UHF RFID TAG Antenna Broadband Design

Hi all,i'm just simulating tag antenna by HFSS and find the BW issue.

I want to get 100M bandwidth(860~960MHz) of return loss below -10dB,but it's not easy,very hard! 860~930MHz or 900~960 is easier which means 60~70M BW. Is anybody has the same concern about it? No matter what the antenna pattern is. Thanks.



B.T.W, return loss is calculated by formular of power reflection efficiency here, CAN'T be calculated by HFSS itself because the S11 formular is wrong.

See this Sonnet application note for a 900 MHz tag layout with extended bandwidth:
https://www.sonnetsoftware.com/supp...Using_EM_Analysis_for_RFID_Antenna_Design.pdf

What do you mean by "HFSS S11 formula is wrong" ?

volker_muehlhaus said:

See this Sonnet application note for a 900 MHz tag layout with extended bandwidth:
https://www.sonnetsoftware.com/supp...Using_EM_Analysis_for_RFID_Antenna_Design.pdf

What do you mean by "HFSS S11 formula is wrong" ?

Click to expand...

volker_muehlhaus,sorry for late reply to u since my so busy work these days. I'll consider your advice in the paper which sounds good for me .

I say" HFSS S11 formula is wrong" ONLY when source and load impedances are complex in which condition S11 should be (Zs-conjg(ZL))/(Zs+ZL). For general conditon,Zs=50,ZL is complex, S11 is (Zs-ZL)/(Zs+ZL) and this is the formula in HFSS.

rf1008 said:

Hi all,i'm just simulating tag antenna by HFSS and find the BW issue.

I want to get 100M bandwidth(860~960MHz) of return loss below -10dB,but it's not easy,very hard! 860~930MHz or 900~960 is easier which means 60~70M BW. Is anybody has the same concern about it? No matter what the antenna pattern is. Thanks.



B.T.W, return loss is calculated by formular of power reflection efficiency here, CAN'T be calculated by HFSS itself because the S11 formular is wrong.

Click to expand...

I'm new in desigining antenna. As such, I would like to know how can we design antenna using HSFF? Do we just need to draw the geometry of the antenna and just let the HFSS do the simulation?

I also need to design an antenna for UHF RFID tag with brodband characteristic and it should be able to be mounted on metallic surface?

Thanks

sriza85 said:

I'm new in desigining antenna. As such, I would like to know how can we design antenna using HSFF? Do we just need to draw the geometry of the antenna and just let the HFSS do the simulation?

I also need to design an antenna for UHF RFID tag with brodband characteristic and it should be able to be mounted on metallic surface?

Thanks

Click to expand...

HFSS can do the simulation on your geometry of antenna you draw.

What's your bandwidth requirement for your metalli tag?

rf1008 said:

HFSS can do the simulation on your geometry of antenna you draw.

What's your bandwidth requirement for your metalli tag?

Click to expand...

I would like the antenna to cover all the UHF range (860-960 MHz) and the aim is to make it as compact as possible since I know there is always trade off between volume and the bandwidth.

Will HFSS be able to simulate the condition of the antenna mounted on the metallic surface?

Currently, I'm using Sonnet Lite and only be able to create a simple dipole for UHF RFID. I'm not sure how to put the substrate and the ground plane for my patch antenna. I can only draw the metal patch.

TQ

I've not designed or simulated metallic tag antenna,but by my view point, 100M bandwidth is very big.

You'd better read some examples and manual, HFSS can creat antenna model,substrate,ground, port,etc you want.It's good for 3D construction.

True, 100 MHz bandwidth is quite big. I found that there are lots research papers of patch antenna design that can achieve this broadband characteristic but most of them consider 3 dB return loss not 10 dB. From your opinion, should 3 dB RL suffice for UHF RFID tag?

Are you working in the industry or academic field?

sriza85 said:

True, 100 MHz bandwidth is quite big. I found that there are lots research papers of patch antenna design that can achieve this broadband characteristic but most of them consider 3 dB return loss not 10 dB. From your opinion, should 3 dB RL suffice for UHF RFID tag?

Are you working in the industry or academic field?

Click to expand...

I'm in industry field,and i think which bandwidth is correct between 3dB and 10dB mainly based on the actual tag performance that you could accept. 3dB bandwidth may be ok for some special applications,such as metalli tag in which the read distance maybe is not the most important.
For biggest energy transfer,10dB return loss is of course the better which means the longer read distance. All must be based on your requirements.

