Help in understanding signal output from a frequency mixer

[samosmireno]

Hello,

I have a frequency mixer TFM-3MH+ where I need to output a low frequency signal of 10-100Hz

The signals that I'm sending at input ports from a signal generator are:

- Sine wave of frequency 6.9 MHz at LO port (+17 dBm)

- Sine wave of frequency 6.9 MHz+10Hz at RF port (-39 dBm)

So, the output of the mixer should be a sum of: frequency sum (6.9 MHz+6.9MHz+10Hz) and a difference (10Hz).

When I filter out the high frequency component, the signal that I'm getting is:

https://pasteboard.co/TuZ7eU7b94wS.png (10Hz)

https://pasteboard.co/hURPGvHgLK2P.png (80 Hz)

Can anyone explain what can cause these glitches in the signal?

How to avoid them?

Can I add some circuit after the output to smooth the signal?



The spec of the mixer is here:

https://www.minicircuits.com/pdfs/TFM-3MH+.pdf

[fourfathom]

I don't know how your scope behaves, but it it possible that it is undersampling the sum signal (6.9M + 6.9M + 10), and so displaying an alias at 80 Hz?  Try reducing your scope time base to something like 10ns/div and see what it shows.

[TimFox]

Also, there is a finite rejection of the LO power into the IF output.  According to the datasheet, this is something like 50 or 55 dB.  Therefore, one expects +17 dBm - 50 dB = -33 dBm or so of 6.90 MHz at the IF output.  The conversion loss (RF to IF) is roughly 4.5 dB, so the expected IF output from -39 dBm at the RF port is lower than the LO leakage, about -43.5 dBm.  How much rejection of the 6.90 MHz signal do you get when you filter out the high-frequency components (6.90 and 13.80 MHz)?  When the 'scope is set to display a 10 or 80 Hz sine wave, it will probably undersample the 6.90 MHz signal that makes it out of the filter, which will give strange aliases.

[edpalmer42]

How did you filter out the high frequencies from the IF output?  In the 10 Hz picture, the vertical part of the glitches implies a high frequency.  If the high frequency components were filtered out before getting to the scope, the glitch must be something generated by the scope.

[fourfathom]

Good points above.  What is the filter you are using, and for that matter, how do you get the separate 10Hz and 80Hz displays?  If you have both of these frequencies on the mixer output the scope should be displaying the additive sum of the two, or perhaps one modulating the other.  But if your two inputs are clean there shouldn't be an 80Hz component in the mixer output, so I suspect some sort of scope sampling alias.  And that glitch is "interesting".

[w2aew]

How did you filter out the high frequencies from the IF output?  In the 10 Hz picture, the vertical part of the glitches implies a high frequency.  If the high frequency components were filtered out before getting to the scope, the glitch must be something generated by the scope.

I agree that this looks a bit like a scope artifact - especially given that you've already low-pass filtered the signal, a step change that is this fast would have been at least softened by the LPF.

[TimFox]

In general, with periodic sampling in an oscilloscope, if the frequency of the displayed sine wave changes when you change the time base (sec/division), then the displayed wave is an alias frequency.
Years ago, the first good digital oscilloscope in our lab at work was a decent LeCroy with CRT display and a deep memory (for that time) of 10k points.  We were troubleshooting a large system and seemed to find a strong AC magnetic field at close to (but not equal to) 60 Hz, using a search coil.  We looked all over the lab for the source, but when we changed the time base and saw the frequency change, we knew it was an alias.  Speeding up the time base, we found that the signal displayed was actually from a 10 MHz system clock:  the apparent 60 Hz (very clean sine wave) was due to the approximately 6 ppm difference between the internal clock of the LeCroy and the actual offending clock generator.

[samosmireno]

Sorry for the late reply, it turns out that the problem was that the two signal generators weren't phase locked to a common reference. When I locked them, i got a smooth signal.

I got my answer at stack exchange, you can read the whole explanation here:

https://electronics.stackexchange.com/questions/591567/help-in-smoothing-signal-output-from-a-frequency-mixer

Thank you everyone for your help!

[radiolistener]

Can anyone explain what can cause these glitches in the signal?

your picture definitely shows aliased waveform. It means that your oscilloscope showing incorrect waveform due to low sample rate (sample rate is not enough to display actual waveform).

It looks that this is alias for 6.9 MHz+6.9MHz+10Hz. Try to setup 10 ns/div and check, if you see high frequency sine, then this is true.

In order to remove 6.9 MHz+6.9MHz+10Hz from output, you're needs low pass filter.