AnsweredAssumed Answered

N8975A input loss correction table not working as expected

Question asked by malibu on Apr 25, 2013
Latest reply on Jul 31, 2013 by malibu

Like • Show 0 Likes0
Comment • 6

The input loss correction functions, Fixed, Table, Temp are not giving me expected results. I am testing a 16Ghz 16db gain amplifier with expected noise figure of 1.7db. I am using an input loss table to correct for cable and probe losses. The input loss at 16Ghz is 1.2db. The output loss table entry is about 0.5db. The cable connecting the output probe to the NFA was included in the user Calibration. The meter measures gain of 16db which matched my s-parameter measurement. The noise figure comes out at 2.3db. A figure that is 1/2 of the input loss above what is expected. If I turn off the input loss correction the noise figure increases by 0.6db to 2.9db and the gain decreases by 1.2db. The gain decrease is as expected but the noise should have gone up! by 1.2db NOT 0.6db! I tried another self consistency check and put in a 3db pad between the noise source and my amp. Leaving the input loss compensation table on. I see the Noise Figure go up by 3db to 5.3db and the gain go down my 3db as expected. Now I add 3db to the values in my input loss compensation table.

The gain is correctly computed to be 16db but the noise is 3.8db... only 1.5db is removed from the computed noise figure!

Why is the noise figure only being corrected by 1/2 the input loss that I put in to the loss compensation system?

If I use the old school method of correcting for additional loss by reducing the ENR value by 1.2db I get both gain and noise figure numbers I expect!

Edited by: malibu on Apr 25, 2013 10:55 PM

No one else has this question

Outcomes

malibu Apr 26, 2013 10:08 AM

I'm such an idiot at times....... I did not realize that the Temp field in the correction setup menu modified the effect of the Table and / or the fixed attenuation's noise contribution I thought it was a third mutually exclusive menu selection... It was so obvious that NO where in the documentation is that made clear!
Thanks any way ... but it is nice to have my confidence in the NFA back!

Actions

rohit @ malibu on Jul 12, 2013 2:30 AM

Hi maibu,

I have Agilent N8975A NFA and I am using it measure Noise Figure of a DUT. when I directly connect the DUT with the NFA (Fixed cable with very small loss), I get Noise Figure well below 3.5dB (within the spec of DUT) but when the same NFA, I connect to DUT in an ATE in which the total path loss of the (cables+switches) connecting the (NFA out) to (DUT in) and (DUT out) to (NFA in) is 7.7 dB, I get the Noise figure around 7.7 dB.
Now I used Loss Comp in NFA. and entered the Forward and Reverse Loos resp as 4.1 dB and 3.6 dB, my NFA shows Noise Figure around 5.5 dB.Then I also entered the Temp settings as 298k (25 c) which is abmient temperature after doing this and guess what I get far better readings and close to what is expected. But I really don't know why and how much temp should I give.Any help will be highly appreciated. :)

Actions

malibu @ rohit on Jul 12, 2013 10:38 AM

Hi,
The problem is that the NFA's default temperature setting for the input / output loss is 0degrees K.
It's purpose is to correctly account for the temperature of the loss element, which may be different than the temperature of the noise source. For example when measuring on wafer noise over temperature the chamber with the probes/cables is cold say -50 C. The noise source is outside the chamber and at room temp ~296 ish K. The loss inside the chamber generate less noise at -50 and this field accounts for the difference. I fixed the bug in my ATE code by updating the temp field at test time to reflect the actual temperature of the loss elements between the DUT and the NFA receiver.
Hope that helps.
.chuck

Actions

rohit @ malibu on Jul 14, 2013 11:43 PM

Hi,

Thanks for the response. :)
Find the attached setup schematic. As you can see only the DUT will be inside the temp. chamber and the instrument as well as the switching matrix both will be outside chamber, so do you mean that I should put ambient (room) temp before DUT and chamber temp after DUT in the NFA programatically.

Attachments

NFA Setup.png19.2 KB

Actions

rohit @ rohit on Jul 31, 2013 3:10 AM

Hello Malibu,

I posted the problem which I am facing with NFA some 2 weeks before, It will great if you could spare some time and guide me some solution/idea.
One observation which I made was, after giving the temp and loss values before and after the DUT, the readings are in specs of my DUT for 1.31 - 1.39 GHz but the values goes beyond specs for 1.3 and 1.4 GHz. Any idea why this could happen??

