Hi everybody!
I recently built a prototype using MSP430F2272, CC2480 and CC2591. The firmware is taken from the ZASA example found in the CC2480+MSP430 demo kit. Schematic was designed following exactly the indications found on this design kit and the cc2430+CC2591 demo kit.
I'm using a two layer PCB and the antenna is a dielectric chip antenna.
The device works properly, but it achieves an extremly poor range (50 meters). Can anybody help ne increasing that range please?
Thank you!
How do you trace you PCB?? The cc2591 VERY sensitive to PCB. I recommend for you take TI design for 2591 and clone it.
Be careful with T-lin . You MUST clone it. It is very important.
I clone TI design for cc2430(2480) and cc2591 and I have about 600 meters.
Jesus,
The first thing to do is to understand if it is the layout or the antenna. I suggest you connect a coaxial cable to where the signals input to the antenna feedline. Then program the radio to a continuous signal at one of the channel frequencies. Measure the output power into a spectrum analyzer. It should be close to the value you program. If not then this tells you the PCB matching is not correct on your board and the antenna is not getting the power to transmit. If you changed the board layer stacking heights from the TI EVM this will change the RF traces line impedances which results in less power getting to the antenna.
Typically a board will have an insertion loss of less than 1 dB. The chip antennas generally are not 50 ohm inputs. The chip vendors will give you a layout using a trace width and length or matching components to create the correct antenna input impedance.
A chip antenna has lower efficiency than a PCB or whip antenna. Typically its around 40 to 50 %. This means your range will drop about 50 % from the TI EVM.
Review the chip vendor layout requirements or use a VNA and measure its input impedance after careful calibration.
RRS
TI Low Power Wireless Group
LPW Support Group
Mi layer stack up is different from TI's reference design. Where can I find specs for impedance and electric length of the T-LIN's?
If you check in the .text file included in the Gerber Zap files you will find the permitivity numbers for the board used in this demo board. The most exact way to get the correct impedance is then use a line calculator and enter the information. I suggest using AWR line calculator which is accurate. With the impedance for the length, width and boad stack up of the existing EVM you adjust your line impedances to the same for different board stack up.
Mos of the RF path lines are of different impedances to help fine tune the signal path since the discretes are not fine enough resolution for low VSWR.
Rgds
Hi Jesus,
I would suggest looking at radio modules with CC2591, like the RC2400HP from Radiocrafts. It include a
CC2530 and CC2591 in a completely stable design with excellent range. Options with other antenna configuration are availeble.
If you are not an RF expert this could save you much time and effort.
BR
Ørjan
We have a design that realizes over 600m with a quarter wave whip, but only around 100-150 with a chip antenna. The reference design and layer stacks were observed very carefully and scaled accordingly. Chip antennas do not typically offer great range performance, can have very odd patterns, and are easily loaded by the host PCB.
Understanding how Small Antennas work on PCB can certainly help optimzed the range link. With so many variables to follow it can be difficult to achieve maximum range. Antennas of course come is various sizes. Since chip antennas are radiators which are reduced in size there radiation energy is reduced as compared to a whip or PCB antenna. Calculating the actual 1/4 wavelength size in air then correcting for velocity changes in PCB material requires knowledge of the antenna design and the effects of the PCB layout.
You select a antenna that has 50 ohm input yet the trace or connecting paths are not 50 ohms and loss significant power to the antenna. So if the antenna radiator is not optimized for maximum efficienncy even though the input is 50 ohms you will not transmit maximum power to the radiated fields. Example most chips antennas have to 10 to 15 +Jx values. If the lossy part of the radiator and matrerial leads to 1-2 ohms your efficiency is low to start with. Chip antennas will have only 40 to 60% peak efficiency with averages much lower.
A PCB antenna provides the designer with the ability to tune the efficiency to a high value. Then when properly matched you end up with maximum power being radiated from the antenna. As example PCB copper and FR4 material leads to approximately a 1 to 2 ohm loss. Setting the radiator to 36 ohms allows for a high efficiencent antenna of 95%. Now Friis equation starts to make sense because maximum power is being transmitted, allowing one to with a well designed ohmi-directiional pattern achieve line of sight distances.
So how does patterns change for PCB antenna. Actually the ground plane on the PCB controls what type of pattern you get for conidtions of line of sight. Granted place next to large bodies of conducting material of course changes the pattern. But most engineers do not control the design therefore end up with what they get. Not smart RF design.
One final comment. Antenna distance above ground sets the final distance. So change antennas from 1 meter to 3 meter above ground increases the link budget. Maybe this will light some ideas about designing LPW products through smart design of antennas.
RRS,
Even though your antenna theories are correct, I think its important to keep in mind that the regular engineer have a less ideal situation. The theoretical possibilities of 95% is not really useful for the practical engineer.
The physical shape of the end product might limit both antenna type/placing and will also dictate the ground plane. In addition many engineers do not have access to the newest RF equipment and their own ambionic chamber. :-)
This means that the engineer have some more basic choices to make.
a) He must get a good and stable match for the chips. (Either through buying a ready to go solution as a module, or put sin the effort him self)
b) For antenna he must look at size available and his requirement for range. For small spaces he can use a chip antenna and get 30-40% average efficiency. I have tested an CC2530 + chip antenna to 350m+ (Only 350 meter before a building ruined LOS). I have done design with CC2530+ CC2591 and gotton this efficiency. For higher preformance he could choose a quarter wave wire or external whip if the product can take the cost.
To avoid impact from his own ground -plane is to use a "external" dipole antenna(often PCB based) connected to main board with u.fl type connector.
Orjan,
I agree with your comments. My points are most designs share in nativity with respect to the antenna theory therefore don't properly trouble shoot their designs. You don't need a PHD or allot of expertise to in making a design achieve equivalent preformance. TI provides reference designs to compare and copy. If prototype preformance it should be somewhat is lower to figure out why. We have built antenna characteration that provide close measurements to a chamber. Then you have simulation programs which also show good correlation.
High speed design is becoming to undersstand.