I'm speculating some of you do a lot of these. . . This regards Calibrating the 1. 00 mW 50 MHz Power reference such as on HP 435B, 436A, 437B, 8902A, etc. , using HP 432A with 478A-H75 or -H76. I had a few Agilent E4417A meters with power reference amplitudes skewed on the high edge of spec recently, which has caused me to dig into "THE FORMULA. " So this question is for you who are mathematically savvy and experienced in this specific measurement. .
The formula in the manuals has been consistent for a long time:
"Pmeas = 2Vcomp (V1-V0)+V0[2] - V1[2] / 4R(Calibration Factor)"
I won't go into the components of the formula, just the formula itself.
I looked in a GIDEP K4 procedure that had:
"PRF = 1. 25 X [2 VCOMP (V1 - V0) + (V0[2] - V1[2])] / CF of Thermistor Mount
I looked in an old HP 432A manual that had:
"PRF = ( (1/4R) X [2 VCOMP (V1 - V0) + V0[2] - V1[2]] ) / EFFECTIVE EFFICIENCY"
First, I got three different answers from the three formulas above. Second, I believe the current formula as in the manuals is ambiguous. It doesn't describe converting to common units, converting the cal factor to a decimal number or not, etc.
If I interpret the current formula as follows, I get in tolerance readings:
"Pmeas = [(2Vcomp) X (V1-V0)] + (V0[2] - V1[2]) / 4R(Calibration Factor)"
With notes to convert all voltages to common units (all V or all mV), and convert cal factor to decimal format
Here are my numbers:
Vcomp = 4. 818 VDC
V0 = 0. 035 mVDC
V1 = 84. 035 mVDC
R = 200. 10 Ohms
Cal Factor = 99. 88%
Any thoughts on this?
The formula with 1.25 in it I've seen and use and there is a note that states something along the lines of 1.25 being the value for 200 Ohm mounts. So unless using 478A H73 or whatnot on a 432A/B I would either throw the formula out or try to find the correct multiplier. It may not even be the 1.25 but one of the values is relevant to the mount resistance in that formula and I know its not CF, V0, V1, or Vcomp.
Gidep 33k4-4-283-1 I believe that is where I saw it. If not gidep by part number 8003 by Gigatronics/Wavetek .
Beyond that formula I'm clueless in St Louis.
I haven't had any problems with it but your numbers look odd based on my experience.
Here's a meter I did a while back and it's typical of many I have done.
Vcomp 5.123 V
V0 97.5 uV
V1 78.798 mV
My meter resistance is 200.19 and cal factor 100.01
Here's a couple things I do
1. Vcomp when I move the lead to a ground I an using the shell on the other BNC connector (recorder out I think it is)
2. Measurng the resistance of the power meter-use 2 K ohm range on your meter, I use a Fluke 8508A and if I let it go to the 200 ohm range I get a lower reading that causes problems, used to see this on the old Tek PG506 when you measure the resistance of the outputs, the range you'd normally use would give a pretty bad reading but by going up a range it would be good. Must be something to do with the current the meter is putting out while measuring resistance.
Are you doing the Output SWR on the E4417 also, same measurement but w/mount resistance to 100 and some wild formula mixing the 200 & 100 readings? Luckily one of the guys here made a spread sheet to do it, I never was able to work it on a calculator. I havn't seen one fail it yet but am looking, I'd like to be able to say if it passes the 1.00 mW test then the SWR test isn't nescessary but have no evidence yet.
It's late Friday evening and my brain is pretty well fried. But I'll trade you my spreadsheet for the 1 mW cal for your spreadsheet for the SWR.
Lets see if I can answer some of your details. . .
I use an HP 3458A as DMM. I do Vcomp with high lead on the Vcomp terminal (I have BNC - SINGLE BANANA adapters on both Vcomp and VRF), and low lead touched against the frame of the 432A. I have the 478A-H76 connected to the 1 mW out on the UUT with 1 mW ON when I make that measurement.
For V0, it varies. My 3458A is on the 100 mV range. I noted my V0 in mV units (. 035 mV) which is 35 uV. Maybe it was the units that looked odd. I have everything connected as above. I just switch off the 1 mW on the UUT for the V0, and measure using the Vcomp/VRF connectors with the Banana adapters and two Pomona banana cables to the 3458A.
For V1, I leave as in V0, turn on the 1 mW source on the UUT, and take my mVDC reading on the 3458A.
Then, leaving everything powered up (1 mW ON), I disconnect the VRF lead, and touch it to the chassis of the 432A and record the VDC reading.
Once I fine tuned my interpretation of the formula in my original post, with all the numbers in that post, I got 1. 0037 mW.
The difference you get in your Vcomp is interesting. I wonder if that is based on a power supply voltage level in the 432A?
As for the SWR, that's an interesting point. I'm thinking because the SWR is so well characterized on the 478A-H75/H76, if you get a good amplitude reading from the reference, you know by definition that the 1 mW is coupling into an accurate 50 ohm load. Therefore SWR can be assumed to be in spec as well (late evening ramblings). I've wondered why they would want to test that when they never did on older models. Sometimes I think SOME tests are designed to give someone job security who creates them.
Okay, enough rambling. I need to go home. Thanks for the inputs.
Unlike the older meters you mentioned, I believe that one can be set to output different frequencies other than 50 MHz. You might want to scroll the menus if you haven't already to make sure that it is actually set to output 50 MHz. You could be inducing significant errors with your H75 / H76 sensor if you are dealing with a higher frequency.
I'm stumped, I hooked my setup to a sig gen and to get a V1 reading like yours I had to kick the power up enough to peg the meter pretty hard. It should be parked right about on full scale when the test is done.
Your Vcomp still seems odd, I mentioned I used the shell of the recorder out for my ground, that does cause about 2 mV difference in Vcomp vs. the frame ground. Vcomp seems to be a constant, it's unaffected by changing the input to the sensor.
My best guess is my 432A is an old used unit bought for us when we opened the lab. Might be inaccurate by normal standards but still works as the voltages measured are based on zero input to full scale ratios and the measured resistance.
You got me. I definitely measure as stated. And apparently it works out, since I was getting about 1. 0037 mW.
The age of the meter should not matter as the purpose of the formula is to remove the meter entirely. You are essentially measuring right off the bridge of the thermistor mount.
At Andrews, I set up a system with an 8840AAF, rear jacks connected to measure Vo and the front jacks to measure V1. A simple push of the front/rear jacks button to switch between the readings.
I also used an old PC with excel and our QA certified, locked down spreadsheet. THis meets the metcal inspectors requirements of keeping control of spreadsheets. It also insures the readings are done correctly every single time.
Once you get the reading correct and verify the value, switch the DMM to the recorder output and measure the meter deflection voltage. You can note this value with correction factor (known mW vs. DCV out) as a cross check for future cals. It is not a certified measurement but a good "2nd" opinion with reasonable resolution to determine of there is an issue or calculation that is wrong. Of course, use the same 432B, DMM, 478A, etc. for this crosscheck each time.