HP8714ET phase problem?

[mrf245]

Hello all
I have HP8714ET and I found a problem these days.
For example, I connect calibration kits directly to RF-OUT port, meas1 set to reflection, freq is 300KHz to 3GHz (full span), after finished open-short-load calibration, leave RF-OUT port open or short, I can see a long line along with smith chat edge, but load is a point, shows below:
Its different from used HP8751 before,after calibration, open short and load are all a point in full span,not line.Anyone
Anyone know how to fix this problem?Thanks

[hendorog]

Here is a reason for this behaviour:
http://www.vnahelp.com/tip7.html

The definition of the cal kit inside the VNA is what 'causes' this. These definitions in the VNA need to match the behaviour of the cal kit you are using for best accuracy.

If you setup a user cal kit in the VNA with perfect opens and shorts then you will see the dot instead of the line.
Of course then the measurements you make will not be as accurate, as the opens and shorts are not perfect in reality.

Presumably the 8751 used such a definition when you calibrated, hence the difference between the two instruments using the same cal kit.

[KE5FX]

As Hendorog says you shouldn't normally see a dot in the open case.  It's possible to build a short standard that will be more-or-less perfect at reasonable frequencies, but the open will always have some fringing capacitance, and the calkit definition will need to account for it. 

As a result, when your calibration is done properly, you will see an arc on the right side of the Smith chart when you re-measure your open standard.  You may or may not see a dot on the left side when you measure your short, but if you do see an arc, it will generally be much shorter than the open.

[G0HZU]

That's a bigger arc than normal if you are trying to measure an open or short at the reference plane. It's probably equivalent to a delay line of about 18 or 19ps. Therefore, are you leaving the cal kit 'OPEN' connector in place when you make this measurement? Don't forget there will be a correction shift in the cal kit to allow for the extra delay in the physical open connector wrt the desired reference plane location. So if you leave the OPEN std from the cal kit in place you can expect to see a bigger arc because you are measuring the effect of this 'delayed open'. In other words you are really measuring an open at the end of a very short delay line when you measure your cal kit open. So this arc is what you would expect to see if you measure the OPEN connector in your cal kit after a calibration. Is the OPEN in your cal kit approx an open circuit 18-19ps delay line? Is your SHORT approx a 18ps shorted delay line? Does your cal kit file correct for this with entries of about 18ps delay for the short and open connectors in your cal kit?

A typical SMA cal kit file might have a delay of about 30ps for the short and the open. Yours looks a bit shorter than this. Is it an N type cal kit?

[G0HZU]

The other thing to note is that your OPEN and SHORT plots show quite a bit of negative resistance which looks very suspicious. Something not right here either? If a user were to measure the OPEN standard after a calibration they would expect to see the equivalent of a slightly lossy 50 ohm delay line with an open at the end of it. The resistive part of the impedance would be less than 0.1 ohm even up at 3GHz. The amount of delay in the line would be in the order of 20-30ps but this depends on the cal kit used. But the resistive part should be very small. Less than 0.1 ohm at 3GHz but you are getting (negative) several ohms here.

Maybe you should show what you are using as a calibration kit and how you are setting up the calibration in the 8714ET menu system.

[mrf245]

Thanks for hendorog, KE5FX, G0HZU answered my question.
Today I found a menu in 8714ET cal page, "Normalize",if turn it ON before use my cal kit do calibration, cal results in open and short are all great.Shows below.
I checked 8714 user guide, this "normalize" means remove measurement errors in 8714 software. Can I use this method to make better measurement than before?
Also show my cal kits below, not sure if it can working up to 3GHz.These are N-type male connector, I also selected "N-type cal kits" in 8714 cal menu. When doing cal, I connect kits directly to VNA port.

[hendorog]

Had a quick look at the manual for the 8714 and I don't think normalise is doing what you expect:

"Normalization is the simplest type of calibration. The analyzer stores
normalized data into memory and divides subsequent measurements by the
stored data to remove unwanted frequency response errors. "

It's just storing the current trace and using it as a baseline for subsequent measurements. So then measuring the same cal standard you normalised on will display 0 dB and 0 degree phase.