Purpose: To provide examples of how to best calibrated Fixture and Device Interface Board (DIB) paths when using a multi-port calibration methodology. This extends the measurement plane from the TIM ports up to the DUT through the Fixture and Device Interface Board (DIB).
Fixture and DIB calibration testplans save measurement data to a specific calibration variable name or path name. This cal name must be unique and consistent throughout the calibration process. The Device Connection Editor is used to create Fixture Definitions (RiFixtureDef) and DIB Definitions (RiDibDef) and to create unique path names for the connections in the Fixture and DIB. The tester configuration loads the appropriate Fixture or Device Interface "instrument" which contains the calibration data as RiFixtureCal and RiDIBCal guru objects. It is the Test Engineer's responsibility to create the appropriate calibration plan for the cal path name created in the "Def" and stored during the "Cal".
A Fixture calibration plan needs three unique measurements to characterize the Fixture transitions and cables lengths that are used to compute an Error Adapter (EA) value that is stored in the Cal Data Name. The final number of frequency points required will be a function of the amount of connections and mismatches as well as cable lengths and the bandwidth desired. All four measurements should have the same number of points across the same frequency. This is accomplished with four test sections, three that prompt the user to attach the correct calibration standard and the fourth to compute the Error Adapter (EA) and save to the Cal Name that is saved in the calibration Guru object as either a RiFixtureCal or RiDIBCal.
Using Calibration OSL Standards:
Best practices dictate that the first two sets of calibration data are to be measured with a complete reflection, as close as possible to 180 degrees out of phase, connected to the port. We normally use a characterized OPEN and SHORT standard with coax transmission lines, but in the case of waveguide or other medium, a SHORT and a characterized OFFSET SHORT can work as well. In addition, we will need to have a well matched LOAD measured and then followed by some array math in a CALC section. The local variable save names for OPEN or OFFSET SHORT, SHORT and LOAD must be unique, but are arbitrary. Any variable names are useable as long as they are unique. To calibrate a multi-port Fixture, it may be best and most efficient to design a set of three calibration DIBs that use the same design and assembly process as the DUT DIBs. One DIB should be loaded as an OPEN, another SHORT and third with LOAD (terminations). Otherwise an adapter can be used to connect OSL coax standards to the Fixture launch (a.k.a. Fixture Insert). DIB calibration can be more complicated, with an one solution being a set of three DUT packages that contain an Open, Short and Load. The "Open DUT" can be characterized and then loaded as a Cal kit, similar to the OSL standards. Alternatively, the cal data can be modeled or measured on the bench and imported into the DIB Cal.
To Design a Calibration Testplan:
- A System > reset cal button must be placed at least once in any Test in the Testplan to reset the current calibration data. This is typically added to the same test panel that the Cal Data Name is Calculated and Saved.
Important! The System > reset cal button causes the stored calibration data to be removed immediately when the testplan is first compiled, not when run. This is to make sure all the calibration data is removed before the new calibration data is measured. If a cal plan is compiled but not actually run, the previous calibration data has been removed and if that Fixture or DutInterface calibration is saved, the original data will be lost. In production, the Cal Exec handles this along with the operator choosing a set of Cal and Validate plans.
- Use the the correct "Fixture path" buttons as described in the Fixture definition to make sure that the hardware path is being set properly and the appropriate signal is being routed and measured.
- Make sure the path data is written to the unique calibration data name using a Fixture > Cal data save or Dut Interface > Cal Data Save button.
- The calibration Testplan should typically be built after a known working DUT Testplan is complete. This way the same Source power levels, Receiver BW, Rec Atten and IF Gain values can be used as those used by the device in the DUT Testplan. If a wide range of DUTs are tested using the same Test Fixture, then the fixture cal will remove the ripple effects in the fixture and another DIB specific cal can be designed to add to the Fixture cal data just like the fixture cal data is added to the TIM cal data. The Cassini system supports three levels of Calibration data active simultaneously. The DIB is often custom to that Device and can use the settings the specific DUT requires to refine the Fixture calibration once applied.
- Follow the steps below to Identify the Instrument's optimum Calibration Frequency Data Interval and Number of Cal Data Points.
