I'm looking for a board mount (IC) solution that will take the differential output of something like ONET1191 @10GHz and convert it to 50 or even 75 Ohm single ended.

I've looked through various suppliers for:

  • Baluns (and other transformer solutions)
  • Differential Amplifiers
  • Receivers/Transceivers

However, I can not seem to find any modules with a pass band around 10GHz. I see tons and tons of ~6GHz modules (makes sense because 1080p60).

Alternatively, I could simply pass it out through 100 Ohm twinax and into a discrete RF amp that takes differential input (those do exist, but they are expensive!!!). I am hoping to avoid this, due to cost reasons.

Just throwing this out into the wind to catch some useful suggestions, thanks!


The typical solution is just to terminate one half of the differential output and deliver the other half to the single-ended transmission line. Your device provides up to 700 mV peak-peak output, which is as high as you're likely to see in a logic buffer at this data rate.

If you need a buffer amplifier to maybe prevent damage to the more-expensive TIA chip due to being connected to an off-board circuit, you could look at something like On Semi's NBSG16 12 Gb/s buffer.

If you need more than 700 mV swing, you could look at something designed as an electro-optic modulator driver, for example Picosecond Pulse Labs (now Textronix) 5865 amplifier. This can do up to 8 V peak-peak output at 12.5 Gb/s.

  • \$\begingroup\$ The only thing that makes me pause before utilizing the dummy termination scheme is that I don't know how long the line to RF mixer might be. The RF mixer has a fixed dBm window for the input. Thanks for your suggestions. \$\endgroup\$ – Steve Novakov May 13 '14 at 20:30
  • \$\begingroup\$ @SteveNovakov, 700 mV peak-peak is a pretty big signal for digital at 10 Gb/s. Also, your part is a limiting TIA. Are you sure you want to send a digital signal into an rf mixer? \$\endgroup\$ – The Photon May 13 '14 at 20:32
  • \$\begingroup\$ You could send the signal differentially to the vicinity of the mixer, then use the NBSG16 as a buffer and dummy terminate there. \$\endgroup\$ – The Photon May 13 '14 at 20:33
  • \$\begingroup\$ Yeah, I'll have to think about this. It will have to be on 1 PCB though. I've inherited a lower-frequency design that I have to now improve for 10GHz. I don't know that I will keep that exact limiting buffer, it was just an example. I could, for example, just terminate the fiber coupling diode and then just deal with a single ended signal to start with. \$\endgroup\$ – Steve Novakov May 13 '14 at 20:36
  • \$\begingroup\$ @SteveNovakov, without knowing about what kind of signal you're measuring (digital or rf at least), and what your overall goal is, I don't think I can help you any more than this. Also, 10 GHz or 10 Gb/s? \$\endgroup\$ – The Photon May 13 '14 at 20:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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