RF receiver design help

[kishore2k4]

I am trying to design a direct-down conversion RF receiver to decode 64-QAM 256-OFDM signals. The input frequency range is 600MHz-1000Mhz. I am planning to use the following signal chain

Antenna->LNA(ADL5521)->Quadrature Demodulator(AD8348)->Low Pass Filter (2Mhz)-> 12-bit Dual Channel ADC(>20MSPS)-> FPGA

The LNA has a Noise Figure of 0.8dB (typ) and the I/Q demodulator has a Noise Figure of 11dB (typical).

My question is: Is the above design a good one, as in high receiver sensitivity for 2Mhz channel bandwidths? My previous thread regarding Zero-IF conversion lead to suggesting that I stick with IF-Sampling but size is also an important factor for my design. Will a IF-Sampling (Single or Double Conversion) design yield better results?

I didn't use any differential buffers to drive the ADC because the Quadrature Demodulator(AD8348) has basedband amplifiers built-in that can directly drive an ADC.

Feel free to suggest changes or entirely different designs as I have never designed RF related circuits before. Thanks.

 

[espirit]

I have not read your previous discussions about Zero-IF, but I hope you are aware of the problems, such as DC drift, and phase noise from the down-mixer vco.

The DC problem is avoided by some vendors by using near-zero-IF. I have not studied this, but guess that you would go to an offset of slightly more than your bandwidth/2, so that you band does not include 0Hz.

It will still leave the VCO noise in your band, unless you shift some more...

Do you consider silicon tuners for your design? There are some from NXP and ST.
NXP can be bought, ST is difficult if you need less that 10^6 or so.

 

[MI3talk]

Which book do u suggest for this kind of application?

 

[khouly]

u can use the LO IF or sampling , the zero IF is very had to impleelnt in discerete IC

khouly

 

[espirit]

Are you designing for volume production? That makes a difference in which door (at chip makers) are open to you.
If you design for a lower volume, you may try to get hold of a can-tuner. ALPS and LG makes them, but they are also diffucult to get, and it is even more difficult to get documentation.
At least it would save you the trouble of designing the RF part. Most of them come out with an IF of 36MHz, for which ceramic resonator band filters are available.
Then you can (under-)sample that IF and continue in the digital domain.

I'm sorry, I don't know or have any books about this.
We designed a series of modulators (QAM and COFDM) with upconvertors for 50-1000MHz and 1000-2000MHz. It took us a while to get phase noise, amplifier linearity and stability at the right level.

 

[kishore2k4]

Thanks for all the helpful replies. I have some more questions:

I couldn't find any single-chip direct conversoin tuner for my frequency of interest. From your suggestions I decided to use Low-IF(Quadrature Demodulation) with the same setup as mentioned in the first post. I can use an IF of 6MHz for 2Mhz channel bandwidths. I don't want to go over 10MHz as that will leave me with little processing gain in the ADC. I am hoping the image rejection problem will be taken care of by using quadrature downconversoin?!?

Are there any books/tutorials that deal with practical aspects of RF receiver design?
espirit: Thanks for your suggestion but I don't want to use tuner modules as they tend to be very expensive at low-medium quantity requirements like mine.

Also can someone suggest a receiver chain that has low noise figure and high linearity, ofcourse with decent trade-off between them. Thank You.

 

[khouly]

as low IF receiver , there willbe also and image problem , so u need some kind of filtering "image rejection filter , or use one of the image rejection architectures

khouly

 

[kooller]

i think for the Low IF structure, you'd better use an IF half of the bandwith, so that the image signal is in the adjacent channel, where the interference is relative small.

 

[mallelauma]

this may be help you to design lna

 

[springf2000]

THE s11 of the adl5521 is poor and it shall degrade your sensitivity, i suggest you change your LNA

 

[rosaeidi]

hi

see "RF Microelectronics" book, by Behzad Razavi.
h**p://search.barnesandnoble.com/RF-Microelectronics/Behzad-Razavi/e/9780138875718