phase noise effect on NF

[kh4n]

Hi all,

My aim is to measure NF of a downconverting mixer, IFout is around 200MHz. I have an agilent PSA for NF measurement and my LO source choices for the mixer are an agilent sinal gen. without ultra low phase noise option and a PLLVCO IC from Hittite. Hittite part has better far out phase noise performance, ie phase noise at 10M offset is 9 dB better than agilent, and it goes flat through higher offset freqs(let me remind my if is around 200M). This effects my NF.
Agilent has better PN at offset frequencies upto 500kHz, as expected.
Previously I used a cavity PBF from KnL around LO, of course that helped. But after finding more about mixer I will desing for a decent LO.
Do you have any comments and/or suggestions on this topic, (biasing and matching improvements are already done at mixer side)? I am especially asking about LO effects on NF.

Thanks in advance...

 

[kspalla]

for mixers LO PN at near i.e. 500KHz offset is critical. IF freq of 200MHz does not mean PN offset is 200MHz.
IF band width makes sense on LO PN contribution.

Try to imagine nd analyse the IMDs or mixer action.

Regards,

 

[jiripolivka]

Hi all,

My aim is to measure NF of a downconverting mixer, IFout is around 200MHz. I have an agilent PSA for NF measurement and my LO source choices for the mixer are an agilent sinal gen. without ultra low phase noise option and a PLLVCO IC from Hittite. Hittite part has better far out phase noise performance, ie phase noise at 10M offset is 9 dB better than agilent, and it goes flat through higher offset freqs(let me remind my if is around 200M). This effects my NF.
Agilent has better PN at offset frequencies upto 500kHz, as expected.
Previously I used a cavity PBF from KnL around LO, of course that helped. But after finding more about mixer I will desing for a decent LO.
Do you have any comments and/or suggestions on this topic, (biasing and matching improvements are already done at mixer side)? I am especially asking about LO effects on NF.

Thanks in advance...

Noise figure of a mixer is the effect of THERMAL noise. PHASE noise of the local oscillator CANNOT affect the NF as it is exactly orthogonal to the thermal noise. In other words, the phase noise is similar to jitter- the PN spectrum shows how fast the carrier is moved due to vibrations and thermal noise pushing the carrier frequency.

Due to this orthogonality, mixer or receiver NF cannot be affected. BUT any oscillator also has the AMPLITUDE noise, directly the result of the thermal noise on carrier AMPLITUDE. A noisy oscillator does have an AMPLITUDE as well as PHASE noise. To counter the effect of the first (amplitude) noise on mixer NF, we prefer to use BALANCED mixers; in such device, the LO AM noise is reduced by up to 20 dB by the balanced structure.

 

[kspalla]

I agree on Thermal noise. the point is PN also has impact.
Amplitude noise in LO is something i am new and interesting to me. if AM noise is important why to use balanced structure why not some clipper circuits?. Also Many questions i am raising here. we see double and triple balanced mixers each for IF, LO and RF.
Does this mean we are handling Amplitude noise of RF and IF signals apart from LO? i guess not.
I believe balance is the method to improve RL on port that intern improve the isolation.

Sorry if i mistake and request jiripolivka to give more explanation.

Regards.

 

[kh4n]

Noise figure of a mixer is the effect of THERMAL noise. PHASE noise of the local oscillator CANNOT affect the NF as it is exactly orthogonal to the thermal noise. In other words, the phase noise is similar to jitter- the PN spectrum shows how fast the carrier is moved due to vibrations and thermal noise pushing the carrier frequency.

Due to this orthogonality, mixer or receiver NF cannot be affected. BUT any oscillator also has the AMPLITUDE noise, directly the result of the thermal noise on carrier AMPLITUDE. A noisy oscillator does have an AMPLITUDE as well as PHASE noise. To counter the effect of the first (amplitude) noise on mixer NF, we prefer to use BALANCED mixers; in such device, the LO AM noise is reduced by up to 20 dB by the balanced structure.

I guess I could not explain myself clearly. I am aware that close in phase noise does not (much)effect mixer NF but far out PN, which may be the same as AM noise of LO you are referring to (i dont know are they the same), has an effect on NF since it adds to noise floor. My concern is that when measuring IF at 200M, LO+200M noise floor level should clearly effect mixer NF, am i right?
If I am wrong please also clearify, why adding a BPF to supress noise floor level at LO+200M, reduces NF?

thx again...

 

[kspalla]

I agree that far noise is also a concern. i am not saying to neglect it. IF far noise level is high then certainly noise floor gets raise.
but I disagree on point "near noise is not important".
I also has Q that does LO+200MHz PN level is important? not the level before or after it?
do we neglect the IF band width and image problems?

Regards.

 

[jiripolivka]

I agree on Thermal noise. the point is PN also has impact.
Amplitude noise in LO is something i am new and interesting to me. if AM noise is important why to use balanced structure why not some clipper circuits?. Also Many questions i am raising here. we see double and triple balanced mixers each for IF, LO and RF.
Does this mean we are handling Amplitude noise of RF and IF signals apart from LO? i guess not.
I believe balance is the method to improve RL on port that intern improve the isolation.

Sorry if i mistake and request jiripolivka to give more explanation.

Regards.

Your message was sent into my spam file. Please make sure there is no virus in it.

