50MHz to 1GHz, freq step 1Hz, RF generator-design discussion

[rfmw]

dds rf generator

Dear friends,

suppose we have to design a RF signal generator with IQ modulation possibility, how would you start, which components would you choose.

The requirements are as follows:

1. Output freq range: 50MHz to 1GHz
2. Frequency step: 1Hz (this is not a mistake!)
3. IQ modulation support is a must
4. All spurious at DC-1GHz band must be below -50dBc, and -40dBc above 1GHz
(5. Application is digital TV with multi transmitters at the same frequency)

I'm thinking of IQ-DDS generator from Analog devices (AD9957), but it's only capable of 400MHz output frequency. There are 2.3GHz and 4GHz clock frequency DDS chips from Maxim, but they won't sell us small quantities to try our design...

Ideas, suggestions, schemes, diagrams, etc are very welcome...


Cheers,
rfmw

 

[biff44]

100mhz rf generator

IF you are happy with the DDS I/Q control available in the Analog Devices chip, then all you have to do is frequency convert the output.

I would use the DDS from 50 to 350 MHz straight, with a steep LPF at 400 MHz. I would downconvert a high side LO to achieve 350 to 650 MHz. I would downconvert another highside LO to achieve 650 to 950 MHz, and again to do 950 to 1000 MHz.

Depending on your spur requirements, you might be able to do it with only 2 LOs.

When you are high side downconverting, you will invert the modulation constellation, of course.

Added after 2 minutes:

If it is only phase or frequency modulation, and you do not care about spur levels, you can simply frequency triple the DDS output.

 

[rfmw]

ad9911 design

Thanks Biff44,

would you mind describing "I would downconvert a high side LO to achieve 350 to 650 MHz" in another words, cos I have some doubts if I got you correctly? So you're suggesting to mix my DC-350MHz with LO, about 700MHz, and filter out the lower band?

The modulation would be QAM, so amplitude matters.

Cheers!

 

[Element_115]

dds-generator

Hi RFMW

I might suggest an FPGA for the QAM Generation, and these parts for
Amps.
http://www.rfhic.com/data/pdf/1-MMIC/29-AP211.pdf
http://www.chipdocs.com/datasheets/datasheet-pdf/WatkinsJohnson-WJ-Company/AH22.html

But that sounds like a tough design!
Good Luck

 

[biff44]

dds generator 1ghz

Do you know what level of spurious sideband spurs you can stand?

 

[biff44]

iq-dds

Also, what RF bandwidth, QAM type (16, 64...)

 

[yendori]

rf generator 1 ghz

I used a AD9911 with a VCO as ref clk to achieve the similar output you need.

Make sure you do a good spur analysis before spending to much time on your design. Do forget the truncation spurs!

Rod

 

[rfmw]

50mhz rf generator

The spurious levels are mentioned above, that includes also the sideband spurs. The modulation BW is approx 10MHz. Modulation type is actually OFDM.

Well, the -50dBc spur content in the whole band is the most troublesome. Indeed the design is not simple.

Thank you all for your suggestions.

 

[biff44]

Re: 50MHz to 1GHz, freq step 1Hz, RF generator-design discus

Well, this is the sort of thing that I do routinely as a consultant. You need to do a subsystem design first. Tradeoff some block diagrams, develop a frequency plan, perform a spurious signal analysis, revise the frequency plan, select components for leakage and especially envelope peaking performance, design some tight filters, breadboard it, make some final tweaks, and away you go. Sounds like it might be a commercial application, so cost is an issue too.

An initial approach might be to have the DDS operate in a sweet spot, maybe 250-350 MHz, upconvert it to a much higher frequency (maybe ~2 GHz), bandpass filter, then downconvert with an LO with 100 MHz step size, filter it in a couple of bandpasses, and use a switch to select the desired output. There are other ways that you would have to study and trade off. There might be other key specs to consider, like phase noise and group delay shape.