Hi all,
i have to directly interconnect a CC2500 and a CC2591.
In CC2591 datasheet seems that no matching components are required between the CC2500 output and the CC2591 input.
This sounds strange to me because in the CC2530+CC2591 reference design there is a matching network between the two ICs and the output impedance of CC2530 is similar to the output impedance of CC2500.
Furthermore in the CC2591 datasheet there is no data about its input impedance (i suppose 200 Ohm balanced, looking at the datasheet of the balun used in the CC2591 ref design).
In this moment i am trying to evaluate the input impedance of the CC2591 by reverse engeneering the CC2530+CC2591 reference design.
Any help is appreciated.
Thanks in advance.
Best regards
Oscar
Oscar,
this is RF not circuit analysis. Since the traces are not multiples of 1/4 wavelength they have an impedance (and some delay). so if you don't include this in your analysis then you will never undestand how this works. As example I've attached a balun design where we have cacluated each traces impedance and suum them with the component values. Remember we are trying to a maximize the power to the load which is done by doing impedance matching.
3463.RF Matching & Layout.docx
Rdgs
RRS
LPW Support Group
You re right, that's because I have done again matching for the CC2500 as I am using a 0.175 um FR4 substrate, obtaining 50 Ohm matching.
I have uploaded my ADS project.
Could you kindly give it a look?
1030.cc2500.zip
Oscor,
The layout schematic layout is correctly set up. The board vendors we use Sunstone and Advance Circuits all quote 4.11 for permittivity at 2.45GHz. So I'm not sure where you got your number it seems a little high. I use ADS and try to keep my impedance matching to less than 1.2 VSWR so yours looks good. One thing we learned to do is parallel the murata parts with ADS model which allows you to parametize the tolerance. Murata has a generic part at the very end of each series but we have not been able make it work.
Using a standard idea part in parallel with murata lets you do a worst case morlo carlo, which we seen about worst case about 1.5 VSWR. Also check you forward transfer S21 to make sure the -3dB point is good and the component in the passband are well behaved. One problem we noted is the self resonance of the inductors can change them to capactor and the 8th and 10th harmonics might not be attenuated properly. This is usually the results of the inductor self resonance moving lower in frequency vs the specs due to longer return paths for inductors. Keeping this short is the key to maintain the component vendors self-resonance specs.
I'm confused in I thought you were connecting the CC2500 to the CC2591. I would thought you connect differentially to the CC2591. The challenge when doing this is much more involved. Because the CC2591 is working on 3.3Volts the dc bias must be centered so the output swing does hit the rails . Since at 20dBM your peak to peak voltage is very close. It actually goes into a slight compression. If it is not a symetrical swing then unwanted distorion appears, resulting in more filtering. The simulator when simulating with s-parameters is really a small signal simulation and will not tell you the complete answer. Generally what is done is load pull at the output to check for stability at maximum power. If the load is not 50 ohms it can cause stability and distortion.
I'm sorry we don't have more models developed that would help you hear. ADS has capabilities to simulate but first creating a model and validatiing that it shows the true results is a good idea. Check TI web site for our latest combo layout. You can easily import the Gerber files and populate the components (murata) to simulate through s-parameter matching.
As you model PCB antennas the setup is done totally different and uses FEM. I can provide a typical file for our inverted F once you get to this point in your design.
Cheers,
Very useful answer! Thanks!
Anyway, the permittivity value comes from Reference Design CC2530-CC2591. In the readme.txt file you will find:
PCB DESCRIPTION:4 LAYER PCB 1.6 MM Copper 1 35 um Dielectric 1-2 0.175 mm (e.g. 1x Prepreg 7628 AT05 47% Resin) Copper 2 18 um Dielectric 2-3 1.14 mm (6x 7628M 43% Resin) Copper 3 18 um Dielectric 3-4 0.175 mm (e.g. 1x Prepreg 7628 AT05 47% Resin) Copper 4 35 um DE104iML or equivalent substrate (Resin contents around 45%, which gives Er=4.42@2.4GHz, TanD=0.016).
I will use the same stack up / material.
And yes, the best way to match CC2500 with CC2591 is differential matching but i don't know the input impedance of CC2591. So in the end I have used this approach:
1) Output matching of CC2500 towards a 50 Ohm load (as you saw in my ADS project)
2) Then use the balun as in the CC2590/91 ref design to match CC2591 input.
3) Output matching of CC2591 exactly as in the CC2530-CC2591 ref design. I just copy paste the trace because my board stack is the same as this reference.
Not the best approach, but I think it should works! ( I hope!!!)
We work directly from the Oslo group which created that document. The permitivity varies from board vendor to vendor. In the US most of the vendors quote a much lower value. The European board vendors quote higher. This is because they use different preg material for the FR4 core. We validate the numbers from the US manufactures and found them to be true. I'm not sure this has been done for the European board vendors. I would suggest you contact the board vendor you are planning to use and ask them specifically what the permittivity is at 2.45GHz. They will try and give you a standard answer of what it is at 1.0GHz but it scales with frequency. It is not the same at 500MHz or 2.5GHz. It doesn't create a big error but helps to center your matching circuit so production variation is minimal, therefore higher yeilds.
Rgds,
If we have to do the load pull measurements for calculating the differential input impedance.
In the following link
http://ecee.colorado.edu/~ecen5014/8-Load%20Pull%20Characterization.pdf
the optimal input impedance for the cc2591 is measured 55 ohm.
If this the correct value?? Beacause in the previous slide the value mentioned is around 100 ohm.
Prem,
We use Murray Electronic load pull and for the CC2591 the differntial input impedance is 105+j35 the single ended output impedance is 33-j6. This is repeatable in out setup and is the satistical average of many lots of parts so it would include the variation. It seems you single piece measurement is in order. The only difference is we have chosen the average values.
Thanks for the information.
Respectfully,
hello,
I want to know the output impedence and input impedence of CC2530. I also want to know the output impedence and input impedence of CC2520.I have looked for many resource.But I can not find the value.Can you tell me?
Thanks.
stones yang
Stones,
The load impedance for the Cc2530 is 69+j29.
The load impedance needs to be 80+j*74.
Regards,
Hi
Above the above post, can you advice which RF application you use to calculate the impedance matching?
Thank you
-Nissim@vitelix.com
- nissim@usbphone.com USB phone inventor Skype: nissim.test