Assessing VNA linearity in VNA Tools

[Martin Salter]

Hello,

 

Is it true that the a step attenuator to be used for assessing VNA linearity in VNA Tools needs to be characterised in terms of both magnitude and phase of S21? We would normally have our step attenuator calibrated against attenuation standards in termas of magnitude (dB) only and not phase. Thanks.

 

Regards,

Martin

 

[Michael Wollensack METAS]

Hi,

You just need to know the magnitude of your step attenuator. The phase error of the VNA linearity is a function of the magnitude error:

1. Both of the IQ receivers have a magnitude error m in db.
2. a in Np = ln(10)/20*m in dB
3. b in RAD = a in Np
4. p in deg = 180/pi*b in RAD
5. p in deg = ln(10)/20*180/pi*m in dB
   
6. p in deg = 6.6*m in dB

Regards

Michael

[Cecile Bachmann]

Hi Michael,

I also have a question about the linearity.

I measured the linearity following the Euramet Guide.

I calibrated at -10dBm and measured with attenuations between 10 and 70 dBm, in 10dBm steps.

I calibrated again at 13 dBm and did the same, to get the positive levels.

How do I now fill out the table in the VNA Tools Database? Why goes the standard entry down to -340 dB? Euramet says it makes no sense to characterize the linearity below -70dBm.  The unit listed is dB and not dBm, so what is the reference level?

For example if I calibrate at 13 dBm, all lower levers are 0.3dB wrong at 50MHz. So it looks like the calibration at 13dBm is 0.3 dB wrong compared to all others. At which power lever do I put this entry in the linearity table?

Thanks and regards

Cecile

[Michael Wollensack METAS]

Hi Cecile,

> Why goes the standard entry down to -340 dB? Euramet says it makes no sense to characterize the linearity below -70dBm

I goes down to -340 dB because that's about the precision of a double.

E.g: 10^(-340/20) = 1e-17 and 10^(-340/20) + 1 -1 = 0

And I want to assign as well a linearity uncertainty if the value is very small (close to 0).

So just extend the -50 dB, -60 dB or -70 dB value down to -340 dB

> The unit listed is dB and not dBm, so what is the reference level?

The reference level in dBm is VNA Device / Settings / Spec Power (dBm)

So for your case set the Spec Power to -10 dBm and enter the observed linearity error and some safety margin (for example 0.01 dB) to the passive side (-340 dB to 0 dB in the linearity table). 

Then enter the observed linearity error (0.3 dB) and as well some safety margin (0.1 dB)  for the active side (0 dB to 20 dB or 23 dB in your case) .

So your linearity table should look like that:

Frequency	Linearity	Linearity	Linearity	Linearity	Linearity	Linearity	Linearity	Linearity
	@ -340 dB	@ -340 dB	@ -100 dB	@ -100 dB	@ 0 dB	@ 0 dB	@ 23 dB	@ 23 dB
(Hz)	Mag (dB) (k = 2)	Phase (°) (k = 2)	Mag (dB) (k = 2)	Phase (°) (k = 2)	Mag (dB) (k = 2)	Phase (°) (k = 2)	Mag (dB) (k = 2)	Phase (°) (k = 2)
0.00E+00	0.02	0.133	0.02	0.133	0.02	0.133	0.4	2.66
1.80E+10	0.02	0.133	0.02	0.133	0.02	0.133	0.4	2.66

and the Spec Power is set to -10 dBm.

Hope this helps.

Regards

Michael

[Cecile Bachmann]

Hi Michael,

thanks. So for the active side it is probably the best idea to calibrate at 10dBm (for the ZNB) and measure a 20 dB attenuation, to get the comparison to -10dBm?

For the ZVRE I can not calibrate higher levels than 0dBm, as this is the maximum output of the VNA Port. Is it correct to assume, that at 23dBm I will have the same error as at 0dBm?

Thanks and regards

Cecile

[Michael Wollensack METAS]

Hi Cecile,

For a ZNB I would measure the passive side at -10 dBm and for the active side you can measure at +10dBm and the measure a 20 dB attenuation. Then you can specify the linearity up to +20 dB with a reference level of -10 dBm.

For the ZVRE I would measure the passive side as well at -10 dBm and the active side at 0 dBm and a 20 dB attenuation. Then you can specify the linearity up to +10 dB with a reference level of -10 dBm.

Regards

Michael 

[cbac...@emc-testcenter.com]

Hi,
as described for the ZVRE we have specified our linearity up to +10 dB. If we now calibrate it at 0 dBm, VNA Tools says that the contribution of the the linearity is not specified, so it can not calculate the Measurement Uncetainty.
Why is this the case?
thanks and regards
Cecile

[Michael Wollensack METAS]

Hi Cecile,

there a points in your uncertainty budget which are larger than +10 dB. So you need to specify the linearity up to +20 dB.

What is your source port power and what is the S21 of your DUT?

Which version of VNA do you use?

Regards

Michael

[cbac...@emc-testcenter.com]

Hi Michael,
We use a ZVRE at source port power of 0dBm. The source port power can not be set higher than 0 dBm.

The DUT has a quite high Attenuation, see Attachment. That's the reason, why we want to measure at a higher source power. When we measure at -10 dBm as usual, the measurement gets quite noisy.

How would it be possible to specify the linearity up to +20dB?

Thanks and Regards

Cecile

[Michael Wollensack METAS]

Hi,

then your 10.58 dB to 11.34 dB values in your uncertainty budget from the post before come from the thru measurement during calibration. I guess you set the Spec Power in the vnadev File to -10 dBm and you specified your linearity up to +10 dB.

If you measure now with 0 dBm source power and you measure a thru (S21) or a high reflect (S11) you will get uncertainty contributions around +10 dB. The linearity specification in the vnadev-File in dB is relative to the Spec Power in dBm.

So you need to specify the linearity up to +20 dB.

Regards

Michael

[cbac...@emc-testcenter.com]

Hi Michael,
so this means my uncertainty budget is only valid for the given Spec Power and also the IF Bandwidth? So if I measure at a different source power I should specify a separate vnadev File, with a different noise, drift and linearity measurement?

Thanks and regards
Cecile

[Michael Wollensack METAS]

Hi,

>  so this means my uncertainty budget is only valid for the given Spec Power and also the IF Bandwidth?

No. You can have different Source Power settings in the vnadev-File and in the measurement. VNA Tools will take care about it.

E.g.:

vnadev-File at -10 dBm Spec Power.

S11 measurement of a short at -10 dBm source power: the linearity uncertainty comes from the 0 dB linearity uncertainty.

S11 measurement of a short at 0 dBm source power: the linearity uncertainty comes from the 10 dB linearity uncertainty.

You can have as well different IFBW settings. VNA Tools will scale the noise uncertainty.

Regards

Michael

[cbac...@emc-testcenter.com]

Hi,
ok, thanks. But how can I specify the linearity up to +20dB? How can I get the values?

Thanks and regards
Cecile