We have two N5230C's, one being overseas, that we've been having trouble reproducing similar data on. I've had the remote site run an ECal confidence check (Both ECals are N4433A modules with SMA connectors) on the most recent calibrations being used and came back with some strange results.
I had them run through and send me the ECal confidence data (Data and Memory data) for S11, S22, S33, and S44. S11, S33, and S44's characterizations closely resemble the graph I've attached in "S11.png". However, S22 seemed rather shocking. This behavior was also consistent in the 5 calibrations I looked at.
Is it possible to say whether this is an issue with the ECal module or with the VNA? What factors would cause this kind of behavior? It seems to me like this is an issue with the ECal as this would be what I assume is a stored characterization.
Any explanation for this would be GREATLY appreciated.
It appears you have damaged port 2 circuitry. Most likely your RF path switch is damaged. Could be measurement receiver also, depending on what damaged it. Either ESD or too much RF power. This is a common failure to VNA's. It will require repair to correct the hardware.
The confidence data does look a bit odd. It has a ripple that indicates an electrically long device (which Ecal is not).
The other cross check it to measure a mechanical open/short or load on port 2 and report back the results. IF you see a simliar ripple then it is likley bad data in the ecal conf check. It might not be a bad ecal but a bad characerization.
Also, you say SMA; do you mean 3.5 mm? Are you using "connector-savers" on your ecal and applying a user characterization?
Yes, they're 3.5 mm. We are using the factory characterization with no connector-savers in this case. I got word earlier this week that our facility in China tried a different ECal unit (re-ran the calibration etc.) with the same result. They got a hold of their local Agilent support and found that Port 2 was in need of repair. They were supposed to have a loaner this week to see if this cleans up any other potential issues, but should have it on Monday.
Personally my first guess was a bad characterization, but if they verified it was the VNA, then I guess that's that.
I unfortunately don't have this particular ECal in my possession, so I've been having them send me data to attempt to troubleshoot.
After further investigation, it turns out it was in fact the ECal unit's characterization. It is following the port on the ECal unit (Port B) when checking the VNA pair connected to it. In addition to the one we caught in China, we're seeing similar effects on the ECal here in the states. Is this a relatively common occurrence?
Also, what kind of impact would this have on our data?
That's not typical. From the ripple I would infer a directivity error of about -32 dB; (See the +-1.5 dB ripple on the -15 dB return loss, or the +-0.5 dB ripple on the -6 dB return loss peak). This would likely make the effective source match a similar level, and its effect on the S21 would be a degradation of 0.05 dB to something like 0.11 dB from 0.06 dB normally.
From a quick check of the specs, the directivity should be on the order of 47 dB to 40dB ; the S21 tracking error on the order of 0.06 dB to 0.07 dB from 3 GHz up to 13 GHz.
If it has been a while since the last calibration, that can sometimes explain it. I've seen cases where connecting the ecal port 2 through a cable, the connector gets over torqued, and then when a wrench is used to losen the cable, it spins off the ecal connector instead. If it is then re-tightened, it can cause this type of error.
Also, we recently (maybe 1.5 years ago) improved the calibration methods in the factory for the Ecal in such a way that each port calibration quality -should- be identical. Older methods could cause different ports to have difference quality of calibrations, but that method hasn't been used for some time.
Dr_joel, many thanks for the insight. Is there any way we can investigate our Network Analyzer's actual performance? I've found the specs you're referring to, but I'm struggling finding documentation on how to check this.