I was hoping to get some feed back from one of the mods about the settings after calibration for measuring cables that are phase equal within a specification tolerance. I have the N5230A which I know will perform the task quite easily. I have read several documents from the internet about how this is done but no one is detailed enough with their explanation concerning the settings on their analyzer. Something I read in this forum said that you can attach a pigtailed cable and measure the delay. I didn't know you can measure delay without attaching the other end of the cable to the test port unless you were in time domain. I didn't think a time domain display of delay would be very accurate when one end of the DUT is open. I thought measuring phase in an S21 display or time in a group delay display was a transmission measurement.
I would like some help with the overall setup. The only way I know is to calibrate with a stepped sweep, 10ms per step, NOP = 401, IFBW @ 100 KHz, power @ +3dB or higher depending on the frequency span, attach the gold standard and save it into the memory and then compare the next cable using the math/memory and displaying mem/data. I can use the phase offset function if the marker at the frequency of interest is somewhere near the + or - 180 degree scale division and set it as close to zero as possible, then I wouldn't need to use the mem/data function. Does this explanation here sound close? If someone can answer the statement about measuring pigtails without attaching the other end, that would be helpful also. Thanks.
The pigtail approach basically refers to a TDR type of measurement where you would make an S11 measurement with one end of the cable connected to the VNA and the other end of the cable left open or terminated with a short or open standard. In lowpass one-port time domain, that open end will result in a large reflection that shows up some x amount of time after zero time, which represents the calibration reference plane. In that case, X represents the two-way electrical delay through the cable, and if you are comparing the delay of two cables (independent of frequency) you can certainly use that technique. However, if you want to make sure that two cables have the same phase at a given frequency, then I think you need to measure the S21 Phase of the cables precisely as you described in your post. If you take your golden cable's S21 phase and do a data/mem and then put in your cable under test, you won't have to worry about phase wraps, since the analyzer will first do the complex divide before computing the resulting phase. The phase wrapping is the byproduct of the Arctan() function used to compute the formatted phase data from the complex real/imaginary values and if the cable under test is less than 180 degrees different from the golden cable at any given frequency, then there should be no phase wraps in the displayed data/mem trace. In the PNA you can always choose the unwrapped phase format as an alternative.
Thank you for your input. I am glad someone finally answered my post and I appreciate your well informed comments. I can certainly measure the S11 (real) TD lowpass of the cable assembly which works great for cables that are electrically short, difficult for electrically long cables (aliasing) unless the span is narrow. The PNA has a nice TD menu function which allows you to select reflection X2 and also gives you a choice to display physical length in meters, feet or inches. Let me get your opinion on a couple of things; First, how would you short the open end of a cable assembly that has no connector attached to it? Second, TDR measurements usually have fast sweep times. Is there a way to slow the sweep speed down without affecting the calibration parameters? I know you can slow the sweep time down in the FD using the sweep menus (stepped, etc.,) but just increasing the sweep time in the TD doesn't seem to accomplish the same thing. The marker's display is changing so rapidly, its difficult to know what value to choose when trying to get an accurate delay measurement in the TD. My goal is to become more educated on creating and testing phased matched cable assemblies using high grade connectors and cable. I want to be able to make cables that are phase equal within +/- 1 degree or at a time base within a few picoseconds if that's possible. Figuring out the math for the amount of trim is one thing, (not too difficult, just labor intensive) but using our PNA to measure cables with one end open is something I would like to do very well. As always, I appreciate all the help I can get concerning this.
" First, how would you short the open end of a cable assembly that has no connector attached to it? "
Short the center conductor to the outer. Solder, wire, copper tape, fixture, are some possible methods.
" Second, TDR measurements usually have fast sweep times. Is there a way to slow the sweep speed down without affecting the calibration parameters? "
The PNA TDR does not require a fast sweep. If you have very long cables, you should set the dwell time high enough for the signal to propagate through the cable. You should probably use step sweep as well. From your original post, it sounds like you are already doing some of this. Maybe your TD measurement is on a different channel from the FD measurement? If so, put the FD and TD on the same channel and they should share the same settings so that the TD trace updates at the same rate as the FD trace.
The FD sweep is performed first, then the TD is a mathematical transform of the FD data. So the TD should always update at the end of the FD sweep (if they are on the same channel).
I think Dan has answered the questions you had, so there isn't much for me to add other than to say that in the PNA you can either explicitly set the sweep to stepped by going to the sweep setup menu or implicitly by setting the measurement IFBW to be 1Khz or lower. However, adding dwell per point does not necessarily switch you to stepped sweep. You need measurements of +/- 1 degree or better for your application and if you say the marker reading is changing too fast, then I presume the problem is that the marker value is changing more than 1 degree at a time. If that is the case, simply slowing down the sweep is not the answer; you have to reduce the noise in your measurement so that sweep to sweep variation in the marker readings becomes smaller than 1 degree. You can do this by either averaging or reducing the IFBW (both of which have the same identical effects). As for your calibration, I would recommend that you calibrate under the same sweep conditions that you are going to perform the measurement. That will always give you the best possible results.