How to Set the Calibration Frequency Data Points:
The number of calibration frequency data points should line up or match the instrument calibration point interval plus a few extra points above a below the bandwidth of interest. If the physical Fixture or DIB is very complex with many transitions, connections and cables lengths, the number of frequency points needed to fully characterize the ripple pattern of mismatches in that setup may be larger than the RF System interval. If the calibration data is desired at specific frequencies that are not at the same frequency points as the instrument's calibration data, the system will use a spline curve fit algorithm to interpolate the existing RF system calibration data. The Fixture and DIB calibration will then correct more precisely at those frequencies. This may be desired if measuring devices that have distinct frequency characteristics that need to be measured accurately to define their function vs frequency. Such devices are filters, coupling structures, amplifiers, mixers, or any bandwidth limiting tuning that needs to be characterized.
- Example A: If the Fixture or DIB being calibrated has a desired calibration bandwidth of 4 GHz starting at 2.5 GHz to 6.5 GHz and the TIM used to measure that data is calibrated every 50 MHz then, the Cal Testplan should match that same Frequency plan using a Freq Range Button staring at 2.40 GHz and ending at 6.60 MHz with a number of points equal to ( 6600 MHz - 2400 MHz = 4200 MHz / 50 MHz = 84 points plus 1 = 85 points )
Example B: If the Fixture or DIB being calibrated has a desired calibration bandwidth of 6 GHz starting at 2.5 GHz to 8.5 GHz and the TIM used to measure that data is calibrated every 10 MHz then, the Cal Testplan should match that same Frequency plan using a Freq Range Button staring at 2.48 GHz and ending at 8.52 MHz with a number of points equal to ( 8520 MHz - 2480 MHz = 6040 MHz / 10 MHz = 604 points plus 1 = 605 points )
Example C: If the Test Fixture or DIB being calibrated has a desired calibration bandwidth of 500 MHz and step size or 1 MHz starting at 3.515 GHz to 4.015 GHz and the TIM used to measure that data is calibrated every 10 MHz then, the Cal Testplan will exceed the number of Frequencies available using a Freq Range Button staring at 3.510 GHz and ending at 4.02 MHz with a number of points equal to ( 4020 MHz - 3510 MHz = 510 MHz / 1 MHz = 510 points plus 1 = 511 points )
To Identify the Instrument's Calibration Frequency Data Interval:
- From the Configuration window (System > Tester), select the appropriate instrument (Testhead, Testhead40, Testhead80, RfMeasure) and and choose Calibration > Inspect from the right mouse button menu.
- Inspect the cal data by double clicking on the variable for the port of choice which will bring up a small window and the double click on the data item.
A cal data inspection window displays each data point used to identify the frequency steps in the cal data by changing the index number and looking at the value in the freq ReFreqD() in MHz.
IMPORTANT: While uncommon, the frequencies and interval may change based on internal component changes to the TIM. This is reflected by the TIMs model number and the date of the Cal plan used. Make a note of the TIMs hardware revision or last two digits of the model (i.e. A0, A1, A2) that is referenced and the date of the date of the cal data or, preferably, the creation date of the Instrument's calibration test plan. If the frequency alignment changes, this note will be helpful for identifying the expected value.
To Identify the Best Number of Cal Data Points:
- Setup the calibration testplan to focus on one measurement by disabling all unnecessary test sections. (See Troubleshooting - Breakpoint & Control)
- Once you have identified the cardinal or specific frequencies and corresponding step size that the Instrument cal data uses, set the Frequency Points button in the Fixture (and DIB) cal plan to match those number of points across your bandwidth of choice.
- Run the testplan and plot that data on a Viewer. (See View Test Results)
- Increase the Frequency Points button to perform a sweep of a finer granularity in Frequency and plot the results to see if all the ripple patterns were visible at the previous system settings. Repeat this operation until no change is observed (See Example D below )
- The optimum number of steps or points are the minimum needed to no longer yields a change in the observed ripple pattern.
- If you notice any spikes, consider adding additional Frequency points around that location by adding an additional Frequency button.
- If there is still variance in repeated runs of the calibration data, then increasing the value of System > Average to reduce the trace noise or data variation of the measurement.
- Repeat this process on different Testers and Fixtures to help identify and quantify Tester to Tester or Fixture to Fixture variation.
Example D: Characterizing a Path from 2.2 GHz to 4.4 GHz
A frequency sweep with 100 MHz step sizes has large ripples in amplitude that are present and would be removed by a calibration. (See Figure 1) But it is not clear if there is any other fine ripple that is not shown due to the resolution chosen. Next, double the number of points to see if that yields any changes. As seen in Figure 2, here is a sweep over the same range with 50 MHz step sizes that shows the same large ripples in amplitude but also start to show some small ripples riding on those large ones that are present. These small ripples need to be characterized if the DUT measurements would use a resolution smaller than 100 MHz. Once characterized, it would be removed by a calibration. In Figure 3, a sweep with 20 MHz step sizes and we can see the same large ripples in amplitude but now also shown is pronounced ripples riding on those large ones that are present and would be missed with a calibration with a less number of points. To be sure we have all the finer ripple characterized we should once again sweep with 10 MHz step sizes. Here we see the same large ripples and the same small ripples riding on the large ones as the 20 MHz step size.
