Harmonics issue while measuring in Spectrum Analyzer

[shashy.br]

HI all,

I recently designed and tested an L band Local oscillator ,

I found that the Module while testing individually is working fine without any harmonics in its output ( or <-70 dBc).

But while testing it with a different spectrum analyzer i found that the harmonics are present at < -50dBc.

Don't know if the issue is with my module or the spectrum analyzer.:-(

Pls comment if any suitable explanations are available. !!!!!

 

[Terminator3]

maybe second analyzer somehow reflects LO and enaches harmonic output?

 

[rf1008]

Hello,you may clarify which SA has degraded measurement performance such VSWR or amplitude display by measuring one GOOD/STANDARD DUT or using one signal generator directly.
Have you compared the output power and frequency of your DUT designed with these two SAs? any difference?

At last,for harmonics measurement, one 3dB SMA attenuator should be added between your DUT and SA, and even one HPF/Notch when your base frequency has large amplitude.

 

[shashy.br]

Terminator,

I am not familiar with such degradation as mentioned by you, do you have any links which explain it in details?

rf1008,

The measured power in both cases is -10 dBm (i.e measured using both spectrum analyzers)

I provided a the input from a Signal generator to the Spectrum analyzer and found that the same harmonics are produced.

Hence i decided that the spectrum analyzer is the problem.Is my conclusion wrong?

 

[rfmw]

Put a 10dB attenuator in between the oscillator and spectrum analyzer. Reduce the RBW to measure the harmonics reliable well above the noise floor. Check if the harmonics are still at the same level (~50dBc).

 

[jiripolivka]

Put a 10dB attenuator in between the oscillator and spectrum analyzer. Reduce the RBW to measure the harmonics reliable well above the noise floor. Check if the harmonics are still at the same level (~50dBc).

Most spectrum analyzers utilize an input mixer, and often several down-conversion stages. The first mixer is the device that can create harmonics when overdriven by an input signal exceeding specifications.

To be sure what happens, always use the RF input attenuator (many spectrum analyzers have one) and check that the signal peak on the display decreases or increases following the attenuator action, usually in 10-dB steps.
Some "lower-frequency" spectrum analyzers can be complemented by external harmonic mixers pumped by internal LO. Agilent sells such mixers and offers them calibrated over appropriate frequency bands.

Many RF oscillators generate harmonics. some spectrum analyzers can in "full-band" operation, display the chain of harmonics, to compare approximately their amplitudes. It is interesting to use a low-pass filter to see its effect on such harmonics. Band-pass filters, however, have a harmonic response, so they are not suitable to clean an oscillator spectrum.

 

[shashy.br]

Thank you for replying,

I will take a measurement with an attenuation, and will get back with the results!!!

 

[RF_Jim]

HI all,

I recently designed and tested an L band Local oscillator ,

I found that the Module while testing individually is working fine without any harmonics in its output ( or <-70 dBc).

But while testing it with a different spectrum analyzer i found that the harmonics are present at < -50dBc.

Don't know if the issue is with my module or the spectrum analyzer.:-(

Pls comment if any suitable explanations are available. !!!!!

shashy.br, in the real world, engineers do not trust their spectrum analyzers without going to some extra lengths.

For instance, I believe someone up thread mentioned putting in a few dB more attenuation with a pad, and this should have a significant impact on the harmonics displayed on the spectrum analyzer *IF* the harmonics are being generated in the first mixer in that spectrum analyzer.

So, applying 3 dB pad to the input the displayed 2nd harmonic may drop by 6 dB if the mixer in the spec an is the source of the harmonic energy.

Also, it is VERY helpful to have some sort of notch filter IN ANY CASE to notch or attenuate the FUNDAMENTAL frequency as a matter of general course ... be sure that the DUT (test device) still see a 50 Ohm load, so place notch filter at some point further down stream. By using notch the filter, one can be a little more assured that the 2nd harmonic energy is from the DUT rather than the spec an 1st mixer!

