Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Design RF transceiver for OFDM/Wimax (using DRP?)

Status
Not open for further replies.

fatboy9345

Newbie level 5
Joined
Feb 17, 2006
Messages
8
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,369
Dear all,

I'm trying to design RF transciever for OFDM (Wimax, in particular) system. Is there any good reference I can study?

Also, the idea of "Digital RF Processing (DRP)" has been proposed recently (see e.g. https://focus.ti.com/pdfs/wtbu/ti_drpwhitepaper.pdf). Do you have any suggestion if I would like to design the OFDM/Wimax transceiver using the DRP?

Many thanks!
 

What are the RF specifications of the transciever ?
How do you design the specs? I need your reply,3X
 

It is not easy:

NF<7.0dB Everything from connector
SYN-EVM < -37dB
TX-EVM PA <-34dB

Frequency stability after RX/TX switch <50e-9!!!

Not an easy task for the Student-CMOS-Abidi-Razavi section.
 

Thank lynnheflex for the reply and thank rfsystem for the RF specifications. My area of expertise is actually in baseband design of OFDM systems, not RF design. I would very appreciate if you could suggest any good reference to help me understand the RF design for OFDM/Wimax.

With digital RF processing (DRP), the conventional VCO and PA driver circuits are replaced by digitally control oscillator and digitally controlled power amplifier circuits, respectively. The resulting digital-to-frequency conversion and digital-to-RF-amplitude conversion transfer functions could be made vary linear and of high dynamic range. Such DRP seems to be a viable technique for the design of Wimax transceiver. Since my knowledge in RF design is limited, I would highly appreciate any advice on the RF design for OFDM/Wimax with the DRP.

Many thanks
 

I think it is very dangerous to substitute analog processing by digital processing because of limited knowledge in the analog domain.

Engineering is to work out a solution with limited skills, minimum effort and maximum result in all domains. If some of the knowledge domain is unsufficient covered you will be overrun by teams having a broader skill landscape.

You should have very good reasons to process something in a digital domain which before was done in the analog domain. Most proposals to shift implementation techniques come from the domain expert. He or she knows what are the exact trades.
 

Dear rfsystem,

I may have an opportunity to work with the experts in RF design to develop OFDM/Wimax transceiver. The system-on-chip architecture design requires the knowledge of both digital and analog domains. It would be very helpful if I have some fundamental idea on the RF design for OFDM/Wimax (and possibly the use of DRP in the design). Would you please kindly suggest me some good reference? Thank you very much.
 

If I understand you right there is only a digital controlled oscillator which drive a PA with digital amplitude control. Every other thing is digital! That is the analog/digital interface for TX. How about the RX? Channel filtering in the digital domain? Or direct RF sampling?

Sorry there is not a single reference. Most of the techniques above are already used to check the performance. Only a mixed analog/digital control of LO is used in production now. All other techniques show performance/power or area issues. You can browse through many IEEE Solid-State magazines, ISSCC or RFIC conference papers. Some of the component results are shown there

Take for example ADC. At the ISSCC the best nyquist ADCs need 300-800fJ (femtoJoule) for each decision step. So a 10bit/20MS ADC for WiMAX need

(300e-15...800e-15)*2^10*20e6=6.1...16.4mW

for each converter.
If the channel filtering is done in the digital domain the sampling rate increase by the number of channels and the resolution increase by the adjacent channel rejection. Take the number from above with 16 channels and 12bit instead

(300e-15...800e-15)*2^12*16*20e6=393.2...1048.6mW

not counting the power of the digital filtering. That is the result of about the 50 best people from academic and industry presented at ISSCC session12.

Otherwise FFT is a good choice because it reduce the computation effort by 0.5*log2(n) over linear equalizer. And FFT is done best in digital. So engineering should choose the best technique. The user only decide about power, area and engineering cost. They are not interessted in digital or analog.

Marketing gives the stamp!

Is digital sexy and analog stupid?
 

WiMax is new enough that you won't find much published in the area of the RF system design. I recently picked up a copy of "RF System Design of Transceivers for Wireless Communications" by Qizheng Gu, and that seems like a decent book for RF systems design, although it is focused on CDMA/GSM/WCDMA.

For Wimax, start trawling through the IEEE publication's database for WiMax papers.

TI's DRP is a "Polar-Lite" approach, which has limitations and problems. I'd suggest that unless you know that trade off's between an I/Q transmitter and a Polar transmitter, you focus on a I/Q Transmitter for implementation. Much more knowledge about how to make an I/Q transmitter work.

Dave
 

is there anybody know the specification of the WiMax's power amplifier such as linearity, P1dB, EVM, and so on ?
 

RFDave said:
TI's DRP is a "Polar-Lite" approach, which has limitations and problems. I'd suggest that unless you know that trade off's between an I/Q transmitter and a Polar transmitter, you focus on a I/Q Transmitter for implementation. Much more knowledge about how to make an I/Q transmitter work

Then please advise me some good reefereces to make an 80.11a I/Q transmitter work. Thanks in advance!
 

Take a look through the IEEE Journal of Solid State Circuits for the past 5 years, and you will see enough papers to make your eye's glaze over. I'd focus on papers from Broadcom, they have published a lot.


Dave
 

fatboy9345 said:
Dear all,

I'm trying to design RF transciever for OFDM (Wimax, in particular) system. Is there any good reference I can study?

Also, the idea of "Digital RF Processing (DRP)" has been proposed recently (see e.g. h**p://focus.ti.com/pdfs/wtbu/ti_drpwhitepaper.pdf). Do you have any suggestion if I would like to design the OFDM/Wimax transceiver using the DRP?

Many thanks!

That article is nice but I'm asking you: where did you'll find NOW a DRP for WIMAX, either for BS or MS nomatter if is designed for unlicesed or licensed band?
The answer: nowhere, you'll have to design it's structure using the fastes DSP on the market, maybe with some help of a FPGA, a baseband and an 0IF transciever (ok maybe a standard 2IF transciever with SAW filters).
Could you do it in other way ? I'm curios about your answer.

Added after 1 minutes:

jcpu said:
RFDave said:
TI's DRP is a "Polar-Lite" approach, which has limitations and problems. I'd suggest that unless you know that trade off's between an I/Q transmitter and a Polar transmitter, you focus on a I/Q Transmitter for implementation. Much more knowledge about how to make an I/Q transmitter work

Then please advise me some good reefereces to make an 80.11a I/Q transmitter work. Thanks in advance!

Why don't you start reading the datasheets of 802.11 transcievers from Maxim's site?
 

sorry sir, went through broadcom and maxim,
still need reference for I/Q calibration details.
 

Usually IQ calibration is not required on Maxim's transcievers. There is a datasheet about this, ask for it at Maxim.
Uncalibrated IQ will affect the EVM. Hoewever the Maxim's 802.11 transcievers have quite good EVM.
 

what you sat if we have low yield lot with IQ imbalance trouble.
 

melc said:
Usually IQ calibration is not required on Maxim's transcievers. There is a datasheet about this, ask for it at Maxim.
Uncalibrated IQ will affect the EVM. Hoewever the Maxim's 802.11 transcievers have quite good EVM.

Maxim utilized SiGe process so that they don't need calibration?
 

For a direct upconversion transceiver IQ imbalance contribute to TX EVM. That is what I know. Depending on TX mismatches either low yield or a calibration is needed. But why SiGe does not need calibration?
 

Calibration could well be handled internally to the part, with no exterior visibility. Also, just because it's not mentioned in the web page doesn't mean that it's not used. On the other hand, you may be able to get the performance of a SiGe part good enough that you don't need calibration.

Dave
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top