Downconversion for spectrum analyzers

[spasmodic1]

Hi,

i have a rigol DSA815, im no million air and dont want or really need to spend that money on a scope to cover up to the 5.8ghz region.

What would be the best way to frequency divide a signal while maintaining a somewhat close to original amplitude so one could display the signal on a lower bandwidth SA?

[DrLuke]

Well, one way is to mix it with a similar frequency, resulting in one very high frequency signal, and one low frequency signal. That's how software defined radios work. For more, see this: http://en.wikipedia.org/wiki/Heterodyning

[Tac Eht Xilef]

(Disclaimer: by no means a microwave expert, though I've dabbled...)

Yeah, a classic IF mixer/downconverter would be a typical choice (e.g. the venerable MC13154, although it's only good to 1.8GHz). At those sorts of frequencies, something like a RFFC5072 (just found from a quick search) would get you in the ballpark of your DSA815's 1.5GHz, or you could do a double conversion.

Would it be good enough for dicking around and looking at stuff? Sure.
Would it be good enough for side-by-side relative comparisons/measurements? Probably.
Would it be good enough for absolute or long-term measurements? Hmmm, not without some serious thinking applied to the design.

[spasmodic1]

Thanks for the replys,

i have a double balanced mixer, but it just a matter of having a vco or tunable oscillator for the LO that would cover a decent range of frequencies.
Although that rffc5072 did look very useful hard to obtain in small quantities though.
But i thought there might be something out there that could give you a clean signal without all the harmonics of a mixer. But i will give the RFFC5072 a go if i can get one.

cheers.

[mhzghz]

Hi,

i have a rigol DSA815, im no million air and dont want or really need to spend that money on a scope to cover up to the 5.8ghz region.

What would be the best way to frequency divide a signal while maintaining a somewhat close to original amplitude so one could display the signal on a lower bandwidth SA?

Make a downconverter with a quality Watkins-Johnson or M/A-Com M1H double-balanced mixer and a 5 GHz local oscillator signal at +7 dBm.

You'll need to attenuate your 5.8 GHz signal down to -10 dBm max.  Apply it to the mixer's RF input.

Apply a +7 dBm 5 GHz signal to the LO port. 

The DC-800 MHz IF output will correspond to 5.0 - 5.8 GHz. You should add a 3 dB attenuator on the IF output.

You could probably use a 2.5 GHz LO and the second harmonic in a pinch.

Losses will be the mixer loss + IF attenuator loss, which will be around 10 dB total.

[spasmodic1]

Thanks quys,

I will have to work on an oscillator suitable for those frequencies.

[notsob]

minicircuits may be worth a visit to see what they have available

[KJDS]

If you're going to want a useful measurement out of the spec ana, then the oscillator should be phases locked back to the spec ana reference. I'd also want a level 13 mixer as a minimum, do some research of mixer third order products and work out what dynamic range you'd be happy with.  To get the best dynamic range from the mixer, drive it harder than nominal, so if it is spec'd for 13dBm +/-2dB, aim for 14 to 15 dBm.

The output of the mixer needs to be terminated in 50 ohms at both IF frequencies. If you have a 5GHz LO and a 5.8GHz RF signal then the IF needs to see a good 50 ohm at the wanted 0.8GHz IF and also at the 10.8GHz IF that you don't need.

[spasmodic1]

If you're going to want a useful measurement out of the spec ana, then the oscillator should be phases locked back to the spec ana reference. I'd also want a level 13 mixer as a minimum, do some research of mixer third order products and work out what dynamic range you'd be happy with.  To get the best dynamic range from the mixer, drive it harder than nominal, so if it is spec'd for 13dBm +/-2dB, aim for 14 to 15 dBm.

The output of the mixer needs to be terminated in 50 ohms at both IF frequencies. If you have a 5GHz LO and a 5.8GHz RF signal then the IF needs to see a good 50 ohm at the wanted 0.8GHz IF and also at the 10.8GHz IF that you don't need.

thanks KJDS
The IF is going to the SA input which has a 50 ohm input so termination isn't needed right?. Or are you saying at .8 ghz the IF will need to be impedance matched to 50 ohms?

[KJDS]

If you're going to want a useful measurement out of the spec ana, then the oscillator should be phases locked back to the spec ana reference. I'd also want a level 13 mixer as a minimum, do some research of mixer third order products and work out what dynamic range you'd be happy with.  To get the best dynamic range from the mixer, drive it harder than nominal, so if it is spec'd for 13dBm +/-2dB, aim for 14 to 15 dBm.

The output of the mixer needs to be terminated in 50 ohms at both IF frequencies. If you have a 5GHz LO and a 5.8GHz RF signal then the IF needs to see a good 50 ohm at the wanted 0.8GHz IF and also at the 10.8GHz IF that you don't need.

thanks KJDS
The IF is going to the SA input which has a 50 ohm input so termination isn't needed right?. Or are you saying at .8 ghz the IF will need to be impedance matched to 50 ohms?

The spectrum analyzer will provide a good match at 0.8GHz. It is very likely that it will provide a bad match at 10.8GHz which is where the other mixing product will be made. It's a non-trivial area of design. If you're not interested in accuracy and just want an idea that the modulation is ok then you can ignore it, but it will mess up conversion loss accuracy and do funny things with the higher order mixing products which may end up appearing when they otherwise wouldn't. The easiest way round it would be to put a 3dB SMA attenuator on the mixer output. That will provide enough loss at 10.8GHz to prevent any nasties occurring.

[Mr Simpleton]

You will have to add filters! If you do look at the mixing products and what frequencies your oscillator will be at, you may find a suitable bandpass filter to help.

Old analyzers had a nifty feature that you could see how the spurs moved, and thus decide that is was a spurioius not something to be bothered about... still it is a pain 

So get a good bandpass filter at the input of the mixer, and keep your 815 at highest possible frequency to move the mirror signals away. And do not forget to add some gain to recover from mixer and filter losses... Looks quite like a major project