RF_Jim

 

[G0HZU]

When measuring harmonics with an analyser you are supposed to follow a basic rule of thumb where the internally generated harmonics (in the analyser) should be about 15dB lower than the harmonic level you are measuring. Otherwise you can get significant measurement uncertainty due to the harmonics summing or cancelling inside the analyser and giving a false indication on the analyser display.

eg if you keep the internal harmonic 15dB below the DUT harmonic then you keep the measurement uncertainty to just under +/-2dB. If the internal harmonic is only 5dB lower than the DUT harmonic then the uncertainty window grows to about +4dB to -7dB depending on summing or cancelling.

 

[FvM]

When measuring harmonics with an analyser you are supposed to follow a basic rule of thumb where the internally generated harmonics (in the analyser) should be about 15dB lower than the harmonic level you are measuring. Otherwise you can get significant measurement uncertainty due to the harmonics summing or cancelling inside the analyser and giving a false indication on the analyser display.

A prerequisite of this method is to know the harmonics specification of your spectrum analyzer. I'm not under the impression that the OP did read it yet, because it isn't mentioned in this thread at all.

To measure low level harmonics of an oscillator/transmitter, some kind of additional preselection is usually required.

 

[shashy.br]

Hello all,

I tested the system again using a 20 dB attenuation at the Spectrum analyzer input,

still the same harmonics were present as measured previously but with reduced power levels due to the 20 dB attenuation.

The attenuation did not help reduce the harmonics dBc

 

[Terminator3]

Maybe i give a strange idea: how is your design powered? Is there any ground connection (for example, coaxial line) with your design? If you use some 220v to DC power source, try to use battery or DC-DC in that case and see what you get. I believe some improper grounds with AC voltages can provide some strange modulation -> "harmonics".

And also, what your harmonics look like (at wich freqs, spasing...)

 

[biff44]

most spectrum analyzers have a step "input attenuator". Modern ones compensate for what the input attenuator is set to with simple math.

What u want to do is change the input attenuator by 10 dB. Ignore what the base noise floor does. But look at the spurious levels in dBc.
Does changing the input level by 10 dB have a significant effect on the displayed spur level in dBc? If it does...then it is your test equipment causing the erroneous reading

 

[shashy.br]

Termonator 3,

I am getting harmonics of the fundamental frequency 1500Mhz (i.e 3000, 4500 only ).

i will check the grounding and get back with the result.

biff44,

I will check for the input attenuation and see what the results are.!!

 

[shashy.br]

Hi,

i Checked the system again,

Found that while interfacing the Exciter and receiver the Harmonics are getting generated in the Exciter .

I am considering a poor return Loss between the Exciter and receiver causes the harmonics generation. Is this correct ?

 

[jiripolivka]

Hi,

i Checked the system again,

Found that while interfacing the Exciter and receiver the Harmonics are getting generated in the Exciter .

I am considering a poor return Loss between the Exciter and receiver causes the harmonics generation. Is this correct ?

I do not think any mismatch or return loss can generate harmonics.

First you had to test your spectrum analyzer to be sure it does not generate harmonics.

Now you detected the harmonics are generated in Exciter. This means there is one stage overdriven by input signal, so it generates harmonics.

Use your spectrum analyzer and check the input signal (oscillator), then each of following stages, to see where the harmonics occur.

If the oscillator itself generates harmonics, then the best way to reduce the harmonic power is to insert a low-pass filter that will only pass the nominal frequency.
If one following stage generates harmonics, adjust its gain and P-1 dB to improve the Exciter.

 

[biff44]

I have seen odd things related to load impedance cause harmonics. but as stated above, it is likely one of the output stages in the exciter is overdriven.

Exciters....are hard to design. are there specifications written down, and was there a first article test done? a high 2nd harmonic would be obvious after the 1st minute of testing.

How about just slapping a lowpass filter in, and then shooting the exciter designer.

 

[shashy.br]

Thank you jiripolivka and biff44 for replying,

I am thinking of Connecting a 2.7pF capacitor between RF trace and the Ground to make it act as a LPF,

That way i can reject the Second harmonics of 1500MHz Fundamental frequency.