It is difficult to offer more explanation. Thermal and phase noise are mutually orthogonal, so the PN cannot affect NF in mixers. Diodes (or other P/N junctions ) in mixers are pumped b LO power into alternate switching ON/OFF. Some clipping could take place but oscillators with AM noise do affect mixer NF. The balanced structure offers to suppress AM LO noise by >10 dB, so the NF degradation is less.
The phase noise can and does affect the communication receiver due to its demodulator. If FM or PSK is used, then phase noise directly expands the vector phase uncertainty. With AM modulation, receiver NF can degrade the S/N ratio.

Mixer RL has nothing to do with any type of noise. RL is the degree by which mixer can receive all or a fraction of power from the LO or RF source. RL = 10 dB means one tenth of power gets reflected back to a source. No noise involved.

- - - Updated - - -

A bit more on "clipper circuits". Clippers affect the amplitude on its peak while noise comes at the "bottom". How can one clip a microwave LO signal? Limiters only clip an excess power, to protect sensitive devices. They cannot affect amplitude noise but only signal peak power. AM noise also has a spectrum around a carrier like PN, but ORTHOGONAL to it. Those spectra cannot affect the mixers but only receiver demodulator far, far away.

 

[kspalla]

Thats great explanation.
Is it correct that bad PN of LO affects FM and PSK receivers and it does not effect AM receivers?
My view is PN tops ( add up may be vector sum) the thermal noise and result degraded S/N. it also affect AM receivers.
If two signals that are separated ( in amplitude dB) by more than LO PN dBc/Hz at any freq offset point, then it is not possible to detect weaker signal as it burred under LO PN. I studied this part in coherent lo phase noise LO for Radar design.

Regards.

 

[tony_lth]

I don't think radar is an AM receiver, doppler radar uses freq shift, so it should be also a FM/Phase modulation receiver.

 

[kspalla]

OK, lets continue brain strom on AM receiver and LO phase noise impact. I agree that radar is generally PM receiver.

Let say two am signals one at f1 with -20dBm and the other f2 located 100KHz offset with -80dBm power. The sensitivity (due to thermal noise etc. addon) is -85dBm. but the LO PN at 100Khz offset is may be 50dBc/hz let say and 70dBc at 1MHz. Can this possible to tune f2?
Not possible in my opinion as it submerged under LO PN. But f2 power level of -80dbm can be tuned if f2 is 1MHz away.
Any one agree /disagree?

Regards

 

[jiripolivka]

In my opinion the problem if any is not the source PN but a possible non-linear behavior of the last stages of the AM receiver and its detector. The front-end mixer will not take any part in the game.
In radar receivers such situations are treated by design, like using a fast-acting AGC, to detect close as well as (latter in time) the far target returns.
In radars, the PN is only important in processing Doppler.

 

[kh4n]

i spent some time on the setup reliability and made some measurements. I guess, far out LO PN, ie noise floor at lets say 100M offset, effects the NF. Close in phse noise is less dominant.
LO source and SA are locked by the way.
I am measuring LO phase noise upto 100M offset at SSA and then measure NF at SA. DUT system has limiting LO amp, passive mixer and if amp stages. If i use pllvco IC as lo source withut any filtering, I can get beter NF compared to Agilent SG. If I use BPF+Agilent as LO source then NF results are better than only agilent case and similar to PLLVCO case. PLLVCO has better farout noise but worse closein PN performance.
For future reference, the DUT system I was trying measure has, as expected by design, more than 10dB NF. PN and noise floor contributions to NF would most probably have changed if I tried to measure a DUT with 2 dB NF.
i am not trying to design a receiver here, my subsystems only contains lo source+mixer+if amps+lo amps. I am trying to improve its parameters such as nf, iip3, cg etc. after that point, some agc things will go on.

 

[tony_lth]

Hi, the root cause of improvement of NF, in my view, is that you used filter to filter other harmonics or spurs of the LO signal, then there are less spurs in mixer in the working band, and that can sharply decrease the NF.
BTW, too bad PN of LO might mean that there are some spurs in the LO signal spectrum.

 

[kh4n]

Hi, the root cause of improvement of NF, in my view, is that you used filter to filter other harmonics or spurs of the LO signal, then there are less spurs in mixer in the working band, and that can sharply decrease the NF.
BTW, too bad PN of LO might mean that there are some spurs in the LO signal spectrum.

If we compage SG with PLLVCO IC, PLLVCO has worse harmonic performance however gives better NF, thus that might eliminate harmonic effect.
But spur performance of SG is better.

I also filtered RF signal to reduce LO harmonic effects but that also did not affect much.

 

[tony_lth]

can you give details of your SG and PLLVCO spectrum data?

 

[kspalla]

Hi,
interesting on LO PN on sensitivity.

 

[jiripolivka]

All the time we have talked about the effect of oscillator phase noise on mixer or receiver NF. I assert that due to the orthogonality of phase noise and thermal noise, in a mixer no effect can be seen on the NF.
Now you said you have measured oscillator phase noise. Can you measure AMPLITUDE noise? This can affect mixer NF directly. To solve this problem, most modern mixers are BALANCED, so LO AM noise is suppressed by the balancing effect. A typical O AM noise rejection by this is >10 dB, so only one-tenth of AM noise power affects mixer or receiver NF or noise floor.
LO noise may affect receiver noise floor in latter stages. Due to this fact, most digital communication systems use PSK of various types, so their limiter reduces any AM noise or signal variation.