How Lo Nulling Works?

Hi,
I have found a circuit for reducing the LO leakage of a mixer. Could you explain how this circuit works? Do you have alternative solutions to LO leakage reduction?
Kind Regards

LO leakage depends on perfect impedance matching which is often frequency dependent. To compensate for filter, load or amplifier impedance mismatch effects on the MIXER LEAKAGE, one can improve LO leakage by injecting a small amount in the reverse direction of output stage of a directional coupler with an equal amplitude and opposite phase.

Phase cannot be measured of the LO leakage vs frequency , but Amplitude can be measured, if no input signal is applied with the detector diode (cut-off in the photo top-right )

By sweeping I & Q bias the level to control amplitude and phase on the cancellation signal , which is then attenuation by the return loss of the Directional Coupler the optimum LO Leakage can be determined by the minimal output on the Forward DETECTOR diode. ( this analogous to an echo chancel) I & Q are 90deg spaced signals from the LO's 3.5dB splitter and 10dB boost Amp.

To minimize loss in this stage the Directional Coupler (DC) can be chosen with lower loss and small taps like -15 to -20 dB. ( e.g. DC-15, DC-20) The LO must be boosted 10dB in order to cover a certain range of mismatch from poor return loss from LO to output which results in the leakage thru to output.

This is a classic approach used by Network Analyzers to obtain very high performance. Now with DSP radio's , and requirements for low LO leakage to antenna, this make modest improvements , when all other methods of impedance matching and effects on Mixer cannot be improved.

This requirement comes from making mixers operating with input frequencies over 9 decades (Hz to GHz) which normally function well in only a limited range of 1 or 2 decades and gives exceptionally good results not obtainable by conventional design. The IQ levels would then be stored in memory until some manual or auto-calibration cycle for updates.

your diagram does not make ANY sense to me. They have a downconverted IF output of the mixer, so why not just use a lowpass filter after the mixer?? And even if they DID want to do it that way for some reason, the directional coupler is drawn backwards. I would say they did not think this thru.

Actually it is almost certainly an UP converting mixer...

The thing you fail to notice is that the input range is 3Hz -> 3GHz!

Given that an input signal near the bottom of the range will be painfully close to any LO breakthru @3.900000003Ghz (LO breakthru @ 3.90GHz), which is surely inside the first IF filter bandwidth (Assuming a first IF centre @ 3.9GHz), when the range extends closer to DC then the passband of the filter you need some way to suppress the leakage to a very high degree.

The hybrid and I/Q multipliers provide a simple means to control the amplitude and phase of the nulling signal, and the coupler provides a means of injecting a small amount of the nulling signal without seriously impacting the level of the wanted components at almost the same frequency.

Regards. Dan.

This is how a Network Analyzer works with Up conversion to 3.9GHz IF to span 9 decades .

The tapped DC is show correctly to improve LO leakage by orders of magnitude over a 3GHz span. THe DC with Mixer S22 can be used to cancel LO leakage, by inject a conjugate LO signal in reverse until nulled and stored in memory for each step, then interpolated with some algorithm.

Hi SunnySkyGuy,
Thanks for the answers.
I guess mixers on the feedback loop functions as a variable attenuator. Also, they can change the phase with 0/180 degrees. Hence, they can generate required feedback. Is it true?
I could not figure out what is meant by
“To minimize loss in this stage the Directional Coupler (DC) can be chosen with lower loss and small taps like -15 to -20 dB. ( e.g. DC-15, DC-20) The LO must be boosted 10dB in order to cover a certain range of mismatch from poor return loss from LO to output which results in the leakage thru to output.”
I am unaware that “This is a classic approach used by Network Analyzers to obtain very high performance.” Since I have not inspected any of block diagram of a network analyzer. Do you have any application note or detailed diagram?
In the photo given, there is a ring resonator on top right. I guess instead of a 50 ohm termination, there is a power sensor, in order to measure LO leakage.
Kind Regards

The operation principle of the LO leakage cancellation circuit isn't feedback, it's feedforward.

And yes, there must be a means to measure the leaked LO power, preferably with a phase sensitive detector. Then adjust the compensation circuit iteratively until the leakage signal is nulled.

I would expect that the measurement is at least briefly discussed in the instrument description where you took the diagram from.

metu said:

I could not figure out what is meant by
“To minimize loss in this stage the Directional Coupler (DC) can be chosen with lower loss and small taps like -15 to -20 dB. ( e.g. DC-15, DC-20) The LO must be boosted 10dB in order to cover a certain range of mismatch from poor return loss from LO to output which results in the leakage thru to output.”

Click to expand...

This Directional Coupler likely uses -20dB taps for low ~0.2dB forward loss. THus LO must be boosted 10dB so that enough reflected LO into the s22 value of the mixer so that it reflects forward to the output So imagine if LO leakage was -30dB and return loss of mixer output was -20dB and output DC-20 reduces that to -40dB

To cancel the -30dB leakage you need to drive +10-40dB to reflect -30dB of opposite phase forward to the output and cancel the leakage for each LO frequency.

This is a fancy directional feed forward cancellation for LO leakage suppression using reverse coupled Return loss of mixer output... for fancy Network Analyzers with high end performance spanning 140dB range with return losses 40dB better than most radios.

Although it is a feedforward correction, the feedback is the Diode DETECTOR to measure the LO Leakage during a calibration run for correction and storage in memory.

yes the I & Q modulators can synthesize any amplitude and phase within a certain range of amplitudes.

You mean, LO is feed back to mixer IF port and reflected signal from mixer output is cancelled with LO leakage. Instead of directly summing at the coupler, I could not figure out why such a feed forward is applied.
In mixer circuits, it is not desirable to have reflections from output. Here, we introduce injection to the IF port on purpose. I believe we may have performance degradation due to feed forward if it exists.

- - - Updated - - -

There is two couplers at the top right. I guess the one on the top is measuring leakage and providing required feedback. The coupler below is providing feed forward to supress LO leakage. Is that correct?
Kind regards

In terms of PCB layout,
your I/Q signal trace length should be as identical as possible.
Otherwise, the LO leakage will become more serious.

I guess, with variable I bias and Q bias settings, one can solve leakage problem. This is the major function of the feedforward circuit.