This past fall a salesperson dropped off a Datamate Mix-Tek connector at my office. It’s one board-to-cable connector that has lines for power, data, and RF. They are available off-the-shelf with various numbers of each type of connector.
- Power Connector - 20A
- Signal Lines - 3A (at room temperature)
- RF Connector - <1.05:1 SWR up to 6MHz
I could not get the manufacturer to provide me a connector with RF coax connectors for testing, so I tacked U.FL connectors on to the board side and the coax pigtail. It looked decent under the microscope, but they cause significant discontinuity.
Board Side:
Cable Side:
I soldered to U.FL sockets together and did the network analyzer nulling calibration with the U.FL plugs mated to it.
When I replaced the soldered-together U.FL sockets with the connector under test, I got a lot of cyclical peaks and valleys in the gain plot. (The vertical axis on the right shows gain in dB.)
You can see, though, that the results oscillate around just under -1dB.
As a sanity check, I de-mated the connectors. The loss went up to 20dB at 2.5GHz and higher at lower frequencies, as expected.
The SWR plot stays under 3:1 and show the same oscillations due to the tacked-on connectors.
I ran the nulling calibration again, repeated the test, and got the same results. We’re seeing the discontinuity of soldering an SMT connector onto coax. I am surprised at how bad the discontinuity is, though, because I have gotten accurate results in the past by soldering a U.FL onto a solder blob on a PCB.
If I ever get sample with RF connectors or if I should have some need to make a board with this part’s footprint, I’ll re-run the plots. I suspect the loss would be well under 1.0dB, and this would be a good way to handle power, signal, and RF going from a cable to a board.
