AnsweredAssumed Answered

Strange results when measuring S11 and S21 of an Airline

Question asked by theamberco on Aug 12, 2014
Latest reply on Aug 22, 2014 by Dr_joel

Like • Show 0 Likes0
Comment • 10

Hi, I have a PNA E8357A and I am using a 85032A cal. kit and a 85055A ver. kit. I did a one port cal and measured the S11 of an airline terminated in a short, I got what is in pictures 2 and 3 then I did a two port cal. and masured S21 I got what is in picture 1. Anyone can tell me what this means and what is wrong with that.
Thanks

Edited by: theamberco on Aug 22, 2014 5:22 PM

Attachments

terminate in short.png8.4 KB
s21 with 50 ohm airline after two ports cal.png8.1 KB

No one else has this question

Outcomes

10 Replies

Dr_joel Aug 13, 2014 8:48 AM
Mark Correct
Correct Answer
You don't say what is the difference in conditions between short 1 and short 2 (maybe two different short standards). You don't say your test setup but i presume it has a test port cable.

The fine grain ripple is due to a bad cable. From the frequency of the fine-grain ripple you can compute your cable length. The slow ripple is due to cal-kit error. The ripple of +-0.3 dB means you have a combination of source match and directivity error of -30 dB. It could be that each is about the same, and each is about -36 dB.

I think the 85032A cal kit is the older one and it looks to be specified to only 2 GHz. Thus it might have a low-band load which is tuned to 2 GHz and has poorer performance at higher frequencies. It might be a case of "you get what you pay for, not what you hope for".
Like • Show 0 Likes0
Actions
- theamberco @ Dr_joel on Aug 13, 2014 10:18 AM
  Mark Correct
  Correct Answer
  That's Joel, I made a mistake I was using the 85032B. With respect to the port setup I was using two APC7 cables Suhner sucoflex 100 with two Agilent APC7 to N adapters.
  The difference between the pictures were taken at different times with the same short standard.
  So those results are telling me that with those cables I wont be able to verify the calibration?. I did calibrate at the end of those cables, why a I getting amplification?
  
  Edited by: theamberco on Aug 13, 2014 11:12 AM
  
  Edited by: theamberco on Aug 13, 2014 11:16 AM
  Like • Show 0 Likes0
  Actions
  - theamberco @ theamberco on Aug 14, 2014 7:57 PM
    Mark Correct
    Correct Answer
    I did check just the airline connected to the port 1 terminated in a short with no cables and this is what I got in the first picture.
    Also tried the airline on port 2 and calibrated with different standards and cables so I got the second picture.
    
    Edited by: theamberco on Aug 14, 2014 8:55 PM
    Attachments
    port 2 s22 different cal standards...png120.7 KB
    airline to port one no cables.png8.3 KB
    Like • Show 0 Likes0
    Actions
    - Dr_joel @ theamberco on Aug 15, 2014 9:26 AM
      Mark Correct
      Correct Answer
      So you see the high speed ripple, which is due to the bad cable, is gone.
      
      The low speed ripple is almost certainly due to the load, which is likely only specified and tuned to 2 GHz. The ripple up to 2 GHz is 0.08 which represents a total error of -40 dB residual. If the load were -46 dB, it would give this error (since the error in the load goes to both the directivity and the source match, and measureing the short at the end of a line gives the worst case addition of these errors at the peak). And there can be errors in the Airline as well as not proper handling (you have to be careful how to set the gap, and it's affect).
      
      Further, these kits have to be handled carefully to avoid connector repeatability errors, and finally, the two ports themselves could have different characteristics, so I think this is what you get with these old-fashioned mechanical kits. That's why everyone uses Ecals today!
      
      These results are not unexpected, according to my understanding of the cal kit. you will need a higher frequency calkit to get better results at high freqeuncy.
      Like • Show 0 Likes0
      Actions
      - theamberco @ Dr_joel on Aug 15, 2014 11:36 AM
        Mark Correct
        Correct Answer
        Thanks, but the kit used is the 85032B goes to 6Ghz but you are saying that Ecal is better than mechanical kits? I will try to fetch one Ecal or another higher frequency kit then and try.
        Like • Show 0 Likes0
        Actions
        theamberco @ theamberco on Aug 21, 2014 7:55 PM
        Mark Correct
        Correct Answer
        I tried calibrating with an 85054B at the port terminating the airline in a short and got this.
        Attachments
        85054B with airline.png8.2 KB
        Like • Show 0 Likes0
        Actions
        Dr_joel @ theamberco on Aug 22, 2014 12:26 PM
        Mark Correct
        Correct Answer
        That 0.06 dB p-p ripple represents a total residual of about -49 dB, inferring a load quality of about -55 dB. So exactly what one would expect. Good loads give better results. The high speed ripple looks like trace noise. Raise the power or lower the IF BW to remove that bit of 0.05 dB noisy error. I recommend 1 kHz IFBW. Also, do you use torque wrenches? They will be needed to get high quality measurements than this. And you will likely need to clean your connectors as well. A little bit of dirt can change the mating distance and limit the quality. But improving beyond 0.06 dB for reflection measurements of shorts means you are in the realm of metrology. In this case, you also have to know that the ripple at the low end of the frequency is a little expected as the airline impedance is not constant with frequency, but increases at lower frequency due to skin effect of not infinite conductivity. And depending upon the length of the airline, you have to deal with the fact that the center conductor is not exactly centered, but droops in the middle causing a change in impedance of the line. This is all accounted for in metrology labs (it forms a catenary curve in fact, just like cables on the golden gate bridge).
        Like • Show 0 Likes0
        Actions
        theamberco @ Dr_joel on Aug 22, 2014 4:00 PM
        Mark Correct
        Correct Answer
        Thanks Joel you are a jewel. I am now checking your book about determining residual directivity with an airline and a load.
        I did clean the surfaces and airline warmed the unit longer time even used gloves to handle the airline (I looked like a metrology hit man) and used a torque wrench but I find no gaging difference in using torque wrench and tightening hard by hand on N type connectors. Also used the 1Khz IF (big difference).
        So now I got this.
        Attachments
        airline with short at port1.png7.5 KB
        Like • Show 0 Likes0
        Actions
        Dr_joel @ theamberco on Aug 22, 2014 10:25 PM
        Mark Correct
        Correct Answer
        The big difference is IF BW and sliding load. I noticed a discrete jump in the trace near 2 GHz, and I presume that is due to you using a sliding load for the calibration. The sliding load is much better than the fixed load, so above 2 GHz you see less ripple because the sliding load sets the directivity and source match.
        
        BTW, the first posting showed the picture. I don't know how to delete repeated postings either.
        Like • Show 0 Likes0
        Actions
        theamberco @ Dr_joel on Aug 22, 2014 4:11 PM
        Mark Correct
        Correct Answer
        Ok
        Attachments
        airline with short at port 1 at 1Khz IF.jpg75.4 KB
        Like • Show 0 Likes0
        Actions

Strange results when measuring S11 and S21 of an Airline

Attachments

Outcomes

Attachments

Attachments

Attachments

Attachments

Related Content

Recommended Content