz/

[tggzzz]

I have the SA124 - and I don't have OCD so maybe I am a poor judge of an SA

Not having OCD is a gift.

The best way to drive an OCD crazy is not to finish a sentence, such

Or worse: https://xkcd.com/859/

[tggzzz]

Why 9kHz specifically? Why not 10?

I know they have to start at some point but this seems like an industry convention. Where does it come from?

ISTR that EMI does not occur below 9kHz - by definition of course. I do not know which is the chicken and which is the egg.

[Kjelt]

makes it unusable for audio applications which su*ks

[ivan747]

makes it unusable for audio applications which su*ks

Yup. Although for audio you'd be better off with a very good ADC.

[joeqsmith]

Re: Why do most spectrum analyzerrs and RF signal generators start at 9kHz?

Why 9kHz specifically? Why not 10?

I know they have to start at some point but this seems like an industry convention. Where does it come from?

Most of my RF equipment is pretty old.   

HP8640B RF generator 447KHz
HP8660D/86603A  RF generator maybe 100KHz before the distortion it real bad
HP8754A  VNA maybe 4MHz on the low side without using an external RF generator
HP3589A spectrum/network analyzer 10Hz
HP8569A spectrum analyzer 10MHz
Tektronix 496P spectrum analyzer 1KHz (30Hz rbw, can see 100Hz, drifts, slow....)

My newer ARBs are all good into the 1MHz and will run way below 1Hz.

[MadTux]

All depends on RBW (to filter LO spike out)and whether it is AC or DC coupled.
HP 3585 is DC coupled, min RBW of 3Hz and goes down to 20Hz
HP 8566/8568 both have 10Hz RBW and go down to 100Hz on DC coupled inputs, 8568 goes down to 100kHz on AC coupled main input.

[cyr]

9kHz seems to be where the regulated radio spectrum starts, might have something to do with it (or indirectly, because of RFI standards?)

https://en.wikipedia.org/wiki/Very_low_frequency

[Bud]

This was answered on Agilent network analyzer forum by their engineers. Sadly i can't remember what it was but believe that had to do with the value of DC blocking capacitor on VNA input.

[ivan747]

I found somehting very interesting. Our national frequency allocation plan officially starts at 9kHz, stating that at 9kHz and below there are no attributed frequencies... Coincidence? Can anybody else compare this to their own?

[hendorog]

I found somehting very interesting. Our national frequency allocation plan officially starts at 9kHz, stating that at 9kHz and below there are no attributed frequencies... Coincidence? Can anybody else compare this to their own?

Snap, its the same here.

http://www.rsm.govt.nz/online-services-resources/pdf-and-documents-library/publications-and-guides/radio-spectrum-allocations-in-new-zealand-chart

[6581]

It's over 9000!


Could not resist. Sorry. Seemed relevant.

[G0HZU]

I always assumed it was to do with common frequency regs. However, all of my regular spectrum analysers here go right down to a few Hertz as they either have no DC blocking cap or the cap can be bypassed with an internal RF switch if DC blocking is not needed.

[T3sl4co1l]

makes it unusable for audio applications which su*ks

(You mean "sucks"?)

You wouldn't want a drifty, phase noisy, 1GHz instrument with excessive bandwidth to read such a signal anyway.  That's what a DSA is for.

Tim

[G0HZU]

makes it unusable for audio applications which su*ks

(You mean "sucks"?)

You wouldn't want a drifty, phase noisy, 1GHz instrument with excessive bandwidth to read such a signal anyway.  That's what a DSA is for.

Tim

For general purpose audio stuff (not serious HIFi), some xGHz analysers do OK down at audio in terms of stability and phase noise. The major issue is the slow sweep time for narrow RBW settings although newer instruments use DSP here and these are a fair bit faster.

But a fully synthesised >1GHz spectrum analyser from the late 1970s can achieve very good stability and the phase noise can be as good as -112dBc/Hz at 300Hz offset on a top model like the old HP8568. These analysers also preserve a decent noise floor right down to a few hundred Hz. So they can be useful way down at AF frequencies if you can live with the very slow sweep speed.

But I'd usually just use a laptop and PC soundcard for stuff like this.