ADF4351 Sinewave?

[SaabFAN]

I'm currently gathering the parts together to make myself a RF Signal-Gen after I decided to spend my money on a Spectrum Analyzer instead of a Signal Gen.

I haven't powered up the ADF4351 Module I bought for a Spectrum Analyzer-Project yet, but from what I read so far, it seems to put out a more or less Square-shaped wave.

It is my idea to use several, maybe even tuneable, lowpass-filters to hammer the square into the shape of a sinusoid. The frequency Span I want to cover is about 40 to 2500MHz (lower frequencies will be generated by a DDS-Module based on the AD9850, of which I have 2 left from a batch of 5 that I ordered some time ago.
I also thought about using the LPFs down to their -30dB-Point and use a +30dB Amp and a PIN-Attenuator between the ADF4351-Module and the Filters to always have the sine-wave at a specified level.

Would you say that this is a feasible idea, or are there better methods to do this?

[RadioNerd]

I would say that's perfectly feasible. However, there are some pints to consider:

- Lowpass filtering a square wave into a sine should work perfectly fine, but I would recommend using a switched filter bank instead of tuned filters as it will give you more stable/predictable and lower attenuation. Minicircuits LTCC filters are predestined for this kind of application.
- I would suggest following component order: ADF4351 -> PIN attenuator-> Amplifier-> LPFs. This minimizes saturation and linearity issues, comes however at the cost of bad output matching at high frequencies. Putting a fixed attenuator at the output could improve that if the absolute output power level is not critical.
- Many PIN attenuators have a rather high lower frequency limit.
- Achieving accurate power leveling over temperature and frequency is not trivial. You could try to implement an ALC loop using a directional coupler and an RF detector. However, you will need either a voltage variable attenuator or a high resolution step attenuator controlled by an MCU to achieve that. 

[Andrey_irk]

I also thought about using the LPFs down to their -30dB-Point

I don't get what you mean. Can you please clarify it?

[SaabFAN]

I also thought about using the LPFs down to their -30dB-Point

I don't get what you mean. Can you please clarify it?

With that I meant that I want to use the LPFs to a frequency at which they exhibit 30dB of Attenuation of the wanted signal. That's why I added a PIN-Attenuator, controlled by an Automatic Level-Control circuit.

The Signal-Path would be like this:

ADF4351 --> PIN-Attenutator --> Bank of LowPassFilters --> +30dB Amplifier --> Power-Detector --> Output at known Powerlevel

Control-Path would then be this:

Power-Detector --> OpAmp trying to equalize the Voltage from the Power-Detector and a digitally adjustable Reference-Voltage --> PIN-Attenuator

I got this arrangement from the Service-Manual of an old Philips RF Signal Generator.


About the problems of using PIN-Attenuators at low frequencies: Frequencies up to 65MHz will be generated by a DDS, which will get its own signal-path to the output-Amp.
I might as well also use a frequency-mixer to create lower frequencies (I bought a few^^) - More about that in the upcoming thread in the Open Source Hardware-Subforum

[nctnico]

I'm currently gathering the parts together to make myself a RF Signal-Gen after I decided to spend my money on a Spectrum Analyzer instead of a Signal Gen.

I haven't powered up the ADF4351 Module I bought for a Spectrum Analyzer-Project yet, but from what I read so far, it seems to put out a more or less Square-shaped wave.

It is my idea to use several, maybe even tuneable, lowpass-filters to hammer the square into the shape of a sinusoid.

Would you say that this is a feasible idea, or are there better methods to do this?

I had the same idea many years ago. You'll need to switch both the input and outputs of the filters otherwise the filters will interact with eachother. It will be hard to push the 2nd order harmonics below -40dB though because at that point the amplifiers will add a lot to the distortion as well.

[SaabFAN]

For the switching I'm planning to use Single-Pole Double Throw Microwave-Switches by M/A-COM (Component: MASWSS0115TR-3000).

Or is there a fundamentally better way to do this? Frequency Mixing for example? I have two MiniCircuits VCOs here - One works from 770MHz up to 1.7GHz.
The best way would be, of course, a dedicated VCO, but I tried building a high frequency VCO and couldn't make it work reliably, so I would like to avoid this route that would just cause massive delays in trying to get the VCOs working

[nctnico]

You'll need more filter banks. I was planning on (IIRC) 7 filters to cover tens of MHz to 4.4GHz and I did find some 4:1 RF switch chips. You'll need switches as both ends of the filters to keep some impedance matching.

[RadioNerd]

For the filterbank switches up to 3 GHz I would suggest something like the SKY13384-350LF part from Skyworks. It is an absorptive switch, i.e. the inactive ports are internally terminated with 50 ohms and it has good isolation between the ports. Using a terminated switch avoids annoying parasitic resonance effects in the unterminated lines (which can very easily occur at these frequencies).

[SaabFAN]

I have posted the block-diagram in the thread in the Open Source-Hardware Sub-Forum:
https://www.eevblog.com/forum/oshw/open-source-hw-rf-signal-generator/