"- The way the averaging works on the 8364, ie it sweeps the entire frequency range even in stepped mode before going back and performing the next sweep to add to the averages. The 8510 performs all the averages at the same frequency before moving on to the next one. I can sweep 401 frequencies at 500 averages per point on the 8510 in about 40 seconds, it would take a lot longer on the 8364."
"- But the main reason is that the calibration path for S11 isolation has been removed from the 8364. Yes, yes, I know, you shouldn't be using Isolation cal for S11, but I'll tell you why I use it. We perform a lot of our measurements in free space, especially to determine material permittivity and permeability at microwave frequencies. We used to do all these measurements in 7mm airline (see product note HP8510-3), but now do most of the measurements in free space using the same technique. We still use the 7mm airline sometimes(hence my interest in the 85060 Ecal system). Basically, 2 microwave horns face each other a certain distance apart, and we put the sample between them and measure S11 and S21 (for magnetic materials anyway, for purely dielectric materials we just measure S21 usually). Obtaining a SOLT (or even TRL) calibration kit is well nigh impossible for free space, so we use a Response/Isolation Cal for S11, and a straight Response Cal for S21. To measure S11 Isolation we put a convoluted foam absorber in front of the send horn and measure the response. All the reflections from the connectors, cables and the horn can then be removed. The response cal is a flat metal plate the same size as the sample. The cal routine removes the noise, finds the peak we need, sets that to be the cal plane and we are ready to go. The S11 cal only lasts about 3-5 minutes before we need to re-cal it due to S11 phase drift, but it's not too difficult.Now, we can measure all these parameters with the 8364B in an uncalibrated state and perform all the maths in an external computer program, but we have found that it is not as easy. Since time gating is an absolutely essential part of the technique, and the position (in time space) of the metal plate peak (calibration plane) changes with frequency space, you have to continually search for the peak and apply the time gate around it. The metal plate peak is not normally the highest (or even 2nd highest) peak in the trace, that is reserved for the myriad of reflections in the horn.However, the 8510 makes it much easier. I simply turn on the gate at zero seconds, 0.5ns span (say) and measure the response and isolation and the 8510 does it all for me! Even though the 8510 is uncalibrated, and the gate doesn't correspond to any peak in the time domain trace, it all works out in the end. "
"I recently read about a newly published cal technique (GRL or something) which appears to do the same thing, but even they say that the metal plate peak need to be the highest reflector (by 50dB!). I have never seen the metal plate higher than the horn reflections, let alone 50dB higher, and with S11 isolation enabled it doesn't need to be. "
"Since you are the Calibration guys, could you enable the implementation of the S11 Isolation cal on the 836X for us guys who don't always measure in the standard way? Just use the same one that's on the 8510, it does a great job The 8364 is "too clever", there doesn't appear to be a way to trick it into doing the measurement we want it to do."
"You don't need to do all of the math external in the computer. You can make the measurements and upload them into the VNA error arrays. The absorber measurement would upload into directivity, the short measurement would be multiplied by -1 (or nearly -1 if you want to account for the thickness of the metal plate) and uploaded into reflection tracking, the thru measurement would be uploaded into the transmission tracking term. That way you would still be able to quickly update the calibration and still take advantage of the internal time domain functions. Someone like OJ would be able to rattle off all of the commands you would need in his sleep. "
" Instead of using averages why not select a narrow IFBW on the PNA--you get essentially the same effect as a point averaging on the 8510. Doing this will also get the results much faster than the 8510. "
" At that time I discovered that if I used two horns on each port I could greatly improve my measurement results. "
" I was able to set up a TRM calibration in free space "
"I guess I could do that - since we use this for free space reflectivity measurements, and the distance between the sample and horns can be 10 metres, I was a little concerned that a narrow IFBW may affect the accuracy of the measurement. Any comment? "
"To use your N4696A ECal with the 85060C you would need a DB25-to-mini-Centronics cable "
""To use your N4696A ECal with the 85060C you would need a DB25-to-mini-Centronics cable "I don't suppose that's a standard cable obtainable from the local computer store is it? "
"Something like this? http://catalog.belkin.com/IWCatProductP ... t_Id=13565"
"The 85060C ECal control unit comes standard with DB25-to-DB25 cables (as the 8506x and 8509x ECals have that type of data connector). To use your N4696A ECal with the 85060C you would need a DB25-to-mini-Centronics cable (orderable from Agilent, Part Number is 8121-1047)."I guess I could do that - since we use this for free space reflectivity measurements, and the distance between the sample and horns can be 10 metres, I was a little concerned that a narrow IFBW may affect the accuracy of the measurement. Any comment? "I'd think it's possible that narrowing your IFBW could cause you to start running into the classic problem of measuring electrically 'long' devices with a network analyzer. If you try that and perceive that to be the case, making an increase to the sweep time would correct for that. Even after increasing the sweep time you might still find it to be faster than the 8510 with its point averaging. But I should note that when it comes to free space measurements I'm a novice compared to Dave so hopefully he might respond on this also Best regards,Brad HokkanenAgilent PNA Firmware Team"
" I am having an 85060C sent to me soon, and then I may be able to try it myself and let you know (although I don't have access to an 8510 in the office I'm currently in, so I'd be trying it with an 872x network analyzer). "
"HI Brad, I'm in England, and just turned in from the Pub, but I STILL remember that for IF BW of 1 kHz and below,we are in step mode in the PNA so we don't have to worry about electrically long devices, eh?"
" I wrote a short paper on the subject for AMTA.David V. Blackham. "Free Space Characterization of Materials." Antenna Measurement Techniques Association Symposium, vol. 15, pp. 58-60, 4 October 1993."Since you are the Calibration guys, could you enable the implementation of the S11 Isolation cal on the 836X for us guys who don't always measure in the standard way? Just use the same one that's on the 8510, it does a great job The 8364 is "too clever", there doesn't appear to be a way to trick it into doing the measurement we want it to do."It is on our list to add back into the firmware."
Retrieving data ...