Transmission measurements below 20 kHz seem to be off

[wickli...@gmail.com]

Dear all,

I am currently working with two VNAs (Keysight P5002A and P5004A) that exhibit the same behavior. Below, you’ll find a measurement of a 20 dB attenuator, normalized to the measurement value from METAS (red). My measurements, shown in black, display a discrepancy below 20 kHz.

This measurement was performed using SOLT calibration with a CalKit calibrated at METAS.  

I suspect this discrepancy may be due to an issue with the VNAs, and I’m not sure if any corrective measures are possible. However, I believe that measurement uncertainties below 20 kHz should be increased to account for this discrepancy. Could this approach be appropriate?

Has anyone else had the same problem?

Has anyone else encountered a similar issue?

For reference, I’ve attached my error terms, as I suspect they may be contributing to the problem.

I anticipated some potential issues with uThru, but not with SOLT calibration. Could anyone advise on how to either correct the measurement or adjust the uncertainties, if needed?

Thank you for your insights.

[Juerg Ruefenacht]

Hi Patric,

we see similar issues with our low frequency VNAs when we measure higher attenuation values at the low frequency end. This is due to common mode effects and can be avoided by placing ferrite cores around the test port cables - towards the VNA test port ends. At METAS we use the following ferrite cores for this purpose: Nanocrystalline cores (HITACHI, FINEMET, FT-3KM F6045G).

Since your laboratory is located very close to METAS, you may borrow a set of these cores and test whether you can get rid of the observed problem.

Best Regards,
Juerg

[Michael Wollensack METAS]

And a picture of our setup with the ferrite cores:

Regards

Michael

[wickli...@gmail.com]

Hi everyone,

I finally had the opportunity to test our ferrites. Unfortunately, they did not produce any noticeable improvement regarding the issue I'm facing.

I've attached an image showing how the ferrites were positioned during testing. These are not from Hitachi; I used "Proterial FINEMET, FT-3kM F6045G," as they were readily available and, to my knowledge, equivalent to the specified product. For this initial test, the ferrites were placed manually to observe any immediate effects, with plans to 3D print a fixture if results were promising. However, I was surprised to see no improvement at all.

Could you provide any suggestions or alternative approaches to resolve this problem? Your insights would be greatly appreciated.

Best regards,
Patric

[wickli...@gmail.com]

Here’s a brief update:
I’ve noticed that this behavior only occurred during the recent measurements. Specifically, I conducted some detailed measurements with a signal level of 0 dBm, whereas the usual level is -20 dBm. To confirm whether this behavior is reproducible at -20 dBm, I repeated the measurements at that level and found that the effect was significantly less pronounced.

This leads me to suspect a saturation effect at 0 dBm during calibration rather than a common mode effect.

For now, due to time constraints, I’ll leave this finding as is. If I gather any further insights, I’ll make sure to share them.

Best regards,
Patric

[Michael Wollensack METAS]

Hi Patric,

thank you for reporting back. 

In that case I think i's a combination of both effects, compression and common mode.

Regards

Michael