SparkFun Forums

[Lucien]

Hi All,

Has anyone tried to do their own layout for a nRF24L01 instead of using a pre-built module? I'm thinking of putting together a nRF24L01 and PIC16LF88 + regulator on as small a PCB as possible.

I'm wondering if there's anything really critical about the component placement and if 0402 parts are a must (RF isn't my strong suit). Also does anyone know if the choice of caps and inductors are important? I'm looking at the Panasonic microwave caps from Digikey for this.

I know Nordic has reference designs available on their website, but Gerber files only. As I'm not familiar with that, they aren't much help.

Any advice would be appreciated, thanks!

And on a slightly related note, the datasheet for the 16 MHz crystal on the Spark Fun site lists the frequency as 14.7456 MHz...

(Apologies to the mods if this should be in the PCB forum. I thought RF might be more appropiate as it's for that particular chip.)

[leon_heller]

At 2.4 GHz the RF parts and layout are very critical. Unless you know what you are doing you should follow the Nordic layout.

Leon

[Lucien]

Hi Leon,

Thanks. Actually that part I do know as I've worked on a tiny circuit for a 199 MHz signal before, and a small difference in component layout made a big difference. But for the nRF24L01, seeing as how I only have the Gerbers and pdfs to go by, how much tolerance would I have with the layout? Could I use the same relative placements with 0603s?

Also, I'm guessing that it's the antenna bits that are absolutely critical. I'm hoping I can be a bit more relaxed with the crystal layout.

[lucicop]

I think the critical aspect is that of the balun circuit. Any modification of the trace length or width would modify its characteristics (R, L, C), thus decreasing the delivered output power.

[RonnyM]

Keep traces as short as possible. Double sided board with lots of ground plane and vias. I've produced a number of variations on the theme with good results. Even some double sided photo etched boards with wire vias worked well. What is the average range people are getting with the 2.4 gig chip antennas at both ends?
Ron

[Lucien]

Thanks for the advice. Do you think I'd be safe moving the balun components up to 0603 case size? Or is 0402 a must. BTW I'm planning to use the ceramic chip antenna with this.

Is it safe to assume the other supporting components aren't that critical in size and/or layout? They're just the decoupling caps, Iref resistor and oscillator bits.

[RonnyM]

0603s are fine. The RF decoupling caps should also be very close to the case. The oscillator components can stick out a little.
Ron

[Lucien]

0603s are fine? That's good to know, thanks!

Wonder if I should push my luck and ask if 0805s are OK too...

RF decoupling caps? I was talking about the caps on Vdd. Did you mean the ones on Vdd_PA / ANT1? I thought those were part of the antenna matching design.

[teekay_tk]

i think 0805 wil be fine as well. You need to keep them close enough to the RF/IO. The 2.4 ghz has approximate wavelength of 8cm, so a 0.5mm here or there shouldnt have disastrous effects. also point to note is that sparkfun 24l01 boards do not follow the nordic reference closely design as they have tried to fit it in to an even more smaller space.

[andersonj55126]

I did a board 0.9"x0.48" using chip antenna, PIC16F690 and voltage regulator. It works but my range maybe less than some. I get about 40-50ft max.

[Lucien]

That's exactly what I was thinking of, except using the PIC16F88. Looking at the datasheets though, the 690 is smaller, may go with that. 0.9"x0.48" is impressive, I don't think mine would be that small.

And 40-50ft is pretty good...

[semicolonthree]

Would I be insane to assume I would be able to use the "toner-transfer" method to create a PCB for the nRF24L01?

[wiml]

Trace widths (and copper thickness!) start to become important at these frequencies. Every time I've done toner-transfer I've gotten sort of wobbly uneven trace widths, which works fine for my random digital stuff but might be important here.

OTOH, the usual rule of thumb is that you don't need to worry about impedance matching if your whole system is less than 1/10 the wavelength in question. Wavelength is 12.5 cm for 2.4GHz, so you're pushing things only a little ... my totally uneducated guess is, it'll work, but maybe with lower performance/range than a perfect layout. Most of the critical RF stuff is hidden inside the chip where you can't easily screw it up anyway. I've never tried something like that myself.