0
\$\begingroup\$

I'm new to rf design, although I'm learning rapidly. I've designed a PCB which uses an ultra-wideband (UWB) transceiver. The center frequency is selectable between 3.6GHz and 6.5GHz, with 500MHz bandwidth. I'm using a surface mount UWB chip antenna.

The application requires fairly precise frequency regulation. The transceiver has a crystal trimming capability of +/-20ppm, in 32 steps. The transceiver manufacturer recommends I assess the output center frequency while changing the trim values, and choose the value that gets me closest.

I have a frequency counter to measure the output frequency. If I connect the PCB to the frequency counter with coax cable, I see the signal and can do the tuning. Here are the specifications for the TF960 frequency counter.

Here's my difficulty:

I've connected an antenna to the frequency counter, but it won't pick up the signal over the air. It is a fairly low-power signal; close to -39dBm/MHz.

In the rf path of the PCB, I left pads to create a pi filter for later tuning. The two locations to ground are unpopulated, and the series location has a zero-ohm "resistor". All three are 0402. When I attached my coax-to-SMA pigtail, I removed the series shunt to isolate the circuit from the chip antenna. Then I scraped away some of the soldermask over a ground fill, and soldered the coax shield there. We obviously can't do this to calibrate production PCBs.

(sorry for the photo color) PCB

In the future, I'll choose higher-tolerance loading capacitors for the oscillator (they are now 5%), but I still hope to be able to do a non-destructive calibration.

My questions:

  1. Should I be able to read the signal over the air? It would make the rest of this easy. I'm currently using this chip antenna, both on my PCB and also plugged into the frequency counter. (I have a dev board from the antenna manufacturer to use with the counter).

  2. Would a signal analyzer be more appropriate for this task than a frequency counter? Are they more sensitive? Budget is tight, hence the inexpensive frequency counter.

  3. If I could find an rf probe (coax probe?) to land on the 0402 placements, can I use it with the chip antenna still in place? I'm considering something like this one, although I'd have to find one for higher frequencies.

  4. Is there a better way to do this? I feel that I may be looking down the rabbit hole...

Thank you.

\$\endgroup\$
  • \$\begingroup\$ Do I understand you correctly and the core problem is measuring the center frequency of an over-the-air received UWB signal so that you can tune your transceiver chip accordingly? \$\endgroup\$ – DakotaD Oct 21 '14 at 22:07
  • \$\begingroup\$ @DakotaD Yes, in a nutshell :) You sure were able to use less words than I did! \$\endgroup\$ – bitsmack Oct 21 '14 at 23:55
  • \$\begingroup\$ As a sanity check, I would connect the frequency counter directly to the UWB source via coax so that you can eliminate the unknowns of the antenna performance and have a transmission channel of known characteristics in which to determine the frequency counter performance. An additional inline coax attenuator would then allow you to determine the minimum detectable power of the UWB signal by the counter. \$\endgroup\$ – DakotaD Oct 22 '14 at 14:27
  • \$\begingroup\$ @DakotaD Thank you. I had connected a coax line connecting my board to the frequency counter, and I was able to trim the center frequency accordingly. The problem is that it required PCB modification (removing soldermask, etc) to eliminate the on-board chip antenna and add the coax. I am hoping to be able to detect the signal through some non-destructive method, to be used in a high-volume calibration fixture... \$\endgroup\$ – bitsmack Oct 22 '14 at 16:26
  • \$\begingroup\$ Can you make the UWB transceiver emit something like a CW tone at the center frequency? If so, what is the approximate peak power? This is probably the only way you'll get your existing frequency counter to work. \$\endgroup\$ – DakotaD Oct 22 '14 at 16:48
1
\$\begingroup\$

I may not answer every one of your questions but I think the following will generally answer your core needs:

With your current frequency counter and its limited sensitivity of -19dBm you are unlikely to be able to measure the transceiver frequency without some modification. Using a wideband high-gain amplifier such as this Mini-Circuits device may help, but noise and limited power could still be a problem.

A spectrum analyzer would be better choice for measuring the center frequency. A spectrum analyzer with a properly matched antenna should be able to detect the transceiver signal over-the-air. Setting the transceiver to a CW tone will ease the spectrum analyzer requirements, otherwise a real-time spectrum analyzer is good choice for UWB detection. The spectrum analyzer could be connected directly to the 0402 pads of the pie network using a coax probe but it would be necessary to disconnect the chip antenna or otherwise compensate for the additional 50 ohm spectrum analyzer impedance in parallel with the chip antenna. Another option for direct detection on the PCB would be to use an active probe.

\$\endgroup\$
  • \$\begingroup\$ Thank you, DakotaD. What you say makes sense. Do you know why the frequency counter was able to detect the signal with a coax connection? It seems to me that it would still be beneath the detection level of the counter... \$\endgroup\$ – bitsmack Oct 22 '14 at 18:16
  • \$\begingroup\$ @bitsmack - The direct connection probably had much lower wideband noise. With the antenna, your signal would be buried in general environmental noise. With the direct connection, you're basically looking at a 50Ω termination, plus your signal of interest. \$\endgroup\$ – Connor Wolf Oct 22 '14 at 18:19
0
\$\begingroup\$

If I read you correctly, it seems you have a complex UWB transmission and you want to determine the "average" centre frequency of that transmission. This may not get you where you want immediately but consider using a PLL (phase locked loop) to give you the average centre fequency of your transmission. Something like an ADF4155 fractional N and integer N synthesizer.

It's basically a programmable divider and frequency/phase comparator. You set up a reference frequency and SPI-bus the internal dividers and with the addition of a VCO you can make a stable output frequency from a wideband signal. It's made stable by having a loop filter characteristic appropriate for the speed at which the centre frequency of your UWB transmission might move.

\$\endgroup\$

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.