Hello,

I am trying explain some “mystery spur signals” which appear on my signal analyser’s display when using my signal analyser with a mm-waves harmonic mixer, and the signal ID feature if turned OFF:

I have on my bench an N9010B-544-EXA connected to an extended V-band mixer, M1970V-002 (50-80GHz), and a 67GHz PSG to act as a source. The 2 instruments (PSG + EXA) share the same frequency reference.

In many cases I came across spur frequencies which I could easily explain. In those "simple to explain cases", the mathematical expression of the spur signal's frequency complies with the general expression I know for harmonic

mixing, namely: f_if = n*f_LO + m*f_signal, where:

f_signal is the actual signal present in the input of the harmonic mixer,

f_IF the analyser’s IF frequency (322.5MHz in the case of the N9010B-544),

f_LO the LO frequency (coming from the analyser into the harmonic mixer) and

N,M are integers.

However, strangely enough, I also encountered some "hard to explain" spur frequencies on the display, for which the frequency equation I just found didn't make sense.

**In order to be clear and to the point, lets discuss the following set-up:**

In all the following examples, Ihave injected a clean CW source at f_cw = 62.5 GHz into the M1970V

mixer. To my best understanding, the analyser's IF input is at 322.5MHz, and the harmonic number N for the LO, selected automatically by the analyser when connecting the smart mixer, is "6-", i.e. the tuning equation for this configuration should be:

** f_LO = (f_signal_in + 322.5MHz)/6**

In addition to the "real" CW signal, seen on the display at 62.5GHz, many spur signals are also displayed across the whole extended V-band.

An example for a spur signal that I did manage to explain is the following one:

**f_explained_spur = 53.295GHz. **

According to the tuning equation above, I have calculated this spur’s related LO frequency to be f_LO = 8.8825GHz (this is, to my best understanding, the instantaneous frequency of the swept LO, when the “blip” of the spur is updated on the analyser’s screen). Therefore, this “easy to explain” spur is related to the actual CW input (at 62.5 GHz) frequency by the following simple relation:

f_IF = f_CW – 7f_LO (i.e. an M=1, N=7 product).

Up to now, everything was OK and as expected...

The reason for my post is to ask for assistance in explaining some other signals that I could not explain.

**I will give now 2 examples for specific spur signals that I could not explain:**

**f_unexplained_spur1= 57.072115 GHz:**

This “strange” spur appears on the display at 57.072115GHz (again, the actual CW input to the mixer remains 62.5GHz), corresponds to an instantaneous LO frequency (at the moment the sweep passes over this spur on the display) of 9.565769167GHz, yielding a "strange" frequency relation of: 2f_IF = 2f_cw -13f_LO.

This equation does practically work, but does NOT explain why the spur is actually present at f_IF (the equation implies the product will be around 2f_IF).

A second example for an unexplained spur signal on the display is:

**f_unexplained_spur2 = 57.6675 GHz:**

Doing the math in this case again for the instantaneous LO frequency required to show this spur, yields an LO frequency of f_LO = 9.665GHz, which complies with the "strange" frequency relation: 2f_IF = 13f_LO - 2f_cw

**My question is as follows:**

I found these 2 spur signals hard to explain, as they obviously do manage somehow to enter the 322.5MHz IF input of the N9010B analyser in practice, while their frequency equation implies these products are only expected at

2f_IF, not at f_IF.

**Can anyone help me explain why these frequencies are in fact on the display, while their equation only shows their product to be expected around 645MHz (2f_IF)?**

Many thanks in advance,

Oren Hagai

INTERLLIGENT RF & MICROWAVE SOLUTIONS

Hello Oren_Hagai,

Good observations! It may seem like a coincidence that you're seeing these spurs at 2f_IF, but this is explained through how mixing is done to achieve these high (in your case, V-band) frequency ranges.

Smart mixers double the LO frequency in a method called second harmonic mixing. The LO is driven hard enough that the second harmonic is excited to a level close to the first harmonic. This technique is common when trying to achieve these very high frequency ranges, so there are actually 2 * 6 = 12 harmonics (the smart mixer chooses N=6 harmonics automatically in your range). Further, the even harmonics are optimized for, so odd harmonics should have a conversion loss of 15 dB, and the end result emulates an N=6 response.

A good technical overview for specifications and usage of our waveguide harmonic mixers is here. I referenced specifications from page 3:

https://literature.cdn.keysight.com/litweb/pdf/5990-7718EN.pdf?id=2089022

Second harmonic mixing explains why harmonics exist at 2f_IF, instead of just f_IF. In your analyzer, are you seeing these unexplained harmonics at lower levels (the specs suggest 15 dB) than your even harmonics? Or are they are similar levels? An image that you are seeing would help here.

If the spurs are too high for your application, turning Signal ID on will help remove some of them.

Hope this helps!

Bertie