0
\$\begingroup\$

I have a circuit with multiple input channels (~10-20), where to each channel I am applying RF voltage signals, usually sine waves, with varying frequency in the range of 1MHz to 1GHz and 0-5V peak-to-peak amplitude.

I am using SMA connectors to send the signals from my Arbitrary Waveform Generator (AWG), however, I am currently looking at creating my own PCB with transistor oscillators on it.

The problem I am facing is that SMA connectors are too bulky and require SMA cables to connect my Device Under Test (DUT) to my oscillator circuit. Moreover, they are a bit of an overkill for a circuit that only goes to ~1GHz.

My question is: Do you have any recommendations for multi-channel connectors that are preferably clip-on (no cables) and support the necessary bandwidth and voltage range? I have been looking at U.FL connectors for a while as they are inexpensive, but as far as I understand, the female connector is always on a cable and only male connectors exist as SMDs.

\$\endgroup\$
3
  • \$\begingroup\$ U.FL-PR-SMT2.5-1(10) - U.FL female in SMT but not cheap \$\endgroup\$ Commented Aug 7 at 4:51
  • \$\begingroup\$ I was checking them out some time ago but, in the datasheet, they specify that multiple male to multiple female U.FL SMT connectors between two PCBs is not recommended. Also, I don't know if slight misalignment between the male and female SMT connectors during soldering can cause the boards to be impossible to connect. Have you tried something similar before? \$\endgroup\$
    – Hikikomori
    Commented Aug 7 at 7:40
  • \$\begingroup\$ yes, fine alignment would be very tricky. \$\endgroup\$ Commented Aug 7 at 11:06

3 Answers 3

1
\$\begingroup\$

You didn't say so, but reading between the lines I seem to understand that you're looking for a board-to-board coaxial connector suitable for many mating cycles between a text fixture and devices being tested. You wish to make multiple connections at once, rather than one at a time.

If so:

Consider a set of compression coax connector: https://www.te.com/en/product-1658260-1.html

You would install a number of these compression coax connectors on the fixture. The DUT would have flush coaxial PCB pads in the corresponding locations. When you place a DUT on the test fixture, you press it against these connectors which would retract slightly as they apply pressure against the coaxial pads and make contact. The fixture would lock the DUT in place to retain the contacts. When the test is done, you would move to the next DUT.

Note that the cost of the connectors on the DUT is zero: pads are free.

TE 1658260-1

{Source TE 1658260-1}

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Thank you for your suggestion! The DUT is an IC, wirebonded to a PCB with fanning out RF pads. I think it would be better to have the connectors on this PCB. On top of it I can put PCBs containing control circuitry. However, I think adapting the current fixture would be a bit of a problem. Also, the spring-loaded connectors might cause some movement/misalignment in the IC when new control circuitry is inserted, unless it is done very carefully, so I think everything becomes more of a mechanical engineering problem at this point. \$\endgroup\$
    – Hikikomori
    Commented Aug 6 at 21:10
  • \$\begingroup\$ "The DUT is an IC, wirebonded to a PCB with fanning out RF pads" -- I wish you would have clarified that in the text of your question. The better information you give us, the more helpful our suggestions can be. \$\endgroup\$ Commented Aug 6 at 22:42
  • 1
    \$\begingroup\$ Sorry for not clarifying earlier. Still, I think that for devices where I have less input channels, say 5-10, the compression coax connectors can be quite useful because I can get rid of the cables of the SMA connectors. It is quite useful information for me! In our lab we are mostly using D-sub connectors for DC to low frequency applications, SMA or SMB for RF, and less often - IDC connectors. \$\endgroup\$
    – Hikikomori
    Commented Aug 7 at 7:53
1
\$\begingroup\$

you're searching for a multi-ganged RF connector. They're available from a few vendors. Here's an example of one such product https://www.samtec.com/rf/cables/micro-ganged/ although similar products are available from other vendors. They're not cheap, but they would give you a PCB mount connector that would support an SMA at one end of a coax, and a smaller PCB connector on the other.

\$\endgroup\$
0
\$\begingroup\$

I recommend ZIF to FPC or micro-ribbon headers with interleaved pairs for Gnd isolation reducing crosstalk.

Add series R and terminate impedance to optimize results if unknown but 100 ohm single 200 ohm differential was common. A Gnd. foil shield will lower the common mode impedance more.

This then may be tested for signal integrity with various tests for crosstalk, Z, and pF per channel.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ I have used ZIF FPC connectors before in other projects and am quite happy with them. However, the application back then required much lower bandwidth signals. They are inexpensive, have multiple channels, and actually building a prototype and hooking it up to the VNA might be the best way to go. I will definitely try using them. \$\endgroup\$
    – Hikikomori
    Commented Aug 6 at 21:18
  • \$\begingroup\$ Glad you like them, adhesive mu metal foil or similar foil will improve signal integrity or maybe even metal duct tape might work with shield to VNA only. Z matching may be tricky at upper limit for s22 => 12 to 15 dB \$\endgroup\$
    – D.A.S.
    Commented Aug 7 at 5:07
  • \$\begingroup\$ Not sure how you will mux all the channels to the VNA unless you use multiple OSX or SMA jacks \$\endgroup\$
    – D.A.S.
    Commented Aug 7 at 5:12
  • 1
    \$\begingroup\$ I have RF probes and was thinking of testing channel by channel. The input RF probe would touch the metal pad of a channel on the ribbon cable that goes to the FPC, and the output RF probe would touch the gold finger on my PCB with the FPC connector, where the DUT (IC chip) will be wedgebonded. \$\endgroup\$
    – Hikikomori
    Commented Aug 7 at 8:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.