Closed nkrackow closed 8 months ago
I believe SMA shield is just shorted to FP in a defined assembly variants:
yes, but the impedance of short-circuit may be non-negligible for 1GHz and above
It does have to do with the panel. Slightly bending the PCB away from it with a finger significantly changes the crosstalk.
We will try to re-create and investigate the problem
To do this, we need more info about the setup. Preferably some photos depict how cables are placed.
@nkrackow sent a photo to @akaminska AFAIK
Yes, I put them here again now. I don't think we've attached anything in a strange way. Simply the output of a Mirny to a spectrum analyzer. We've seen the most crosstalk between roughly 1-2 GHz, above and below it was better. Pressing on the PCB close to the front panel (area central in the first picture) has quite an effect on the crosstalk.
we plan new revision. What if we remove the assembly variant and connect the SMA connectors to solid GND?
added direct connection like in Urukul
The outputs of current Mirnies V1.1 suffer from significant crosstalk:![PNG1](https://user-images.githubusercontent.com/17088194/208652565-3d9ec3f0-719e-4621-88f7-bebf6d5884a2.png)
Pushing on the PCB close to the front panel makes the crosstalk vary by more than 10 dB. The SMA shield is shorted to the front panel at DC.
This implies that there is a problem with the front panel, probably the SMA pins making connection. The same issue appears on other high frequency sinara modules: https://github.com/sinara-hw/Urukul/issues/68
@gkasprow @akaminska