This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

WEBENCH® Tools/CC2640R2F: Antenna Matching Circuit design for CC2640R2F

Part Number: CC2640R2F

Tool/software: WEBENCH® Design Tools

we would like to design a matching circuit to tune the antenna to perfect 50ohm for the Bluetooth to work well. we simulated the balun (attached here) and realize that the balun is unbalanced.why the balun is unbalanced? if yes, how one can get maximum output power out of the balun. i have some simulation results but will share it in private message or email.

  • Hi,

     The Balun in our Reference design is optimized to get an optimum performance. So, we recommend to use the Balun as it is. 

    C7 input/output is a 50 Ohm port. We recommend to put a PI Network in between C12 and Antenna for Antenna matching.

    Please refer to SWRA640 for Antenna matching guidelines. The following is the link.

    Thanks,

    PM

  • hi PM,

    PI network meaning impedance matching network I believe.

    I cannot see C12 in the schematics I attached. what is C12? Are u talking about C11?

    Br

    pragash

  • im simulating the balun S31 and S21. a good balun should show -3dB S21 and S31. current balun exhibits S21 of -9.6dB at 2.45GHz and S31 of -1.3dB at 2.45GHz. im using complex load impedance given in the hardware configuration manual as below. 

    "TI have found the recommended balun and matching circuits through simulations and load- and source-pull measurements over the full operational range. The RF circuits are designed to give best overall TX and RX performance (output power, sensitivity, current consumption, and harmonic and spurious emission)."

    above statement is from the hardware configuration manual. can i have the simulation results of the balun and antenna matching network by TI? this is because im using different component sizes and different PCB stack up so the component value will certainly change. 

    i suspect the balanced port impedance or differential port impedance that im using is wrong. im using 50ohm in my simulation. what is the correct differential port impedance?

    can i also request to talk to an RF designer from TI. Only a RF designer will be able to understand my questions.

  • even in the hardware configuration manual, it is requested to ask compare simulation results between TI simulation and ours in case if we use different component sizes, which we do..

  • Hi,

    It was a typo. It is C7 (12pF).

    Thanks,

    PM

  • hi PM,

    Could you please refer my follow on questions to your colleagues working in RF design? thank you.

    BR

    pragash

  • Could you share the results of your simulations and how you have done the simulation (I assume that the simulation is done with models of the components you use with the layout you use in ADS or similar) 

  • hi Ter,

    can i email it to you? could you please give me your email address?

    BR

    pragash

  • If you have confidential information to share, please send me a friend request and you will be able to send me the data in a private message.

  • hi ter,

    thanks. i just emailed you the simulation results and the spec for a good balun design. the simulation results off by far. 

    hope to hear your comments.

    BR

    pragash

  • Not fully sure if I follow what you have done/ simulated.

    If you want to simulate the balun you have to use the component models from Murata and import the layout and simulate the substrate. Then we can discuss the results. 

  • hi Ter,

    im an expereinced RF designer. based on my experience, the balun simulation results will not change that much even when i include component models and the layout. 

    i will change ideal component model with murata, colicraft and AVX model as requested by you and will compare the results of ideal vs component model.

  • hi TER,

     I have done the simulation with component models from Murata and Coilcraft even when I believe it is not going to make a huge difference. I was right. It didn't make a huge difference.

    We don't have much time to tune the antenna + balun. Could you please direct me to RF guys. They could quickly tell what when wrong with the simulation or my assumption.

    For example, the impedance of the differential arm is not 50ohm. That's what I'm expecting. Some of my assumptions is wrong. I need an RF engineer who designed the balun before to tell me what went wrong. 

    BR

    pragash

  • i sent the new simulation results in PM.

  • See the below snippets for our setup and results:

    Note that a 50 ohm termination on the antenna port is used. Not sure why you have terminated with something else. 

  • in order to answer your question, please refer to TI document as below. 

    on page 30, CC26XX has a load impedance of 42+j21ohm when biased using differential internal bias. since I'm using CC2640R2F, I believe I should use load impedance of 42+j21ohm instead of 50ohm. However, even when I use 50ohm load, the results do not change much so this is not the root cause of the issue which I'm seeing now. 

  • Another significant difference I noticed in the simulation you uploaded is there is no inductor in the middle, between two differential arms. I have highlighted the missing inductor in your simulation. 

    I quickly removed the inductor I highlighted below from my simulation to check if I'm getting the right results as you got. The answer is NO! There is still some other component value error occurred in the BOM uploaded in the reference designs. 

  • As the datasheet state: The target load impedance should be 42+j21. That means that the differential impedance seen from the RF_P/ RF_N pins into the balun should be 42+j21. 

  • Bad checking on which set-up I got from R&D: The simulation set-up I attached is for 7x7, using internal bias. 

    Could you also check the values on L3 and L5 in your setup? 

