I am designing a bluetooth low energy device. It basically reads a sensor over I2C and collect data. You can check data in your phone whenever you want. I have two questions, one related with antennas and another one related with battery.

first of all, this is my layout so far

enter image description here

It is based in this schematic

enter image description here

IC2 is B1 from schematic, IC3 is X1. The left part is my specific sensor.

My questions are: First, the one related with antennas. Since this is a homemade project I do not have a vector network analyzer so I can't tune the antenna. However, I found this antenna from Texas Instruments:


They claim its impedance is 50 ohm, so could I just attach its 50 ohm feed point to my 50 ohm output from the balun? with a extremly short wire so I don't add extra resistance or inductance (or as low as possible)

Second question is I need it to be battery powered and small. I went for a CR2032 battery. However I checked its current specifications and it says 4mA maximum discharge current. I need around 20mA in some moments when the radio is working. I thought I could add a large capacitor (C14) so it can provide higher current for a short time and then it will be charged again when the system is sleeping. It is possible?

Thank you.


I added a pi-network just in case. My layout is this now

enter image description here

and this is the schematic

enter image description here

I'll use a 0 ohm resistor as R3. R4 and R5 are not mounted initially. This would allow to eventually tune the antenna, but it won't be easy since I don't have lab equipment. I am wondering if this would work without tuning?

  • 1
    \$\begingroup\$ Is this for a product? If so, using a separate antenna (vs. an FCC-certified module with an antenna already included) could result in 10's of thousands of dollars in FCC testing. Really. \$\endgroup\$ – tcrosley Feb 2 '15 at 16:24
  • \$\begingroup\$ Yes, it is for a product. We are considering it too. We are expecting profits above ceritifcation cost. however it could be a good way to start the project. \$\endgroup\$ – zapeitor Feb 2 '15 at 17:29
  • \$\begingroup\$ ok, as long as you are aware of the costs. A company I contracted with spent over $100,000 testing a PCB with three radios on it, the testing had to insure the radios didn't interfere with each other as well as not generating any harmful inference to others. For a new project I am working on, I am using a BLE module with an antenna already certified. \$\endgroup\$ – tcrosley Feb 2 '15 at 17:45
  • \$\begingroup\$ wow, that's a lot of money. Well, I hope it's less if you only have one radio. I am looking for modules, but in my product, cost is key, so those bluegiga are not an option. I am checking some chinese options. Anyways, do you know if what I'm trying to do would work? Even if it's only for educational purposes. \$\endgroup\$ – zapeitor Feb 2 '15 at 18:38
  • \$\begingroup\$ I updated my answer to address your current design. \$\endgroup\$ – tcrosley Feb 2 '15 at 20:04

First, referring to your existing design, the important thing about a PCB antenna is to have no ground plane under the entire antenna, and to follow the layout shown in TI's application note exactly. It's not a bad idea to include the Pi network, it may be useful for during RF testing.

You should be able to continue the 50 ohm feed for a little ways, as long as it is surrounded on both sides by a ground place as shown in the app note.

Re the 2032 battery, they are designed for a pulse drain of 15 mA, although they can go higher, say 30 mA, with a minor (9%) drop in capacity. So you shouldn't need a large cap (although it wouldn't hurt).

However I suggest switching over to a BLE module that has a built-in antenna, like the BLE113 from Bluegiga. It includes an 8051 which you can use to run your own application code, and has SPI, I2C, UART, and 12-bit ADC interfaces.

Even though the maximum transmit power is 14.3 mA, this is for a very brief time, and the BLE113 is specifically designed to work with a CR2032 battery -- in fact their development board includes a holder for a CR2023 for testing that configuration.

  • \$\begingroup\$ Hi, like I answered above, unfortunatly my cost requeriments are tight and Bluegiga modules are not an option. However, I am checking other ones. Thank you for your answer. \$\endgroup\$ – zapeitor Feb 2 '15 at 19:11
  • \$\begingroup\$ Thank you for this update. I'll finally use a module but I'll fabricate this board too for educational purposes. \$\endgroup\$ – zapeitor Feb 2 '15 at 21:07

To avoid the high cost of FCC testing when you do your own antenna, try to use a pre-certified module if you can. Check out the PTR5518 module if you like using the nRF51822 (it's a great chip). It's FCC certified, pretty tiny and quite cheap. You can avoid a lot of the expensive FCC testing with it as tcrosley said. I'd recommend the nRF51822 chip and modules based on it rather than Bluegiga or TI's CC2540. The 32-bit ARM Cortex-M0 core is much better and can be programmed with just GCC. TI and Bluegiga (which is based on the TI chip) require spending a lot of money on IAR's compiler and IDE for Bluetooth functionality.

