CC1200 - Thevenin equivalent circuit of the PA

How the CC1200 PA looks like does not matter. If you have a 50 ohm port this should see 35+j35 into the network connected to the PA pin.

Note that you should include the layout in your simulations.

Dear TER,

thank you for your answer.


> If you have a 50 ohm port this should see 35+j35 into the network connected to the PA pin.
LTSpice is not a microwave simulator so I can't set a 50 Ohm port at 868 MHz.

LTSpice has voltage sources.

Yet, 0402 packages and a compact layout (Electrical Length less than Lambda/20) allow me to simulate the DC BIAS and MATCHING NETWORKa as a lumped element model circuit.


QUESTION 1
-----------------
Is it fine if I use a sinusoidal voltage source with a 35 Ohm series resistor and 5.22 pF series capacitor in my simulation?


QUESTION 2
-----------------
If the PA of the CC1200 transceiver is a standard 50 Ohm output port, how come the datasheet requires a load imepdance of 35+j35?


QUESTION 3
-----------------
The load impedance of 35+j35 required by the datasheet refers to the whole network (DC bias + matching + antenna), that loads the PA, right?


> Note that you should include the layout in your simulations.
I agree with you.
At the moment though, I'm just looking for a first order solution therefore I can neglect parasitic and coupling effects.


Thank you for your time,
Enrico Migliore

First,
let me ask why you want to run this simulation and what you want to use the results for?

Q3: Yes. If you remove CC1200 from a EM and measure the impedance on pin 17 you should see 35+j35. In this is also the answer to Q1 and Q2: You want to measure the impedance seen into a network and at least when doing this with ADS the source impedance does not matter.

Dear TER,

> let me ask why you want to run this simulation and what you want to use the results for?
I've been designing analog and microcontroller circuits for 17 years but this is my first RF design.

I want to simulate the matching network in order to understand the weight of each component (L and C) in the (output power/ input power) transfer function.


> You want to measure the impedance seen into a network and at least when doing this with ADS the source impedance does not matter.
I agree with you.


QUESTION 1
----------------
What's the output impedance of the PA of the CC1200 transceiver?

Could you please answer this question? I need to have this answer otherwise I can't simulate the circuit.

After that I will click VERIFY ANSWER :-)

Thank you for your time,
Enrico Migliore

Dear TER,

thank you for your answer.

> What I have tried to explain to customers many times is that what impedance the PA has have
> no importance at all to simulate the network.
> Actually we don't have this number since the impedance with small signal (S parameters) will not be the same as
> the impedance when the PA gives 14 dBm. In this case the S parameters is not valid.
Ok, got it.


>The only information you need is the ideal load impedance which is found by load pull measurements.
Ok, got it.


> If you are interested in the transfer function, simulate the match and filter as any other filter you are interested
> to find the transfer function of.
Ok


QUESTION 1
----------------
Could you please tell me the output power value under which the "Optimum Load Impedance" must be 35 + j35?

The datasheet doesn't report that value.


QUESTION 2
----------------
Are available for customers the Load Pull measurements?


Thank you,
Enrico Migliore

Dear TER,

thank you for your answer.

> This is typically measured for max output power but I have checked for 10 dBm at one point and the impedance is still valid.
Ok, perfect, thank you.

QUESTION
--------------
In the Load Pull measurements what was the antenna impedance?

Thank you,
Enrico Migliore

Dear TER,

thank you for the links.

QUESTION 1
-----------------
To measure or to simulate the impedance of the filter + matching circuit I need to put an antenna at the end of the chain. Right?


QUESTION 2
-----------------
"Optimum load impedance 35 + j35" was calculated pretending that an antenna of 50 Ohm was loading the filter + matching circuit, right?


Enrico Migliore

Hi TER,

I simulated the circuit by LTSpice.

The filter and matching network, when loaded with an antenna of 50 Ohm, do show an impedance of 35 + j35 to the PA.
That confirm that my simulation is correct.


Yet, if I add an impedance of 35 -j35 to the source voltage I got the following results:

1. The power sourced by the voltage source is almost entirely real.

2. Virtually no power is dissipated by the 10 Ohm resistor of the DC bias network (3%)


Thank you for your help and for pointing out that load pull issue.

Enrico Migliore

1) Not sure if I understand the question. Conjugated match is not valid for PAs
2) The 10 ohm resistor is there to lower the DC on the PA for reliability. Not sure why you are surprised by low power dissipation in this component.

Hi TER,

I see what you're saying:
> Conjugated match is not valid for PAs

But still, if I add impedance of 35 -j35 to the source voltage, the simulation says that the power sourced is real and that means that voltage and current of the source are in phase.

> The 10 ohm resistor is there to lower the DC on the PA for reliability.
Ok, thank you.

> Not sure why you are surprised by low power dissipation in this component
As I told you, this is my first RF design.
When I first looked at the CC1200 EVM circuit I noticed the presence of a 10 Ohm bias resistor in series with the RF choke inductance and I wondered how much power it could dissipate at 868 MHz. The RF choke of course blocks RF currents from flowing into the BIAS network.

Thank you,
Enrico Migliore

Would you be able to attach your simulation? I am trying to determine the impedance at the PA as well and do not seem to get the same results.

pa_out.zip

See processors.wiki.ti.com/.../CC26xx_Optimal_Load_Impedance for why the impedance of the PA is not relevant. The PA on CC26xx and CC1200 is similar enough to make the wiki post valid for both devices (the optimal load impedance is different)

ElectroNerd: What are you trying to simulate?

I am trying to simulate the load impedance TO the PA, in which I hope to achieve the optimal 35 + j35 at 915 MHz. With the recommended circuit I get 22 + j30 @ 915 MHz by plotting "V(pa)/I(V1)". This is with the LTSpice circuit I attached.

3343.pa_out.zip

I don't have spice installed so I'm not able to see how you have run the simulation. But note that if you are using the component values from the ref design in a spice simulation you will not get the same impedance. To simulate on the match/ filter network you have to use ADS or similar to take into account the transfer function of a real component and the parasitics in the PCB.

Hi,

> I am trying to determine the impedance at the PA as well and do not seem to get the same results

I simulated the LC passive network placed between the CC1200 transceiver and a load of 50 Ohm (the antenna)

I assumed that PA was a voltage source 30 -j30.

I verified that, at 868 MHz, the LC network transforms the 50 Ohm load into a 30 + j30 load.

The simulator is LTSpice.

I grabbed the models of the components from Murata.

At 868 MHz all 0402 components can be simulated as lumped components.

If you are interested I can attach the schematics.

Best Regards,

Enrico Migliore