This shows the added effort in doubling the number of calibrations points would not yield any improvement in accuracy. Therefore 20 MHz step size seems to be the optimal number of points.
Figure 1: 100 MHz Steps
Figure 2: 50 MHz Steps (Double number of Points)
Figure 3: 20 MHz Steps
Figure 4: 10 MHz Steps
In the next few pages, the example of a Fixture calibration test plan and associated fixture designed to be calibrated using Dut Board calibration Standards. The Fixture definition this is written for is a RI 12 RF port fixture, where four fixture paths are defined as follows: RF2 >>DutRF2; RF3 >> Dut RF3; RF6 >> Dut RF6; and RF7 >> Dut RF7. The example is a template that can be modified for additional paths (switching or different DUT board connections).
1.) OPEN or (OFFSET SHORT) Cal Plan:
Note the reset cal, the fixture path and the appropriate routing of Source 1. Note: Reset cal only needs to be done once for each path. It is also only done during the compile. If the cal plan is run outside of the calibration executive. It must be recompiled every time.
2.) SHORT and LOAD
Are the same, note the absence of the reset cal.
3.) CALC - Note the selection of the appropriate path on the save calibration button:
To add additional paths, we need to create a path and then make the appropriate modifications in the test plan to use the path:
- Add them to the software fixture
- Activate the Fixture
- Copy an open, short, load, and calc cal test panel for the test ports involved.
- Modify the copied OPEN plan's Fixture>RESET CAL button to select the new path name
- Modify the OPEN plan's Fixture>HEAD RF X button to select the new Dut board Pin (DutRf1-Rf12)
- Create a new Data save name (System>SORTED BY FREQ) on the OPEN plan appropriate to the new path
- Repeat steps 5 and 6 for the short and load
- Modify the new CALC plan for the 3 new local variable sources (open, short, load)
- Modify the new CALC plan's Fixture>Cal Data button to select the new path name
This same procedure can be used to include switching as well by including the switching in the fixture definition.
Here is an example of a Cassini TestPlan that can be modified to create a custom fixture calibration plan.
Once the calibration plan has been created and saved into Guru, it can be automatically invoked from the System Configuration window, by a right mouse click on the Test Fixture Name and selecting the "calibrate" menu pick.
To open the Tester Configuration window, first press the "System" button on the main Cassini window, then chose the "Tester" Button on the pop-up window
The Tester Configuration window will appear that lists all the components currently installed in the Tester. Click to Highlight the Test Fixture.
From the Instrument pull down menu, select "Calibration" and another sub menu will appear, from that select "Calibrate"=
The Fixture Calibration Executive window will then appear but there will be no test plans listed in the upper left window.
From the pull down menus, select testplan>add.
You will get a dialog box with all the fixture types that have calibration folders created.
1.) Select the fixture type
2.) Select the testplan you previously created.
3.) When you close this dialog, it will ask if you want to save changes. Select yes, and the fixture will then always point to that file.
Calibrations can then be automatically run when in the calibration executive window by selecting the desired plan and "run selected."
Here is an example of an RI7100A TestPlan that can be modified to create a custom fixture calibration plan.
Once the calibration plan has been created, it can be automatically invoked from the fixture container window, by a right mouse click and "calibrate". To enable this capability the calibration test plan must located in the correct directory.
1.) Create a directory named for the appropriate fixture under riapps\caltests\fixture\
2.) Place the calibration plan in that directory
3.) From the fixture container window right click and select the calibrate option
You will then see the Fixture Calibration Executive window. There will be no test plans listed in the upper left window.
4.) From the pull down menus, select testplan > add.
You will get a dialog box with all the fixture types that have calibration folders created. (This is the folder you created under riapps\caltests\fixture\"your fixture name")
5.) Select the fixture type
6.) Select the testplan you previously created.
7.) When you close this dialog, it will ask if you want to save changes. Select yes, and the fixture will then always point to that file.
Calibrations can then be automatically run when in the calibration executive window by selecting the desired plan and "run selected."