matching of CC2591 + CSR chip for class1

Naoki Kimura, 

When you say "loss" what does that mean. Are you talking about TX output power or LNA gain? Also can you share the schematic, or at least the portion of the schematic pertains to the CC2591.

Regards,
/TA 

TA san,

Thank you for the comment.

"loss" mean, I'm talking about TX output power at BPF (FL1) out.

and currently schematics are as attached.

L408 and L409 are reserved for matching with CSR BC5,

If some of these value (L or C) makes matching better.

Thank you.

- The output impedance of the CC2591 change when the impedance on the input change.

- The impedance on the AVDD_LNA, AVDD_PA1 and AVDD_PA2 lines are important

- You need to run load-pull at the CSR output to know the optimal load impedance. Second guess is that the output is 50 ohm and you can tune from 50 ohm to CC2591 required impedance.

- Please check the connection to pin 3.

To make a board with CC2591 and a non-TI radio require good RF knowledge and RF lab.

TER san,

Thank you for the advice.

1. Impedance on the AVDDs.

 Yes, They are.

 We had try to modify PCB rayout to improve, already.

 In reference to your application document.

2. CSR output impedance.

 Acaccording to the cut and error we tried, it seems not 50 ohm.

 And, on the CSR's application, it recommend to use balun which special tuned for CSR chip.

 the characteristic is different with general used 50 ohm. It means that impedance of BC5 is not 50 Ω.

 So this conditions makes the matching difficult.

 And,  Asked about impedance with CSR,

 any sales and FAE answer that they can not disclose it. It is confidential sure.

3. check the connection to pin 3.

 What kind of measurement/checking can find a problem and improve the characteristic?

 please advice, thank you.

Do you know the example that CC2591 connect with CSR BC5, in your customer's case?

Thanks and Best reagrds.

Kimura,

I don't know about any designs which uses  CC2591 with a non-TI chip. This is normally fairly complex and the support we can provide on this is very limited. Try CSR with balun to 50 ohm and then 50 ohm to the CC2591 input.

TER said:

I don't know about any designs which uses  CC2591 with a non-TI chip. This is normally fairly complex and the support we can provide on this is very limited. Try CSR with balun to 50 ohm and then 50 ohm to the CC2591 input.

I too am trying to interface a TI front end (CC2590) to a non-TI chip and am planning to use a balun into the CC2590 input.

The datasheets for the CC2590 and CC2591 show a Murata LDB182G4520C-110 on the evaluation module. This is a 50 Ohm unbalanced to 200 Ohm balanced part which leaves me a little confused because, from everything I've read on here, it seems the CC2590 input impedance is close to 100 Ohms differential.

Can you confirm that a 50 Ohm unbalanced to 200 Ohm balanced is the right balun to use? or would a 50 Ohm to 100 Ohm provide a better match?

I recommend follow an existing and verified design in detail, if you not is prepared to measure/design your own impedance matching circuit. My experice is that it is too easy to else fail and more or less not getting any system improvement with circuits like CC2590-91.
I design impedance matching circuits including CC2590 regularly, maybe 4-5 times each year. Sometimes feed by CC2500 and sometimes something non TI. It is always nice to know manufacturer measured impedance for these chips but at these relative high frequencies plays also PCB and transmission lines a big role for total impedance. Even if transmission lines are well designed do they add a delay and chip impedance is then of little value for calculating an impedance matching network. Measure with a VNA, via a differential probe at point for the network both against CC2590 and against the source chip. From that is a L/C network calculated with aid of an automatic impedance optimizing software (AnTune) which takes account of L&C S-parameters.
Often forgotten are taking in account importance of matching or mismatching for harmonics. It can be of very high importance for some  types of PA stages behind CC2590 to get high efficiency and low harmonic levels.

For new designs look into http://www.ti.com/product/cc2592 which is easier to use than CC2592.