Balun/filter design for Atmega256RFR2

Go To Last Post
10 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,

I would be interested to test out if the balun/filter could be designed for Atmega128/256RFR2 series by using discrete components.

My plan for some future projects is to use a PIFA or other PCB antenna with a balun+filter made from discrete components. I guess this would be a quite good compromize regarding performance, component cost and board space requirements.

Is anyone aware of any example designs that could be of help? Especially for the balun design. I have seen the Atmel App Note "AT01030: Low-cost Ethernet to Wireless Gateway", but that uses a differential antenna which is WAY too large to be practical.

Last Edited: Fri. Oct 16, 2015 - 01:47 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm not aware of anyone doing this. It is not very likely that you will get a good performance with a balun on discrete components.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmm, strange. Custom made balun seems to be very popular in the industry though. For example, i think TI and Nordic Semi use them in most of their example designs for low cost 2.45 GHz wireless.

The problem here is that Johanson Balun 2450BM15A0015 costs around 0.5 USD which is a A LOT in a low cost wireless module. By choosing Atmel chips to begin with, I am already in trouble cost-wise because of the super high price of MCUs compared to the competition. If I must further add to the price by using integrated Balun, I'll just drift further and further away from the price budget...

Maybe one solution would be to come up with as small as possible differential antenna. But I have not found a way to make one small enough.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How to you say designing discrete balun is cost effective?

Maybe the components shall be less expensive when compared to Johanson Balun.

But the design investment you need to develop such a discrete balun is not cheap.

Costly equipment, development time everything plays a role here.

Even after doing it you will end with a low performance balun design.

So people prefer to go for proven designs

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Please do elaborate on "low performance". How many dB do you estimate to loose with custom balun vs. integrated balun? Many projects are not critical for link budget.

If the product is sold 100k per year, and the Johanson integrated balun costs about 0.2 euros in such quantities, we are talking about 20k euros in savings IN ONE YEAR. This is not peanuts.

The design investment is there anyways if we are to design a RF PCB in the first place. We must have access to the equipment, and we must measure and adjust components like antenna shape and matching capacitors anyway. If not for any other reason, it should be done for ensuring ETSI compliance. We are only talking about some increase in development costs, but that is not a huge increase. Certainly not over the 20k euros.

If Atmel would provide us with a proven reference design for a discrete balun, it would greatly help their position against the competition.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

slintone wrote:
Please do elaborate on "low performance". How many dB do you estimate to loose with custom balun vs. integrated balun?
I don't have exact numbers.

slintone wrote:
If Atmel would provide us with a proven reference design for a discrete balun, it would greatly help their position against the competition.
I've talked to the RF engineer and he has seen some passive balun designs. In his experience they are very susceptible to variation in components and soldering process. To the point where the same resistor from different suppliers or from different batch will require new RF matching of the entire circuit.

This is the reason why we recommend using integrated baluns.

I've been working in this field since 2008 and I've never seen a customer (small or big) going with the discreet components design. As a software guy, I'm assuming there is a reason for this :)

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thank you for your valuable input, I really appreciate it. This is an interesting topic indeed.

I have been drawing, simulating and measuring more or less RF related PCBs since about 2006. So I'm used to measuring and optimizing PCB antennas, transmission lines and RF filters. But I have not yet designed a custom balun, I have only used integrated ones or reference designs.

Roughly speaking, how bad problems have you seen from the component/process variations regarding a discrete balun? Are the problems so bad that the circuit is completely unsusable without a new RF matching, or do we just lose some dBs of link budget? If it's impossible to manage, how come TI and Nordic Semi get away with it? :)

I'm still open to the idea of taking the Atmel App Note AT01030 and trying to make the antenna smaller. I assume that the discrete filter there is not as susceptible as a discrete balun.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

slintone wrote:
Roughly speaking, how bad problems have you seen from the component/process variations regarding a discrete balun? Are the problems so bad that the circuit is completely unsusable without a new RF matching, or do we just lose some dBs of link budget?
In most cases it will work, with some loses. The problem is that you never know, so each time you substitute a component, or use different supplier, you need to do the check, or acknowledge and accept the risk.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The issue with discrete baluns and manufacturing tolerances is that the tuning at 2.4GHz is critical. You only have 50MHz bandwidth on each side of the centre frequency, which is +/- 2%. Thus, if you use components with a 5% tolerance (common for capacitors and inductors, infact a typical capacitor spec is 10% tolerance) you can very easily end up with your tuning point totally outside the band. If you want high-accuracy capacitors you will likely end up paying more for them than an integrated balun, and you still have to do the design of the balun yourself.

The other thing to bear in mind is that SMD inductors use extremely fine wire, this has resistance and resistance lowers the Q (quality factor) of your circuit. Again, unless you are an expert at RF design this will ruin your day. The integrated balun has taken all of these aspects into consideration with guaranteed performance.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you can use a symmetrical antenna, you don't need the balun. Only a couple of
pi-style filters is needed then to supress the harmonics. For a really good antenna
done that way, look at appnote AVR2006. (It will probably be too large for your
design, I'm afraid, but it really reaches the 5 dBi gain claimed in the appnote,
something most other antennas don't even scratch at.)

Jörg Wunsch

Please don't send me PMs, use email if you want to approach me personally.