  • i tried to zoom into the component values to simulate the exact components in my simulation. However, i cannot see the component part number clearly. i need your help to upload clear snapshot of the component p/n so i can recreate the simulation here. 

    i also would like to highlight here that there is an error in the TI reference design for balun because it has middle inductor as highlighted above but your simulation which shows good results does not have the middle inductor. 

    another comment from me based on my EM simulation experience is, you don't connect the ground symbol on the layout simulation. usually, I populate ground pins on the ground layer adjacent to the signal pin which believed to be the ground return path for the signal. if you do that, the term ground should go to the layout instead of the ground symbol. 

  • I changed to the differential impedance,  RF_P =  42+j21 and RF_N  = 42+j21 and single-ended impedance to 50ohm. the results still bad so that's not the problem. 

  • all the values in my setup are attached in the schematic i uploaded. those values are from TI reference design and TI recommended values for CC2640R2F.

  • Dear TER,

    just to highlight that this issue is not resolved yet. I'm still waiting for my questions to be answered. i saw something which says "TI thinks resolved". please change this or do not take this seriously! 

  • I haven't found a ADS setup that simulate the balun using external bias so I haven't been able to find how they have terminated the RX_TX pin in the simulation. I see that you have not terminated this node with anything and since L1 is different from L5 (referring to your first post) the result will show unbalance. If you look at the reference designs that uses internal bias which is closer to your simulation setup, L1 and L5 is equal (that is the case in the simulation results I posted)  

  • TER said:
    since L1 is different from L5 (referring to your first post) the result will show unbalance. If you look at the reference designs that uses internal bias which is closer to your simulation setup, L1 and L5 is equal (that is the case in the simulation results I posted)  

    not true. i changed both L1 and L5 to be 2.4nH and 2.7nH but the results still bad. my simulation cases;

    1) L1 = L5 = 2.4nH

    2) L1 = L5 = 2.7nH

    both cases above still giving failing results for the balun.

  • TER said:
    I haven't found an ADS setup that simulates the balun using external bias, so I haven't been able to find how they have terminated the RX_TX pin in the simulation. I see that you have not terminated this node with anything

    I believe RX_TX (DC) should not affect RF/microwave simulation. Even if you don't terminate RX_TX, the balun should behave like a balun simply because balun operates at 2.45GHz whereas DC (RX_TX) is 0Hz and we don't have active devices (e.g. transistors) in the balun design. When i say "the balun should behave like a balun", I mean, it should meet all the parameters specified for a balun. 

    I still think there is a mistake in the balun design of CC26XX or the impedance of the differential arm or single-ended arm is different from what has been specified in the datasheet/documents. Could you please verify this asap. Sorry to push you but we are trying to get the bluetooth up and test the product.

  • First: What is the purpose of these simulations?

    RX_TX: First of all, this is not a DC node since it provides an external inductor for the LNA in RX when external bias is selected. And this is RF meaning that no nodes are ideal and provide a load impedance meaning that if you what to simulate with this node you should provide a load on this node to reflect the real world. 

    I sat up a new simulation (using internal bias) without the layout for simplicity.

    Note the dB(S(2,1)-dB(S(3,1)) goes down to -1.4 dB and the phase goes to 177 when the layout is included.

    Do you consider this out of spec? 

  • Question: First: What is the purpose of these simulations?

    Answer: to validate that we are getting the best Bluetooth performance available. Before matching the antenna, we usually check the balun as well. When we check the balun of C26XX, we got shocked. that's why asked for an explanation. We would like to know more because we have changed the component size to smaller than TI recommendation and due to that, used different components and layout then what TI suggested. Due to that, reoptimization or redesign of balun is an inevitable process. We have a lot of experience in RF design, but TI balun is unique. Trying to understand further to make sure we do an excellent job in optimising for our individual PCB. 

  • Question: Do you consider this out of spec? 

    Answer: I believe is not acceptable. Ideally, S21-S31 is 0dB.  anything closer to 0dB is acceptable.

  • RX_TX: First of all, this is not a DC node since it provides an external inductor for the LNA in RX when external bias is selected. And this is RF meaning that no nodes are ideal and provide a load impedance meaning that if you what to simulate with this node, you should provide a load on this node to reflect the real world. 

    My question: it's still not clear to me on the purpose of RX_TX. Based on the datasheet, I assume its C6 (please refer to schematics i attached) bypass cap for the DC supply of the on-chip LNA. Could you please explain the use of RX_TX? This is a unique balun if RX_TX is RF. We would like to know more because we have changed the component size to smaller than TI recommendation and due to that, used different components and layout then what TI suggested. Due to that, reoptimization or redesign of balun is an inevitable process. We have a lot of experience in RF design, but TI balun is unique. Trying to understand further to make sure we do an excellent job in optimising for our individual PCB. 

  • Which component size are you aiming to use? 

  • 0201. I don't understand how this question will help me to get the answer that I'm wanted. Please clarify. 

  • - Component size: I wanted to check if we have a reference design using the component  sizes you are using since the reference designs online for CC2640R2 uses 0402.

    - You say that you got shocked of the balun performance but I see that when I re-checked your results that you are using 0201 components in your simulation but you have taken the component values from a 0402 design. This, in addition to simulating without the layout will give a false result.

