Unexpected IF Outputs

Thank Enash, to answer your points:

1. Yes, around 120mA as per the documentation. Current consumption increases when I'm in offset mode to around 140mA, but this seems reasonable to me.
2. The two generators are not locked together with a reference, but the offset seems consistent - I don't think that is playing an issue here. I don't have the ability to lock the instruments within my lab - sorry.
3. That makes sense to me, thank you for confirming.
4. Unfortunately I only have one RF gen capable of GHz frequencies, so an easy setup like this is impossible. I am trying to perform the mixing externally, but this will take some time to put together for a reason I'll go into below.

To me, the output looks like what I might expect to see if the offset mixer was not rejecting the image frequency. My theory is a bit rusty in this area, apologies, but if the image is present, then when that is mixed down to the IF, it cannot output an arbitrary phase - if we have:

\textbf{Desired signal:} \[ A e^{j(2\pi f_{\text{IF}} t + \phi)} \]
and
\textbf{Image signal:} \[ A e^{-j(2\pi f_{\text{IF}} t + \phi')} \]

Then in amplitude and phase are similar, when they are summed one gets:

\[ x_{\text{total}}(t) = A \left[ e^{j(2\pi f_{\text{IF}} t + \phi)} + e^{-j(2\pi f_{\text{IF}} t + \phi)} \right] = 2A \cos(2\pi f_{\text{IF}} t + \phi) \]

i.e. a real signal only - quadrature information is gone. In effect the phase space collapses because the complex part is gone. To me this would occur when either the mixer is not rejecting the image, or the LO filtering is poor.

As I say, this is only my guess based on what I see when connecting short and open - I imagine I am driving//configuring the setup wrong, rather than there being an issue with the silicon.

To try to test with a different LO, I will try mixing myself - I don't have an image reject mixer so I will try using a tuned yig filter to isolate the LO+IF component (using a venerable HP 8445A Preselector! ). Trying to isolate the frequency component is probably at the limit of this instrument though, so I can't promise a result.

-

Good morning.

Unfortunately my plan to externally mix and isolate a single LO frequency doesn't work in practice - the insertion loss of my setup is too high for me to amplify with components available to me at present.

Interested to see what results you find, enash.

Hi enash, quick update...

I've actually managed to use an old LNB as an LO - yes terrible precision, but it's enough to be able to test at a single frequency at about 10 GHz. With this, the outputs behave as I would expect - producing sinusoidal waveforms instead of square, which freely move in phase. I measure the reflected waveform about 3.9dB down when compared to the incident, which I think makes sense if the device has ~2dB loss each wave (out and in from the test port). I have tested with open shorts and a load, and the output behaves as expected for each - I'm able to calculate quite sensible real and imaginary values for the loads I have tested.

To me, this suggests three options (no particular order):

1) I understand how the device works in bypass mode, but am missing something important in the documentation about the setup of offset mode.

2) The documentation is missing some important context on setting up offset mode.

3) There is some issue with the device itself.

Hope this is useful to assure you that the device is functioning in at least one configuration. Hope your Linduino issues get fixed without too much bother.

Pictures of Offset Mode Setup related to previous thread

Thank you Eamon, very useful to know we get similar results.

Whilst you wait to hear back from colleagues on the square wave output, could I ask if, when you are in offset mode, are you able to see continuous phase shifts between input and reflected waves? Ie, can you observe frequency shifts other than 0 and 180 degrees? It might be a pain to test, but are you able to observe these intermediate phase shifts when using the output from the splitter?

I am expecting a second RF generator to arrive soon. However, the Marconi 6311 cannot be reference-locked, as outputs are directly derived from yig oscillators.