OSL calibration kit quality question

[antenna]

After seeing a video comparing the quality of various calibration loads, I am wondering what mine, which says it is good up to 4.5GHz, looks like compared to a much better load.  The one I have is the Siglent F503ME kit, but I have nothing better to compare it to.  Has anyone calibrated with a better load and then looked at the F503ME?  I don't have a VNA that goes that high, I am just curious how accurate it is near the top end of what my equipment can measure, which is 3.2GHz.

Thanks!

[Andrey_irk]

In the F503ME's datasheet they say that the VSWR of the load is <=1.02, which is about -40dB of Return loss. It is pretty decent, especially considering the fact that the repeatability of precision coaxial connectors and cables is not that far off from this number.
Can't tell you from the top of my head, but I saw cal kits with loads that have RL of -50dB in your frequency range, but these are very expensive and fragile. A better load than this one can potentially give a better residual directivity for your VNA, but you have to look closely to find whether it is going to make any sense considering the drift of your VNA, repeatability of your connectors and the VSWR you are going to measure.
There is a nice table from Anritsu where you can find the uncertainty of your RL measurement from the difference between your directivity (which you can just take equal to the RL of the load) and the expected RL of the DUT. Find it, play around with it and see for yourself if that load is good enough for you. Personally, I would be pretty happy with it.

[Joel_Dunsmore]

-40 dB load will give you almost exactly 1 dB ripple (error) on a -22 dB return loss measurement.  Good Keysight loads are on the order 1.005 SWR (about -52 dB RL) up to a few GHz but the hard spec is something like -46 dB RL (in fact, they have to pass -52 dB to guarantee they are better than -42, at least back in the day).  Above a couple of GHz, TRL standards are actually better for Return Loss than using fixed loads.