Adjustable Resolution Bandwidth Filter (RBW) of a Spectrum Analyzer How Its Done

[Mechatrommer]

as the topic stated, i've been searching in the net how the adjustable RBW filter is implemented in circuit to no success... if we look at the W2AEW's video on this...
youtube=Ffhs9Ny03lM

RBW is adjustable in FW so i guess there must be switches on the pcb to change RBW. but when looking at Dave's video, there is no such thing...
youtube.com=-8fr_otW0q4

can anyone give some light?

[capt bullshot]

Many way to implement this:

Older stuff uses a bunch of filters, each one made for the specific RBW, and switching them into the signal path. Especially the small RBW (e.g. 30Hz) filters are difficult, they operate at the third IF (usually in the 10 MHz range, usually you'd require a crystal filter made from a bunch of selected and thoroughly tuned crystals here)
In one particular case I've seen a cascaded crystal filter, where the Q of each stage could be varied by some electrical means (don't remember the details atm), resulting in an adjustable RBW
Less modern devices use a mixture of analog and digital filtering, changing the BW of a digital filter is rather simple by exchanging the set set of coefficients.
Modern devices most probably use some kind of digital processing for the RBW filter, and everything is hidden inside a DSP or FPGA chip.

[BrianHG]

Dave even says explicitly and shows a block diagram at the end that the fine filtering at the IF stage is being done all digitally.
Seek to time 42:20, watch and listen...

[Mechatrommer]

Dave even says explicitly and shows a block diagram at the end that the fine filtering at the IF stage is being done all digitally.
Seek to time 42:20, watch and listen...

what i understand is he's talking about envelop detector thats missing that maybe went into all digital. the rbw filter should be before that. referring to diagram dave made below, i dont see any selectable filter in 1st and 2nd stage mixer, there's saw filter in 2nd stage, but not adjustable and not sure what they do there. in 3rd stage mixer (1GHz), there are selectable networks with many unpopulated components, it could be it but it seems too simple. GOHZU described it as another band pass filter, isnt rbw is just few selectable band pass filters?. need more clarification from more knowledgeable people. after the IF OUTPUT, the signal will go direct to 40MSps sampler on the next board (where the ref clk comes from), but not sure how 40MSps ADC is any good sampling 190MHz IF sine (1GHz - 810MHz) without an envelop detector/rectifier.

[capt bullshot]

what i understand is he's talking about envelop detector thats missing that maybe went into all digital. the rbw filter should be before that. referring to diagram dave made below, i dont see any selectable filter in 1st and 2nd stage mixer, there's saw filter in 2nd stage, but not adjustable and not sure what they do there. in 3rd stage mixer (1GHz), there are selectable networks with many unpopulated components, it could be it but it seems too simple. GOHZU described it as another band pass filter, isnt rbw is just few selectable band pass filters?. need more clarification from more knowledgeable people. after the IF OUTPUT, the signal will go direct to 40MSps sampler on the next board (where the ref clk comes from), but not sure how 40MSps ADC is any good sampling 190MHz IF sine (1GHz - 810MHz) without an envelop detector/rectifier.

Can't tell anything about the SP4T labelled components, but where's the problem with 40MSps sampling a 190MHz signal? That just works fine and is called "undersampling" - the signal is
folded into the 0..20MHz output band of the ADC - or "mixed down to 10MHz" (that would be the equivalent operation). So you need a BPF to ensure there are no signal components outside the desired maximum IF BW (e.g. 185MHz to 195MHz - that just a wild guess, I don't know anything about this particular SA). Then sample the 190MHz centered IF with 40MSps, folding the signal down to 10MHz in the digital domain, and do the RBW filtering digitally.