8485A Power Sensor Calibration Question

[mango]

Hi,

I have a question someone may be able to help me with.   I am calibrating an 8485A power sensor.   The Ref Cal Factor at the last cal was 100%.   I ran it 5 times today on the PMCS and the Ref Cal Factor keeps dropping from 98% to now 94%.

Is this adjustable?  Or is this a sign of a damaged sensor?  Repairable?

Thanks,

Mango

[mdbuike]

1.  As far as I know, unless you have the 26.5 GHz mount and the PMCS running on Next Gen, you shouldn't be calibrating it to just 18 GHz.

2.  If you have to, run it manually (sucks, but it doesn't really take that long) and compare your results.

3.  If you are limiting the mount to 18 GHz, shame on you...we've calibrating mounts to 50 GHz for years..you just need to get the right standards from AFPSL, and if you show a need, they can send them.

Mike

[CalLabSolutions]

Mango..
That is odd. Usually an 8485A power sensor's cal factor @26.5 is lower that the 50 MHz Ref.  I have never ran the PMCS system but I am assuming no cal factor can be about 100% (Limits of the 436 Power Meter).

No, there are no adjustment to 848x Power Sensors (Just Cal Factors).  Yes, the sensor is repairable.

The biggest problem with Power sensor is people damaging the connectors.  Usually they over tighten them and push the center pin in.  If this happen, when the power sensor is connected, the center pin is not able to make contact.  Under the right conditions a power sensor a power sensor could be more responsive to higher frequency than lover frequencies.

Do you have a calibration kit with pin depth measurement standards?

I would check the pin depth of the DUT Sensor.  If that is good, verify the PMCS Power Sensor Calibration system with a known good Power Sensor.  The system should be able to create cal factors for your know good power sensor with in the uncertainties of the station.

If everything looks good.  I would suspect the last cal with a 100% Cal Factor at that frequency..

[mango]

Hi again,

Thanks for the responses.   I think I might need to re-clarify my original question though.

The mount is an 8485A with a range of 50 Mhz to 26. 5 Ghz.   It has a limited cal on it to 18 GHz by a previous technician.   We have no need to cal above 18 Ghz at the present time.   Not sure I understood the "shame on me comment" however I will not take it personally.   However, if you are telling me to run the cal from 50 MHz to 26. 5 Ghz for a more repeatable 50 Mhz reference cal factor, please say so.   This would be easier to understand.

On the previous cal chart for this mount the reference cal factor at 50 MHz is 100% for standardizing the sensor at 1 mw (0 dbm).

When I run the cal, my reference cal factor at 50 MHz@1 mw on the 436A Power Meter is ranging from 98% to a low of 94%.

I am assuming I will have to find the most repeatable reading for my reference cal factor by running the PMCS manually and comparing it to what I get when I run it automatically.   I appreciate that suggestion.   Thank you.   

If I accept a new 50 MHz reference cal factor for the sake of "discussion purposes only" right now, say at 94%, and do my comparison as is directed by the 33K procedure between the old cal chart and the new chart, I would not meet the 33K specs.   

How would this be accepted for purposes of Quality Review or AFMETCAL audit?  Is this sensor still serviceable?  Should I trust the previous cal chart from when it was last cal'd by the previous technician (who is no longer with us)?

What happens to a power sensor when it does not match within limits of its previous cal chart as directed by the 33K procedure? 

Appreciate more feedback on this.

Thanks,

Mango

[RichMojo]

If a power sensor does not meet the K-pro specs, it is just considered a "K" action.   But, the sensor must be repeatable.   We run them twice, once as a warm-up and then the second as the actual cal.   These two runs must be close or it might point to a problem with the sensor.   If you are getting readings from 94 - 99 for the reference, there seems to be a problem.

[mango]

Thanks Mojo

[mdbuike]

Sorry about the shame on you..but I have found over the years, that it is best to calibrate, and if necessary, adjust the equipment to thier full range, whether a sensor or spectrum analyzer, or even an oscilloscope..

