output impedence of cc2591

Necdet, 

Please follow our reference designs closely, however I can tell you that the output impedance is very close to 25Ohms with not imaginary component to speak of.

Regards,
/TA 

Thanks TA,

We are also in need of s-parameters. Could you share that? I have read in the forum that 2 layer PCB designs with cc2591 may be problematic and may even reduce the performance of the radio module. Is it true?

Regards,

Necdet

Hi Necdet 

You could check
https://e2e.ti.com/support/low_power_rf/int-low_power_rf/f/162/t/128791.aspx
It might provide additional insight into CC2591 design and operation.

 Tor-Inge

Hi Tor-inge,

This link appears to be broken (opens "Group Not Found" page), can you please provide

another link.

Hi Igor
Sorry, my mistake. The link is to an internal TI support forum. I have added the essence of the thread, below. 

Hello,
A customer has requested S11 data for the CC2591 in order to design impedance-matching circuitry between the antenna and PA/LNA. They have also asked to which points the S11 data has been de-imbedded (i.e., pin, pad, chip, etc).

The CC2591 is in TX a highly non linear device hence S-Parameters are not accurately representing the device.
The CC2530_CC2591EM reference design is developed using load/source pull techniques while continuously monitoring overall performance such as output power, current consumption, sensitivity, harmonic content, spurious emissions and more. The reference design is the circuitry that currently is identified to provide the best compromise of these parameters.
In the CC2530_CC2591EM rev 3.2.1 reference design, the effective load impedance at CC2591 ANT pin is approximately 13Ω at 2440MHz The impedance is measured by removing CC2591 and measuring the impedance on solder pad CC2591pin 11. Note, this impedance is not necessary the only optimal impedance as impedance on pin AVDD_PA2, AVDD_LNA and others also influence the effective load/source impedance in normal operation. This means that i.e. changed AVDD_PA2 impedance can generate different optimal ANT pin impedance measured on pin 11.
Remember that impedances at other frequencies also influence circuit operation. Especially the filtering characteristic holds dramatic influence on overall circuit performance.
The differential impedance between RF_N and RF_P when CC2591 is terminated as in reference design is approximately (105+J30)Ω at 2440MHz
Optimal CC2530 load impedance for CC2530 is approximately (69+J29)Ω
Given the complexity of dependencies in CC2591 designs we recommend copying a reference design as closely as possible to minimize the number of design iterations necessary to get acceptable overall performance. 
Tor-Inge 

Thanks Tor!
I will forward this info along with AN086, AN068, and the CC2530-CC2591EM Ref Design. Also, while we will definitely suggest that the customer follow the reference design as closely as possible, making modifications only to accommodate changes in board thickness per AN068, etc., do we have some load-pull and source-pull data (e.g., in the form of a Smith Chart, etc) that we can also share with the customer since that said that they would also like to run simulations if possible?
Regards,

Hello Tor,
The customer has built a simulation model and has asked if we can confirm the complex impedance that should be seen on pin-11 looking back into the CC2591 and with the antenna removed. They are simulating 28+j13, and have asked if we can confirm that they are close or need more work. In your previous email you had said that the optimal load impedance at the CC2530 was 69+j29. Similarly, what would the optimal complex impedance be at the CC2591 output?
Regards,

Measuring the impedance on p11 into CC2591 will as previously mentioned depend upon the impedances connected to CC2591 on AVDD_PA2 and AVDD_LNA and will hence hold some variation from design to design. The optimal load on CC2591 p11 (ANT) is measured to approximately 13ohm, no significant imaginary part. The impedance is measured at CC2591 pin11 on the CC2530_CC2591EM rev 3.2.1, CC2591 removed and the RF signal routed to SMA (through 0ohm resistor) and the SMA terminated with 50ohm. 
On other designs with different impedances connected to AVDD_PA2 and AVDD_LNA this impedance might be different we have seen load impedance values up to around 30ohm generally with a small inductive imaginary portion. Based on this a load impedance of (28+J13)ohm seem a realistic value.
Measuring impedance into pin 11 will provide varying results dependent upon CC2591 mode (TX, RX, PD), supply voltage, measurement power (network analyzer test signal amplitude) and AVDD_PA2 and AVDD_LNA impedance.
Tor-Inge

Can you confirm the final number of cc2591 impedance ? 

 The output impedance of the CC2591 best load match is at 33+j7.   CC2591  the differential input impedance is 105+j35 (RF_N RF_P)

Thank you

Nissim@vitelix.com

Hi

Removing CC2591 from CC2591EM, measuring impedance presented on PIN ANT result in close to 33+J7 ohm.

On CC2591EM the differential impedance between RF_N and RF_P is close to 105+j35 when CC2591 in TX and ANT connected to 33+J7 (CC2591 re-soldered onto the board.)

Tor-Inge

Hi

Thank you for the help. 

In the "CC2530_CC2591EM reference design" the input to cc2591 the "Complex Impedance Matching Networks", is very long. see photo 1. And has 7 component's between cc2530 and cc2591. that may be good for reference design, but in the real world , it take too big PCB space, and  make it near imposable to use the cc2591 if this PCB space must be wested. 

In demo reference design cc2430 and cc2591 it use only one coil, see photo 2. Can we use this reference design? and we can use only one coil between cc2530 and cc2591?

I read in some RF guide paper, that : "If Traces Shorter than 1/20th λ Long, Zo Matching is usually Not Important. "

What I missed ? 

Photo 1: CC2530_CC2591EM reference design                             

 Photo 2: cc2430 and cc2591  reference design

Nizim,

I see you point, however the output stage of the CC2430 and CC2530 are so different that the same match could not be re-used. Here is a layout study we did not show that a compact layout can be acheived with a CC2530+CC2591 device. The match between the CC2520 and CC2530 is the same.

Regards,
/TA

Hello TA12012, 

Thank you for the excelant PCB space saving example. 

Can you share with me also the Gerber file of this design with it's PCB stack up, PCB material (Dielectric Constant, loss Tangent, Copper Conductivity), so we can imitate that 100% to get same RF result ?  (maye also RF tests also ...)

Thank you

-Nissim@vitelix.com

Hi everyone!

I would like to build wireless project use cc2500+cc2591.I have known output impedance (RF-N -- RF-P) of cc2500 is 80+j74 and input impedance of cc2591 is 105+j35. So need to matching them.Use the Smith chart, i build matching circuit as the picture. I got VSWR for cc2500+cc2590: 1.14, VSWR in output is 1.07. IS this working good and properly?

I would highly recommend you to use the CC2592 (http://www.ti.com/product/cc2592) instead of the CC2591. It is easier to work with and will give you better performance.

CC2591 is not NRND but we recommend CC2592 for new designs since this is a improved version of CC2592. Could you repost on the internal forum with more details about what their problem is?

Their problem is that they are way behind schedule, their prototype (having many flaws) based on '2591 is badly mistuned, and they have little RF expertise or equipment.  They are renting equipment and I'm training them how to use it, but I think we would all agree it would be much easier to use '2592. 

Not sure what else to add, if I need more help I'll contact you internally.  If you are still interested, look me up.

Best, S. Dunbar, AFA Colorado