AnsweredAssumed Answered

Difference in TDR Plot of cable assemly using PNA vs. ENA

Question asked by dweinstein on Feb 12, 2016
Latest reply on Feb 16, 2016 by dweinstein

Like • Show 0 Likes0
Comment • 4

Recently a customer returned a cable assembly (15 ft long) claiming the impedance of the cable was 53 Ohms. We tested using the N5225A PNA setting it up for a Low Pass TDR calibration from 10 MHz to 8.01 GHz with 801 points using an 85052D cal kit. See the attached plot. The customer uses a E5071C ENA and we have one also so we set it up the same way and got a significant difference in the plot.

The main difference is that on the PNA, at time less than t=0, the plot is at -10 mU and on the ENA, it is +10mU giving a 2 Ohm difference for the cable. Without the cable attached or while measuring a short cable (6 inches), the reflection at less than t=0 is 0 (no offset).

Why is there a reference offset below t=0 and why are the 2 plots not giving the same result? What is the actual impedance of the cable?

Thanks,

David

Attachments

CABLE TDR E5071C.PNG25.5 KB
Cable TDR N5225A.png45.8 KB

No one else has this question

Outcomes

drkirkby Feb 12, 2016 9:20 AM

It's probably worth attaching the Touchstone files. The someone else can analyze them.

Dave

Actions

dweinstein @ drkirkby on Feb 12, 2016 10:03 AM

I no longer have the cable assembles and I did not save the data, just the plots.

Actions

Dr_joel

Feb 15, 2016 9:10 PM

The offset at T=0 is due to several factors, the main one being the load that is used to calibrate the system, and the point spacings used. Because a DC value is not measured, it is inferred from the low frequency RF data. The offset of +-1 ohm represents a return loss error on the order of about -40 dB. You can get this offset also for complicated reasons due to the termination impedance at the far end of the cable, and how the time domain transform (which must be repetitive) causes an alias of the data.

One way to avoid this is to use a short length of well defined impedance line, such as an airline, before the cable to set the reference independent of the calibration or the reference load. The offset from zero line is a kind-of artifact of the math and settings. If the cable is very long, you really need to have enough points to allow a long enough span to go to the end of the cable to get a reasonable result. for a 15 meter cable, with maybe 20 nsec of delay, times 2 for reflection, means point spacings on the order of 10 MHz to get good results.

Actions

dweinstein @ Dr_joel on Feb 16, 2016 5:29 AM

Joel,

Thank you for the very thorough explanation. We do have airlines we can use as a reference and will try the measurement again using it. We will also ensure that we have adequate number of points to get the spacing required based on your guidelines.

David

Actions

4 Replies

drkirkby Feb 12, 2016 9:20 AM
Mark Correct
Correct Answer
It's probably worth attaching the Touchstone files. The someone else can analyze them.

Dave
Like • Show 0 Likes0
Actions
- dweinstein @ drkirkby on Feb 12, 2016 10:03 AM
  Mark Correct
  Correct Answer
  I no longer have the cable assembles and I did not save the data, just the plots.
  Like • Show 0 Likes0
  Actions
Dr_joel Feb 15, 2016 9:10 PM
Mark Correct
Correct Answer
The offset at T=0 is due to several factors, the main one being the load that is used to calibrate the system, and the point spacings used. Because a DC value is not measured, it is inferred from the low frequency RF data. The offset of +-1 ohm represents a return loss error on the order of about -40 dB. You can get this offset also for complicated reasons due to the termination impedance at the far end of the cable, and how the time domain transform (which must be repetitive) causes an alias of the data.

One way to avoid this is to use a short length of well defined impedance line, such as an airline, before the cable to set the reference independent of the calibration or the reference load. The offset from zero line is a kind-of artifact of the math and settings. If the cable is very long, you really need to have enough points to allow a long enough span to go to the end of the cable to get a reasonable result. for a 15 meter cable, with maybe 20 nsec of delay, times 2 for reflection, means point spacings on the order of 10 MHz to get good results.
Like • Show 0 Likes0
Actions
- dweinstein @ Dr_joel on Feb 16, 2016 5:29 AM
  Mark Correct
  Correct Answer
  Joel,
  
  Thank you for the very thorough explanation. We do have airlines we can use as a reference and will try the measurement again using it. We will also ensure that we have adequate number of points to get the spacing required based on your guidelines.
  
  David
  Like • Show 0 Likes0
  Actions

Difference in TDR Plot of cable assemly using PNA vs. ENA

Attachments

Outcomes

Related Content

Recommended Content