Design Review of NRF51822-QFAA board

[tamojit]

Hello,

Greetings from India!

I would like to have my design review by your engineers or any concerned person in the related field. The design is not yet complete but the reference layout is taken from here.
I am using inverted F antenna with default passive pi-network as given in the reference layout.

I have attached the gerbers below at your perusal for review.

Please provide me with the detailed analysis as I am a beginner in the RF world. Also I do not have any spectrum analyzer to carry out any necessary testing. If possible please provide me with a correct design(preferred in EagleCAD) .

Below is the snapshot of the board designed in Eagle.

Hoping to hear from you soon.

Tracker.zip

[tamojit]

I have updated the board layout by adding a shunt capacitor. Please have look.

[AmbystomaLabs]

As mentioned earlier do not use 0603 for matching components.  At 2.4GHz, it's about 10 degrees just going from one side of the device to the other.  Matching is a pain.  Use 0402 or preferably 0201.

You failed to stitch the interior of the board.  The whole things needs to be stitched.  

You should use the Johanson matched balun for it.  Likely it will work without a lot of post matching.  Also, you get a harmonic filter in the same package and a smaller footprint than all those discretes.

You should model the antenna.  Use the free version of Sonnet.  It will easily model that antenna and the learning curve is easy. This way you can get the antenna close to 50ohm in the model and the board will come out close.

[tamojit]

AmbystomaLabs Thanks for your review!

AmbystomaLabs said:

You failed to stitch the interior of the board.  The whole things needs to be stitched.

By stitching you mean to connect the 2 layers with via, right? Both the planes are GND.
I will do it eventually as I go through the design process.

AmbystomaLabs said:

You should use the Johanson matched balun for it.

Ok. I am getting these parts from Mouser, please verify the following:
1. 2450BM14E0003T
2. Chip Antenna
This is the antenna I prefer to use along with the balun. Please verify, if I need any additional filter/matching network along with the balun.
Actually I do not prefer to use the discrete matching network. They are pain in the a*s!

AmbystomaLabs said:

You should model the antenna.  Use the free version of Sonnet.  It will easily model that antenna and the learning curve is easy. This way you can get the antenna close to 50ohm in the model and the board will come out close.

Is this step necessary if I use the chip antenna?

Waiting for your reply.

AmbystomaLabs said:

Also, you get a harmonic filter in the same package and a smaller footprint than all those discretes.

I did not understand by what you meant by harmonic filter. Please elaborate as I am a noob in RF designing.

[AmbystomaLabs]

1. Yes I meant vias on the interior.  Just because they are both ground doesn't mean they are at the same potential all the time.

2. I'm sorry I simply don't have the time right now to review individual parts.  Printed antennae like your F antenna have much better efficiency than chip antennae therefore greater range.  Chip antennae are easier to get to 50 ohm by rigidly following the design guidelines.  Take your pick.

3. If you use a chip and rigidly follow the design specs yes you will probably get something that works well without a lot of effort.

4. Just google harmonic filter.

[tamojit]

I am just worried about the inverted F antenna tuning. So can I avoid the discrete matching network and use this 2450BM14E0003T balun instead?

[tamojit]

Please help!

[AmbystomaLabs]

The balun question is moot.  You should use the balun since it will give a predictable result unlike a lumped element balun. So you should use the chip balun for either antenna.

On the F antenna, if you plan on using it you should model it or really trust the source that they modeled it. Sonnet is an easy app to use and the free version should take care of it.

If you still feel like you want a printed antenna without trusting your source for the F and modeling it yourself, then copy either the F-meander hybrid that both Nordic and TI have used. TI even has a whitepaper out on it. Then you are getting an antenna from a reliable source that if duplicated faithfully will likely work.

If none of the above options appeal to you then hire an RF engineer since you will have some matching to do with an unknown antenna and a lumped element balun.

[tamojit]

Thanks for your valuable suggestion. I just updated the board layout a bit.

I plan to use the balun, for now I am sticking with the discrete balun network provided here
In the design I just added a shunt capacitor [C25] for tuning.

The thick line in the design from the feed of the antenna to the common point of C6,L3 has a width of 1.42mm and clearance from the TOP GND plane is 0.254mm. However the bottom GND plane is present under that line.

As the board thickness will be 1.6mm,  the impedance of the line becomes 50ohm.
I got the value from Saturn PCB Toolkit.

Screenshot of Saturn Toolkit:

Am I going right ?

What else is needed to be considered?
Please help!

[tamojit]

I expected some suggestions...

[Andreas]

Hi,

 

Your design looks good now, you should be good to go.

 

Best regards,

Andreas

[tamojit]

Thanks Andreas for your reply. But I am worried about the F antenna I am using. I haven't seen any proper results or post regarding the successful use of the inverted antenna with nrf51822. 

So I again modified my design and copied the meandered antenna from the official NRF51822 EK Board. I have attached the zip provided by Nordic.
nRF51822 Evaluation Kit - Hardware files 2_0.zip
I copied the design from the PCA10001 folder.

Here goes my updated design.

Can someone help if this is good ?
Or, the F antenna is better to go with in terms of efficiency?
Please provide me corrections(if any).

Thank you in advance for taking out some time to review the design.