I was not saying to cal the sensor (8485A) to 50 GHz, but we have been sent sensors from other labs (8487A's) and the new MMR sensors, because they say they can't do them...

Well, if you have a 50 GHz generator, a 50 GHz MMR, get the standards to do them..

It's late, and I'm venting a bit, but we are supposed to be metrologists, not mechanics..

So, off to tuck my granddaughter in, listen to my new Craig Chaquico CD, and play bass at church tomorrow

Mike

[pmelgeek]

While you may be getting consistent calibration factors from your PMCS, this does not necessarily mean they're good calibration factors. Have you checked the SWR of the sensor? The uncertainties you're trying to compare too (33K4-4-35-1 Table 1) are based upon the SWR of the PMCS and the SWR of the 8485A itself. If the SWR of the 8485A is too high (bad pin depth, damaged center pin, etc..) you will still get consistent calibration factors from your PMCS, but the "uncertainty" of those factors will be higher than what's listed in Table 1. ...and unfortunately you'll never know this unless you actually measure the SWR of the 8485A (which the Air Force 33K4-4-35-1  doesn't currently do.)   

Of course since you mentioned only calibrating it to 18GHz, would you happen to be using an N-3.5mm on the end of your F1109 to do this?  If you are, I would highly suspect the adapter you're using is the problem (or was the problem the last time it was calibrated). This method is tongue-in-cheek at best, and quite arguably "untraceable" since you never know the SWR of the adapter to actually compute the uncertainty. If you have access to the F1117A, use that instead and loose the adapter...you reduce a big chunk of your measurement error.

Sorry to get on my power sensor soap box here, but quite honestly I don't know how the Air Force gets away with calibrating power sensors the way we do...it's total BS. How do you ever know if the sensor is actually good or not simply by comparing it to the last chart!? ....and where in 33K4-4-35-1 does it tell you to even do that much? It just says to "prepare a chart".  It doesn't say to campare it to anything...we all just assume that part (and so do evaluators). The SWR specification is where the measurement uncertaintly lies, yet it's never checked. You might as well be stamping them in scheduling for all the good it does...and don't even get me started on calibrating MMR sensors. The PMCS itself is a good system...but it's only half the calibration. When the Air Force ever wakes up and starts actually "calibrating" power sensors instead of just charting them...there's going to be allot of NRTSed sensors sitting in DRMO!   

   

[RFCAL]

From your posts,I would suspect the pin is loose or the pin depth is oot.This would account for the ref factor variance

[piter_c]

Hello,
I have very similar problem: I have calibrated 8481A power sensors with PMCS (surecal, 1805B and F1135 thermistor mount).  Ref Cal Factor drops from 100% to 94%. 
At the beginning everything is ok: picture "1wykrespocz. jpg".
After zeroing 1805B - chart drops about 6% down: picture "2wykresprzed. jpg"
Caf Factors are entered to file properly: picture: "3cf. jpg"
But after calibration Surecal changes it: picture: "5wynik. jpg"
I have tried on three different 8481A - all the same.
I ran it manually - Cal Factors are in tolerance to previous ones.
Pins seemed to be ok - no damage.  Adapter's attenuation is added - I charted it with N5530S.
I'm out of ideas what can be wrong.  Can anybody help?
PS.  English is not my native language so I'm sorry if I messed up  :-).

[CalLabSolutions]

piter_c ,
Your English is great, better than some of the people I know..

I am not a  SureCal expert, I have never used it. We wrote our own power sensor calibration solution because our customers wanted to calibrate linearity on the Agilent E-Series power sensors.  A required step according to Agilent.

From  your description it looks like the software is testing the Cal Factors correctly..

