Zero-IF receiver DC offset

In Zero-IF receivers after mixing the LO with the incoming signal the carrier would appear as a DC offset. If the LOs are slightly offset, due to crystal tolerances, there appears to be a large signal near zero, at a couple of kHz. Is this signal just filtered out with a high-pass filter? Is this phenomena still considered to be a DC offset, even though it's not DC? Why I'm asking, is that I can't seem to find literature on the subject, but a lot on the DC offset problem.

I think the "DC offset problem" isn't about the received signal being mixed down to DC. That's simply a normal function of the receiver chain and mixer. The problem is when the LO leaks into the RF chain and ends up on the RF port, and then mixes down on the IF port to DC, resulting in a DC error. This is very problematic in systems where the RF receive chain has high gain.

DC offset in a ZIF receiver due to L.O. leakage, it is not caused by crystal tolerances. It is caused by L.O. phase shift in the RF input path compared to the L.O. applied to the mixer. A fixed leakage, both phase and amplitude can be compensated for. Variable leakage, like variation as hand is placed on case of cellphone, must be minimized.

Mixer circuitry itself is often the cause of DC offset. Matching of Gilbert cell mixers for IQ baseband must have tight tolerance and some method for compensation. Variation in L.O. quadrature applied to mixer is also a problem. Variation is baseband gain and phase must be compensated.

Just about all recent cellular receiver frontends use ZIF topology for receivers. They have a self calibration algorthms that comp out much of the variation. DC offset is an important factor in second order I.M. which is very important in ZIF receivers.

Receiver frequency offset to desired signal is something different. Within limits the I/Q demodulator can wash out frequency offset. The demodulator also provides frequency correction info that is used to zero in the RF frequency.

Thanks for the clarifications. Can you please give an example of an I/Q demodulator IC or RFIC which has this frequency offset compensation? If such compensation is not present, can a filter be used to wipe out the non-zero carrier.

booblik said:

Thanks for the clarifications. Can you please give an example of an I/Q demodulator IC or RFIC which has this frequency offset compensation? If such compensation is not present, can a filter be used to wipe out the non-zero carrier.

Click to expand...

The TI TRF3710 Zero IF demod receiver IC family (like the TRF371125), which we had used (in canceled Navini/Cisco WiMAX P4 product) has programmable (internal D/A) for trimming DC offset.

Chip info and the datasheet: Broadband RF/IF & Digital Radio - IQ Demodulator - TRF371125 - TI.com

See also page 37 of the PDF datasheet which describes the "Automated DC Offset Calibration" which can be invoked in the chip for this purpose (or, one can do their own DC calibration if preferred). The chip is 'talked to' via simple SPI bus interface to send simple commands to the chip as described on page 30 of the datasheet.

Jim

(PS I am not the rep, but I worked a lot with this chip during development of H/W and S/W on P4 product.)

Depending on modulation properties of the transmitted signal, AC coupling caps can be used in the baseband but the low frequency pole must be compatible with the signal. It will put a low frequency hole in the recovered signal spectrum. It is not used much in modern designs as digital offset controls are usually much better. A cap coupled baseband can also cause problems in achieving initial frequency correction because of the hole in recovered I/Q baseband signal.

For WCDMA HSDPA modulation, a cap coupled baseband is very damaging to multiphase signal recovery. For GSM phones, very low I.F. frequency baseband is often used to avoid some of the DC offset problems and avoid the poor low frequency 1/f noise of CMOS mixers. The I.F. is selected to place the image near the 200 kHz adjacent channel spacing.