Bruce, Steffan,
I think you are forgetting that there is another verification you can perform - completely remove the ECal module and connect the test ports together and verify the error in the calibration is close to 0.00dB across the frequency range.? That is your error ;-)? ? You can also confirm the phase starts at on the right side of the smith chart for the open port as one would even in manual calibration.
Answering your question the signalling on the DB25 uses the same protocol, in fact the first board is the same between the low and high band modules.? The primary difference is the low band uses 7 switch signals to the lower board and the high band module uses a total of 16.
Analog voltages to the PIN switches are different with the low band using 0 and -12V whereas the high band unit uses +5 and +24V for its pin switches, so the final driver board is completely different between the 2 bands.
The digital interface controlls access to the EEPROM which contains the measurement data, and a way of enabling any of the 7 (or 16) switches in the RF path.
I did open my 85097 (which I use on my 8753ES analyzer) but haven't reverse engineered it other than a conceptual understanding of it taking the signals from the parallel port and using a control line (or maybe register) to multiplex between the A and B ports.?
Following onto answer Steffan's questions,
1.? Yes recalibration this should be possible.?? The contents of the EEPROM are simply the actual S1P measurements for each PIN switch combination, so its a matter of taking a full measurement (on a mechanical calibrated VNA) against each of the PIN switch combinations to get a set of known values.? Note there are only a few standard combinations used? which have been chosen by the design engineers who have a better understanding of the phase delays along the transmission line required for best calibration.
2.? As per #1 above, because you know the complex impedance of the ECal port the VNA will measure the impedance it sees and will de-embed to solve the 12-term matrix for calibration (or 3-6 terms depending on the calibration type selected).???? Instead of a short, open or 50R in the solution matrix we use the known values for the ECal.
It should be noted that the ECal is never a perfect thru, open or short at any time of the electronic calibration.? From memory the short is a few ohms but to "hammer it home" the following 2 PIN's are also energized to ensure there is a stronger reflections.
For successful calibration there must be a ~180deg reversal observed at each frequency (perhaps 90 deg - I know quarter wave is required for unknown through).? This can be easily achieved in the high band cal with the length of transmission line.? For the low band? module there is also a PIN switch halfway through which simulates a open to create a larger phase reversal.
If you have ever used a sliding load during calibration then you will know what I mean by 90 degree observations to get a solid calibration point.
This 180deg requirement is also what sets the lower frequency limit of 45MHz for our units.? As noted in Joels book more modern ECals use a GASFET switch which allow for a more traditional SOLT standard at lower frequencies.
The Arduino board I built is able to reprogram the EEPROM but I haven't added that into the code.? The datasheet wasn't an easy read at the time! ?? After dumping the contents I found the ECals to be better than what I can achieve mechanically, so if I re-calibrate mine it would be a step backwards. ? Given I don't posses a 4 receiver VNA and lower grade test cables my calibrations will always be inferior.
Finally I believe there must be a version of firmware for the 85060C which supports the calibration process.? This is based on the references in the dumped EEPROM data.? Maybe its hidden in the current version somewhere.? If anyone is able to get a copy it would be greatly appreciated.
--
Wayne
