I'm trying to make my own board for a CC1101 transceiver that will operate at 915 MHz. To do that, I'm just copying the reference design layout from the eZ430-Chronos-915 RF Access Point that uses a CC1101 with a MCU to be used as a dongle. This is the reference design PCB, I could only find it in PDF format as an image:

enter image description here

(Inside the black square are the 0.25 mm tracks and inside the red square is the 0.4 mm track)

And this is the schematic section corresponding to the CC1101:

enter image description here

By measuring the image I got that the tracks coming out from the RF_N and RF_P pins are 0.25 mm wide, this is on a two-layer FR4 board with a thickness of 0.8mm, so that will be around 100 Ohm impedance for the tracks which is right because it's a differential signal AFAIK (to maintain 50 Ohm impedance, correct me if I'm wrong please).

These two tracks go to a Balun (this one) and the antenna connection is made on pin 1 of the balun. This is the typical layout:

enter image description here

The balun says the impedance of the track going to pin 1 must be 50 Ohms. I measured that the track that goes out from pin 1 is 0.4 mm wide, with a two-layer, FR4 board with a thickness of 0.8 mm that is an impedance of around 88 Ohms, not even close to 50 Ohms! To get a 50 Ohm impedance the track should be 1.5 mm wide.

What am I missing? I know this design works because I have this board and it works as expected. I was planning to put a 50 Ohm SMA connector in the end, should it then be 50 Ohm or something closer to 88 Ohm?

All the pictures are taken from here.

  • \$\begingroup\$ With traces as short as shown and for low frequency as 1 GHz, the small characteristic impedance deviations do not matter. \$\endgroup\$ – Ale..chenski Sep 12 '18 at 3:42
  • \$\begingroup\$ And it would be really nice is you could provide actual links to your sources of your pictures and layouts. \$\endgroup\$ – Ale..chenski Sep 12 '18 at 3:49
  • \$\begingroup\$ @Ale..chenski Added the link to the PDF! \$\endgroup\$ – Andres Sep 12 '18 at 4:27

The trace segments in question are just few mm long. For a 1 GHz signal the wavelength is about 160 mm on a typical microstrip on FR4 PCB (~150 ps per inch). Any characteristic impedance imperfections shorter than 1/10th of wavelength has no real effect on signals. So no worry, use the reference designs as-is.

  • \$\begingroup\$ Just a comment: If lambda/8 can transform an impedance 90 degrees in smith chart, lambda/10 is not so diffent. \$\endgroup\$ – user94729 Sep 12 '18 at 7:34
  • \$\begingroup\$ @Mikey, the segments in question are about 3mm long, so you have 3/160 ~ 1/50th of lambda. \$\endgroup\$ – Ale..chenski Sep 12 '18 at 15:42
  • \$\begingroup\$ That is true, but i was speaking just in general terms. 1/10th rule is not that a good rule. Maybe shouldn't have commented it in here. \$\endgroup\$ – user94729 Sep 13 '18 at 3:47
  • \$\begingroup\$ @Mikey, maybe I shouldn't mention the 1/10th bad rule... \$\endgroup\$ – Ale..chenski Sep 13 '18 at 3:49

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.