What I believe is happening is the 1805A is an absolute RF power source.  Meaning when you have set the system up correctly zeroed the 1805A and bolometer.  Then turned on the RF powe,r you have a calibrated output of 1805B setting times the Bolometer cal factor.  (1mW * CF)

The HP\Agilent 848x power sensor are relative measurement devices.  Meaning they measure power relative to the REF Output of the Power Meter.   The REF Cal factor and the Frequency cal factor can be scaled.  So you can subtract 1% from the Ref Cal Factor and every Frequency Cal Factor and still make accurate measurements. 

From your description, what it looks like SureCal is doing is scaling all the Cal Factors to Ref Cal Factor = 100%.  This is an acceptable practice for 848x Power Sensors.  The reason you have to scale the power sensor's cal factors is because some power meters like the 436A cannot have Cal Factors above 100% or below 85%.  While the Agilent E-Series Power Sensor and N848x Power Sensors must have a Ref Cal Factor of 100%

[Hawaii596]

As I have some ANCIENT experience using the old Weinschel version of the system and in my recent previous job, we had power sensors cal'd by our mother lab on Surecal, I can address a little of the shifting of reference cal factor to 100%....

We had a problem with a customer who used 11722A's (for any not familiar, it is the sensor module used on an HP8902A and in part, functions kind of like an 848xA series power sensor; and is calibrated like an 848xA series power sensor.  The software the customer used them in, in a factory environment, was used by operators.  Some of these 11722A's had cal factors (frequency response curve) such that some of them were higher than the reference cal factor.  Surecal was automatically shifting the reference cal factor to 100% and referencing to that.  The customers software would not allow any cal factor to be entered that was higher than 100%.  The Surecal was effectively shifting so that some cal factors were higher than 100%.  Unacceptable to the customer.

So (and here's that part that may relate to this circumstance), after some discussions, we were able to have them manually change the reference cal factor down.  Bottom line, it has been my (indirect) experience that Surecal does indeed shift reference cal factors up to 100%; and it can be manually overridden to force it down to where ever you want it.

[CalLabSolutions]

Our Software does the same thing..  Only not automatically.

The user has the option (on 848x Power sensors) to scale the cal factors.  We built in 3 options

1) Scale the Ref Cal Factor to 100%
2) Scale the Ref Cal Factor to an entered number
  * Handy for comparing the Old cal factor to the New one.
3) Scale to max 100%.  This will adjust all the cal Factors to <100%.  However it will use a whole number for the Ref Cal Factor.

It also has the option to store an autoscale operation in the calibration template.  So after the cal factor data is taken, the software will autoscale the cal factors. (The technician does not have the option to choose).

Thanks for you input, Hawaii..  Glad someone who uses SureCal chimed in..

[CalibratorJ]

In my experience with SureCal, you can have it force the ref cal factor to 100%, or let it autoset the ref cal factor to keep all of the cal factors in XX-100% range (set by user).

From what it looks like to me, someone may have modified your test profile to set the ranges of the cal factor to 94% max. I would definitely go in and double check the test profile.

But, then again, I have never used the 1805B, just the 1806 and some 3458s. I used the bridge strictly as a warmer for my F1135 or M mounts to do intercomparisons.

[LarryH_ET]

For part of your original question Mango: 
To test for a damage sensor, see how well it holds ZERO.   A blown sensor will drift VERY bad as soon as the zero LED goes off.   You WILL know it when you see it.   Good sensors drift VERY slightly depending on warm-up time, maybe 1 digit every couple of seconds.

Another good cross-standard check is to use an HP 8478B thermistor mount to check the PMCS.   This thermistor mount is the STANDARD at Agilent for calibrating power sensors.   I usually run a calibration on the F1109 with it BEFORE and AFTER the AFMETCAL exchange.   I did catch a BAD error on the new F1109 with this method - they must have had a loose connection because it had a bad 14GHz drop out in the cal factors loaded.   I even did up a spreadsheet that compared before and after to the RSS uncertainties.