    - Note that that a simulation is just that, it will not take into account all effects in a real design. If you look at the results I have posted you see that the balance between RF_P and RF_N gets better when the layout is taken into account. If the balance is measured on a physical board it's even better. 

    - RX_TX: This pin is a part of the LNA design meaning that a traditional balun is combined with an external inductor giving ~1 dB better sensitivity. 

    - I haven't looked at the ADS simulations for a 0201 balun yet. The schematic snip below shows the component values used for a 0201 design:

  • hi TER,

    you are just wasting my time with you false hypothesis. if you read my post carefully, i have told you that i have done layout simulation also and the results does not differ much from schematic simulation. based on my experience, layout simulation should be very close to schematic simulation provided, good layout is designed.

    im telling you TI balun/information provided is wrong. instead of proving there is nothing wrong on TI side, you keep giving your hypothesis on what went wrong in my simulation. please read my post from the begining till the end to understand what i have done. Its really frustrating to deal with TI now a days as we dont get straight answers for our questions.

    you keep showing balun without the inductor between two differential arms. are u admitting that TI reference design is wrong?

  • If you read through what I have written I wrote that I could not find any simulation using external bias, only internal bias hence I have shown results for this.

    You claim that our balun is wrong but I'm not sure if you have simulated on our reference design since you are using 0201 and the reference design uses 0402.0201 components behave somewhat differently than 0402, at least for the smallest values since the values could have a different offset. 

    We have done a full characterization on EMs with the 0402 balun you find on the web and also the 0201 balun I posted yesterday. This includes a full automotive checkout. 

    The datasheets uses numbers from the char and therefore also the balun. Are you claiming that the datasheet numbers are no good or could have been better? 

  • If you read through what I have written I wrote that I could not find any simulation using external bias, only internal bias hence I have shown results for this.

    Reply: yes, I got that. Someone should have designed balun with internal bias. If its non-existant, does it means balun with internal bias never been simulated before?

  • TER said:
    You claim that our balun is wrong but I'm not sure if you have simulated on our reference design since you are using 0201 and the reference design uses 0402.0201 components behave somewhat differently than 0402, at least for the smallest values since the values could have a different offset. 

    i don't believe this. However, i can try this if you send me a screenshot of simulation with clear visibility of the 0402 component part number. but you will say you don't have an external bias simulation. then how can i verify this? whatever I do, you are saying you don't know how to simulate with external bias! I'm using external bias! just assumption will not help at this point. if external bias balun is not simulated and tested by TI, please declare that. 

  • TER said:

    We have done a full characterization on EMs with the 0402 balun you find on the web and also the 0201 balun I posted yesterday. This includes a full automotive checkout. 

    The datasheets uses numbers from the char and therefore, also the balun. Are you claiming that the datasheet numbers are no right or could have been better?

    Can I have balun simulation with 0402 components for external bias? If you are able to provide that, my problem is solved.

    I have sent you my simulation results it contains two sets of data;

    1) ideal component simulation

    2) simulation using 0201 component models from Murata and coilcraft.

    Both didn't meet spec. I can simulate 0402 component models and send you the results, but you are going to say TX_RX is not terminated and so forth. Moreover, I don't believe 0402 simulation is the solution because the 1) and 2) above is far off from the spec, which is unusual based on my experience. Unless this balun has different impedances as I mentioned before, I think it's not possible for the balun to good.  

  • Internal vs external bias: For this family the balun for 7x7 was developed first and this was simulated. The 7x7 does not have a RX_TX pin and hence this was not included in the simulation. When the balun for 5x5 was developed I believe the balun was adjusted slightly to adjust for the RX_TX pin using trial and error. As I understand it a sensitivity gain of < 0.5 dB compared to use a fully symmetrical balun.

    0201 vs 0402: Have you tried to simulate the same component size with 0201 and 0402 models? Simulating a 1 nH inductor gives a fairly different result for 0201 compared to 0402. 

    I have sent you component values we have used on our 0201 internal bias balun. Have you tried these values in your simulation? 

    -- 

    Given that you believe that our balun is good the method to port to a different layout/ component size would be the following:

    - Extract and EM simulate the layout used in the reference design.

    - Use 50 ohm termination on all ports 

    - SImulate S11 etc and the transfer functions S21, etc

    This is the impedances that your design should have.

    - Do the layout with the component sizes you want to use. 

    - Do EM simulation etc

    - Adjust the component values until you get the same S11, S21 etc as on our design.

    - Test in the lab. Since not all effects are possible to include in an EM simulation it's normally required to adjust the values some in the lab if you want to get the last couple of dB. This will also be dependent on what range you need in the end. 

  • TER said:
    - Use 50 ohm termination on all ports

    including RX_TX? in the hardware configuration manual, the target load impedance is stated as 45 + j43ohm. why is it stated wrong? 

  • i will do the simulation and update the results.

  • All ports to 50 ohm since this is a relative simulation. The objective is to simulate a passive network and compare with a new passive network.