Regards,
Rohit.

Actions

malibu @ rohit on Jul 31, 2013 7:17 PM

Looks to me like all the loss of your system is outside of the temp chamber.
Therefor the temp setting of both loss elements should be ambient. not the temp inside the chamber.
how much does the loss vary with frequency over the band you are measuring?
if it varies by more than a few percent of the expected measurement you should be using a table that accounts for
the frequency changes of the loss.
hope this helps.

Actions

6 Replies

malibu Apr 26, 2013 10:08 AM
Mark Correct
Correct Answer
I'm such an idiot at times....... I did not realize that the Temp field in the correction setup menu modified the effect of the Table and / or the fixed attenuation's noise contribution I thought it was a third mutually exclusive menu selection... It was so obvious that NO where in the documentation is that made clear!
Thanks any way ... but it is nice to have my confidence in the NFA back!
Like • Show 0 Likes0
Actions
- rohit @ malibu on Jul 12, 2013 2:30 AM
  Mark Correct
  Correct Answer
  Hi maibu,
  
  I have Agilent N8975A NFA and I am using it measure Noise Figure of a DUT. when I directly connect the DUT with the NFA (Fixed cable with very small loss), I get Noise Figure well below 3.5dB (within the spec of DUT) but when the same NFA, I connect to DUT in an ATE in which the total path loss of the (cables+switches) connecting the (NFA out) to (DUT in) and (DUT out) to (NFA in) is 7.7 dB, I get the Noise figure around 7.7 dB.
  Now I used Loss Comp in NFA. and entered the Forward and Reverse Loos resp as 4.1 dB and 3.6 dB, my NFA shows Noise Figure around 5.5 dB.Then I also entered the Temp settings as 298k (25 c) which is abmient temperature after doing this and guess what I get far better readings and close to what is expected. But I really don't know why and how much temp should I give.Any help will be highly appreciated. :)
  Like • Show 0 Likes0
  Actions
  - malibu @ rohit on Jul 12, 2013 10:38 AM
    Mark Correct
    Correct Answer
    Hi,
    The problem is that the NFA's default temperature setting for the input / output loss is 0degrees K.
    It's purpose is to correctly account for the temperature of the loss element, which may be different than the temperature of the noise source. For example when measuring on wafer noise over temperature the chamber with the probes/cables is cold say -50 C. The noise source is outside the chamber and at room temp ~296 ish K. The loss inside the chamber generate less noise at -50 and this field accounts for the difference. I fixed the bug in my ATE code by updating the temp field at test time to reflect the actual temperature of the loss elements between the DUT and the NFA receiver.
    Hope that helps.
    .chuck
    Like • Show 0 Likes0
    Actions
    - rohit @ malibu on Jul 14, 2013 11:43 PM
      Mark Correct
      Correct Answer
      Hi,
      
      Thanks for the response. :)
      Find the attached setup schematic. As you can see only the DUT will be inside the temp. chamber and the instrument as well as the switching matrix both will be outside chamber, so do you mean that I should put ambient (room) temp before DUT and chamber temp after DUT in the NFA programatically.
      Attachments
      NFA Setup.png19.2 KB
      Like • Show 0 Likes0
      Actions
      - rohit @ rohit on Jul 31, 2013 3:10 AM
        Mark Correct
        Correct Answer
        Hello Malibu,
        
        I posted the problem which I am facing with NFA some 2 weeks before, It will great if you could spare some time and guide me some solution/idea.
        One observation which I made was, after giving the temp and loss values before and after the DUT, the readings are in specs of my DUT for 1.31 - 1.39 GHz but the values goes beyond specs for 1.3 and 1.4 GHz. Any idea why this could happen??
        
        Regards,
        Rohit.
        Like • Show 0 Likes0
        Actions
        malibu @ rohit on Jul 31, 2013 7:17 PM
        Mark Correct
        Correct Answer
        Looks to me like all the loss of your system is outside of the temp chamber.
        Therefor the temp setting of both loss elements should be ambient. not the temp inside the chamber.
        how much does the loss vary with frequency over the band you are measuring?
        if it varies by more than a few percent of the expected measurement you should be using a table that accounts for
        the frequency changes of the loss.
        hope this helps.
        Like • Show 0 Likes0
        Actions

N8975A input loss correction table not working as expected

Outcomes

Attachments

Related Content

Recommended Content