It's made by a Chinese company so the best place to find the chips is on places like Aliexpress.com (for one-off samples at retail prices, around $8 to $11) but hopefully your product gets into higher quantities, then you can often negotiate better rates with those same distributors on Alibaba or through the company that makes the PTR55x8 modules themselves.

Nordic also lists them on their partners page so they might have contact info for reps too if you get in touch with them. They also list other companies who make modules based on their nRF51822 chips if the PTR one doesn't work out for you. Wouldn't hurt to contact a bunch of them and best of all, your code should work regardless of which module you choose, maybe just changing some GPIO's based on what works best for your board.

Good luck!

  • \$\begingroup\$ Thank you! Actually I just requested for a quote of that module in Alibaba. Hopefully for my quantity requeriments they have a good price. About the Nordic chip instead the one from Texas, that's exactly why I choosed it. Also besides its ARM cortex core, it's better in power management, and this is also important in my device. \$\endgroup\$ – zapeitor Feb 2 '15 at 21:00
  • \$\begingroup\$ Also, although your answer was really helpful for my project, the answer above actually answered my question about my circuit layout and schematic. I'll choose it as the accepted answer \$\endgroup\$ – zapeitor Feb 2 '15 at 21:02
  • \$\begingroup\$ Of course, that's what the question was about anyway. Tcrosley knows his RF stuff pretty well! \$\endgroup\$ – nemik Feb 2 '15 at 21:06

Nordic Semiconductor supplies example layouts (including with PCB antennas). For best results, start with their example and modify it as little as possible.


I see you are using the balun from Johansson Technology. Request a reference board layout from them for a balun and chip antenna (also one of theirs) and copy it exactly. I can't emphasize this enough; it's not "approximately 50ohm" but "you're off by a few mil on one trace and nothing works" lol. Nordic also has a couple of layouts you can copy, some with balun and trace antenna and some with discrete matching network. Trace antenna is cheaper, balun is also cheaper than discretes in mass production (but somewhat of a pain to hand solder).

In my experience pi networks are a waste of time. If you do copy a reference design, it will work well enough, and you won't be able to improve it by swapping components blindly (without a vector analyzer). "Well enough" in this context is maybe within 5dB of the theoretical best performance, or 3dB if you can control the board impedance (material, thickness) more precisely. This is not too far off from the variation between production units of the same design anyway. This may limit your range slightly, but will by no means prevent things from working.

Check out my comparison of BLE chips in another answer: https://electronics.stackexchange.com/a/151493/26394 Nordic was best when it came out, now there are other good options, and Nordic is no longer cheapest/easiest to program/most capable/most power efficient: there is something better in each of those categories, at least as far as volume production goes.

As others have suggested modules are a great option for smaller volume devices because of FCC.

  • \$\begingroup\$ Hi Alex. Thank you for your answer. I found a good module for my design based on the Nordic SoC. I assume you mean the DA14580 as a better device, right? Also, about my layout you are right, I'll find a reference layout from Johansson Technology. \$\endgroup\$ – zapeitor Feb 3 '15 at 9:49
  • \$\begingroup\$ @zapeitor: Indeed, DA14580 is smaller, much cheaper (helps that is has balun included) and much more power efficient, yes :) I haven't used yet but have an eye on it for a future design. Broadcom's chip (SIP version) may be the cheapest and smallest already-FCC-approved option: it is even slightly cheaper than CC2540-based modules from Shenzhen. I have used that, it is not bad to program, maybe a bit clunkier than Nordic, and support is not in the same league, but there are tons of examples that come with it. Hope that helps. \$\endgroup\$ – Alex I Feb 3 '15 at 10:53
  • \$\begingroup\$ @zapeitor: Out of curiosity, which Nordic module? Pricing? (If not confidential) \$\endgroup\$ – Alex I Feb 3 '15 at 10:53
  • \$\begingroup\$ The module is PTR5518. It's 3.5USD@3000uds. I want a Nordic chip based module because I have their development kit. However, DA14580 looks good. I'll probably use it in the future. For now I don't want to restart again with a different chip. \$\endgroup\$ – zapeitor Feb 3 '15 at 12:02

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.