We are trying to improve our measurements of the components in this kit so we performed one port cal of the E8364B with data based 85054B OSL (with new part number broadband loads) and the male open does not quite (200% of spec) meet the phase model (as provided by Agilent cal lab data sheets showing the nominal model.)
When we check same components using 85032B kit (checked as conforming by Agilent cal labs) we get results near the nominal model, but we realize its about 1:1 for the comparison (1.9 degree uncertainty and 1.9 spec at 6 GHz test point.)
At the 6 GHz test point we estimate about a quarter degree phase uncertainty using the data based kit, and 3.5 degree measured error.
We are trying to determine whats the likely error: both 85032B opens off model; the uncertainty estimate off (phase at 6 Ghz is out 180 from the UUT open, or???
Ah...If I remember, the male open is very simple: just an outer conductor with a hole. No center pin extension.
Since you are using a high-frequency analyzer, I infer that you are using an adapter to make a type N port. What adapter are you using ?(in fact, I know, it is NOT a good adapter!, due to your results).
When using the 85054 kit, the male open has a test port extender that creates the proper length open male, even if the adapter center pin is not the correct depth. But the 85032 does not have an extender pin for the male standard (female test port), SO, you rely on the lenght and collet style of the adapter to give you the proper delay of the open.
When you cal with 85032B kit, you are just saying : bad length open is defined as good length open; 85054 doesn't make that assumption.
So, you must have a precision adapter with a set of fingers that matches the connector adapter of the 8753(for which the type N kit was defined) to have a proper test port for 85032B female test port to test a male standard.
Thanks for the troubleshooting idea. Yes we had two good north California adapters with no slots and replaced them with one southern California adapter with four slots. Minimal change. We also added another adapter to the test so we are testing three devices.
If we add double the difference between the open and short offset delay to the offset delay, using 18.223ps instead of 17.411ps for the open (17.817ps short) we seem to track the results we are seeing.
I'm inclined to use this changed model and its phase value as nominal model for checking our other cal kits with the PNA since three random opens from good kits agree.
Any idea if the E5071C N(f ) ports would better match the 8753 and the open model?
The type N cal kit is defined with a 6 slot adapter. Here is a picture of that shows different type N adapters. The upper ones are not acceptable. The lower one shows the 6 slots. It also likely has less recession than the first upper ones.
I believe the ENA uses the same Type N connector as the 8753.
The difference you see represents about 4.7 mils of offset (0.24 mm). Did you guage your connectors? What is the recession or procession (should always be recession!).
On the first pass "I'm using two Agilent adapters. One to go from the 2.4 mm connection on the PNA to 3.5 mm (model 85130F) and one to go from 3.5 mm to type N (model 85130C)." "The pin depth averages at -0.00014 inches and it is a slotless connector."
After the change of adapters. "The pin depth on the Maury adapter for the female type N connector averages at +0.0009 inches and it is a four-slot connector."
We will repeat the tests with an E5071C - it will be about ten days before we can arrange that.