direct digital synthesizers 6 ghz

Hello,

i need to get between various RF channels a very good resolution and precition in phase shifting between them. The RF range i'm interested in is broadband, at least from 2 to 6GHz. I've been told that the best solution to get very high resolution is using DDS. Is this right?

Can anyone indicate me any document or article where I can read some basics about this?

Thanks so much

Not really. A DDS synthesizer might be a component in the synthesizer system that you need, but such a system will have many more parts than just a DDS. The key component will be some bank of VCO oscillators that span the 2 to 6 GHz range, and the DDS can be used to slightly change the frequency of those oscillators through mixing.

Yes you are right.
I'll need a mixer. But how will I get an specified phase at the output?
I'm trying to figure out a topology or scheme of components that allow me to get 8 RF channels simulating a received signal by an 8 antenna array.

So, in an ideal way, the simplest thing one can think of, is a sweeper, a 1 to 8 divider, and for each channel, an atennuator and a switched line phase shifter. That's all it talkes. BUT. Here is the tricky part. I want the RF output to be able to go from 2 to 6 at least (to the maximum possible). and I need very high resolution and precition in phase shifting. I can't do this with phase shifter, although I have to think about it and look the state of the art, but it seems that DDS will have to play a key role in the system i describe you.

What do you guys think?

Thanks,

(i love this forum, share knowledge!)

You need 8 phase coherent signals. If you had 8 different DDS synthesizers, with programmable phase, you would have to be very careful indeed to insure that they were phase coherent. Most cases they would not be. You could watch the phase of each chanel rotating around at maybe a 10 hz rate or so! This is because a DDS generates its small frequency steps by applying a stepped ramp of phase. Having the DDS's be coherent would require some way of reseting all 8 of them at one point in time, and possibly avoiding a whole bunch of operating frequencies.

If I had to do it, I might make one central frequency source, an 8 way splitter, and 8 vector phase modulators, and 8 ferrite isolators. I would measure the 8 outputs on a network analyzer to calibrate the phases at each frequency, remember those calibration settings, and then proceed on with the simulatred antenna test. At each frequency you would program each vector modulator for the desired phase shift PLUS the calibration error correction term.

If this is a commercial job, I do do that sort of design work as a consultant. Be glad to design one for you.

Added after 3 minutes:

Here is a chip that almost does that bandwidth:
http://www.hittite.com/products/view...view/HMC697LP4

Hello

don't the DDS's remain coherent thanks to the crystal oscillator common to all of them?

I think the problems would arise when trying to take the system into ranges of frequency were the DDS don't work, we would have to use up-conversion and that would complicate our goals, which is:

- having 8 RF signals with relative phase shifts between them of 1o resolution at least. The idea is equivalent to a phase array, where we want a 1o phase resolution.

Thanks guys

Actually when you drive more than one DDS with the same referance, they all be coherent and if they drive some kind of PLL, they will be phase lock also.
I suppose main problem about using DDS is the frequency limit as you said. You should not exceed 350-360 MHz output with DDS.
Rather than mixers etc. I strongly recommend that you try to have DDS output as referance signal for commonly used PLL that drives an VCO of your band.