Hello,

We have several 8753D network analyzers and several cal kits, an HP 85032B and an HP 85032F.

We just recently leased the 85032F.

Since this calibration kit is not listed in the call kit menu in the 8753, I have loaded the standards from the disk that came with the kit. However, when I do this, the user call kit now expects an offset load when there is none included in the kit. I have tried downloading the call standard file directly from Agilent--this produces the same results.

I can modify the call kit to use the broadband load. This appears to work but I am not confident that this is the correct approach.

Could you please clarify this? Specifically, after loading the call kit file with a "restore state" command, should I modify the call kit definitions in order to eliminate the offset load?

Thank you very much for your help.

We have several 8753D network analyzers and several cal kits, an HP 85032B and an HP 85032F.

We just recently leased the 85032F.

Since this calibration kit is not listed in the call kit menu in the 8753, I have loaded the standards from the disk that came with the kit. However, when I do this, the user call kit now expects an offset load when there is none included in the kit. I have tried downloading the call standard file directly from Agilent--this produces the same results.

I can modify the call kit to use the broadband load. This appears to work but I am not confident that this is the correct approach.

Could you please clarify this? Specifically, after loading the call kit file with a "restore state" command, should I modify the call kit definitions in order to eliminate the offset load?

Thank you very much for your help.

You have a call logged in our technical support area as well. This call is under case #3400071 and is in your rental contacts name. The 85032F should have no references to 'offset load'. Please call me at 800 829-4444, 9, 400071, # and we can step thru the kit definitions as loaded in your 8753D and either ensure they are correct or modify them to be correct. I have been informed that some of the cal kit downloads on our web site MAY be pointing to the wrong files. Not sure if this is true or not. But with some quick checks we can offer the assurances you need to proceed forward with your 8753D + 85032F Type-N kit.

Tom.

If possible, I was hoping to ask one more related question.

The manual (page A 11) supplied with the 85032F calibration kit specifies two different sets of calibration model parameters, one for use with the 8753 and one for use with the PNA series analyzers.

For example, for use with the 8753 the model parameters for the short are specified to be:

L0=3.3998 x10^-12 H

L1=-496.4808 x10^24 H/Hz

L2=34.8314 x10^-33 H/Hz^2

L3=-0.7847 x10^-42 H/Hz^3

Offset delay=45.955 ps

However, for the PNA series these are specified to be

L0=0 H

L1=0 H/Hz

L2=0 H/Hz^2

L3=0 H/Hz^3

Offset delay=45.955 ps

That is, all of the coefficients in the power series expression for the inductance of the short are set to zero in the model for the PNA analyzers. The offset delay is the same for both sets of data.

Of course there is only one physical cal kit. From the application note 8510 5B, one can see that there is a uniform definition for the calibration standards which uses both the power series coefficients as well as the offset delay. That is, it appears that there is really only one model in which the parameters are used.

Thus, it seems odd that the power series coefficients for the short would be set to zero when using the 8753 but take on non-zero values when using a PNA series analyzer.

Joe Lubecky from Metric Test surmised that this might be related to the different frequency ranges of the two VNAs. That is, the parameterized model for the calibration standards, in particular the model for the short, is more accurate over the 8753's frequency range of 30 kHz to 6 GHz with the power series coefficients set to zero, but when considering the much greater frequency range of the PNA is more accurate with the power series coefficients set as above.

If I compute the impedance of the short ( the parasitic inductance computed from the power series) at 6 GHz it is only .056745j Ohms.

If this is true, it would seem to me that the PNA using the parameterized model given for it would suffer some accuracy loss below 6 GHz.

Can you tell me if this is true?

Thanks again for all of your help.

We are very much concerned with accuracy of these measurements--that is why I am concerned with verifying that the calibration is correct.

I think the more proper way to think of this is that 8753 has some error in the model of the short due to lacking a more complex model. This was a decision that was made back in 1982, when the 8753 only had 3 GHz capability, and the model of the short provided essentially no benefit. The 8510 had always provided the full model of the short with inductance, but as you noted, it doesn't have much effect until higher frequencies.

Since the PNA was intended for higher frequency, it used the more complex model.

Actually, you have it backwards. The 8753 didn't include the inductance terms for the short model. Usually, the coefficients for the 8753 had a slight change to the offset delay to provide a better fit to the polynomial coefficients. At low frequencies, the differences are small enough to not cause worry.

The calibration kits from Agilent come with a generic model for a family of devices. Due to machining tolerances there are variations in the actual devices from the generic model. These variations contribute to the residual errors after calibration. At low frequencies, the variations due to part dimensional tolerances contribute more than having or not having the inductance terms. I am a little surprised that the manual indicates that the delays are equal because several years ago I created a set of delay terms that would provide a slightly better match for the 8753 which did not include inductance terms to the model that included both delay and inductance terms. It has been so long ago that I don't remember which version of 8753 firmware picked up the new delay terms. At the time I remember thinking that the changes did not really affect things very much.