Dear rf_1008,

I used CST MWS for RFID antennas but some years ago I intensively used HFSS for other things. As you already know, the input impedance of the chip has generally a capacitive reactance. Because of that, the antenna itself must offer an inductive input reactance.
Firts of all, in order to simulate the matching, you'd better need a complete S-matrix file (s1p) of the chip. Draw a lumped port in HFSS, in place of the chip. In this case, the impedance does not matter. Ansoft HFSS has a co-simulation interface with Ansoft Designer. If you have the possibility to use Designer, you can easily simulate the whole system. That is called "dynamic link": in the Designer schematic you insert a black box with this dynamic link and you start the simulation from the cricuit simulator. You can perform full wave parametric or optimisation analyses controlling everything from Designer. You can connect the two blocks but if you di just this you don't have the port to extract the s11. To extract the s11, place an ideal circulator (I guess HFSS has one) between the two blocks and from the middle one extract the s11 using a port. if you do cannot use designer, define the lumped port to have the impedance of the chip at 896MHz and after you have matched the antenna at 896MHz export the s-matrix as s1p and perform post processing in any other program (ADS, AWR or also in MATLAB) and calculate s11 referred to the correct chip impedances in the desired frequency range.

As for bandwidth enhancement there are some techniques, as far as I know. If you have an IEEE access you can download some papers inserting proper key words. I know tehcniques using some "defected" dipoles: some small holes close to the feed can be used to alterate the input reactance. Furthermore, you can use inductive coupled feeds if you need a very high inductance.
Another thing one can do is to design the dipole off-resonance: you will get a slow slope in the real and imaginary part, so that it is better to get good matching at frequencies around the 896MHz central frequency.
Also the type of the dipole can help: for example you can use a bow-tie structure to get some benefit in therms of bandwidth.

As for the bandwidth, I guess 100MHz is not necessary. 40MHz or 50MHz (6%) is still very good at this frequencies. The application itself is normally not broadband.

I hope this can help,
Ivan

- - - Updated - - -

Sriza85,
3dB means that 50% of the power is reflected... I am very doubtful about that... Indeed, the signal encounters a first reflection when received from reader to tag and again to the path back. One has a very bad radiation efficiency in this case. In the path reader-tag, for example at the minimum far-field distance (if the antenna is 8cm-long, that's about 40cm), you have an additional path loss of more than 3dB. That means that the bidirectional attenuation due to path loss and mismatch will be at least 12dB. At a bigger distance of course you get lower power. Anyway, it is always dependent on the application and power requirement at the reader. The antenna will still probably work at relatively short distances. Anyway, I always prefer to define the bandwidth referring to the -10dB level.

I.

to Ivan, Say thanks firstly to you for the detailed thinkings you obtained these years. and I'm glad to discuss sth. or consult sth.from you if permitted.

1)I've been using HFSS for yrs for RFID antenna designs,and just begin to learn CST because i want to compare the two results, i think i'll get the comparison recently. By your point of view, which software do think has more accurate simulation to the real impedance performance of the antenna itself?
2)I simulate the rfid antenna mainly focus on the imput impedance which is the most important for RFID tag design because the two impedances of tag antenna and chip are complex and differ very much to 50 Ohm. Of course other performance,gain,efficiency are still be considered.
3)I will not define the HFSS port impedance as chip impedance and leave it to be 50Ohm. And I'll calculate the power reflection coefficient manully based on the antenna impedance from simulation. Matlab of course will do it for me.
4)For designer,i haven't used yet,maybe by your word,i'll try to learn it.
5)I've been trying to measuring the impedance of tag antenna,as you know it is balanced for most cases. Do you have any advice on differential impedance measurement on tag antenna? thanks.

Dear rf1008,

1) The CST transient-solver is extremely faster than the finite element if the structure does not have multiple resonances. For the design of the UHF dipole it is perfect and the speed comparad to a frequency solver is considerably better. If you start designing RFID coils at 13.56MHz you need a finite element solver (in freq. domain) as the time domain solver of CST (based on FDTD) does not meet the convergence criteria. But at 900 MHz I found CST very efficient and accurate. You can nalso compare the two CST solvers (time and freq.) with the HFSS solver (freq.) to get a feeling about accuracy and design speed. Summarizing, for the RFID design I would do this:

-->Coils @ 13.56MHz --> FEM solver
-->Dipoles @ 896MHz --> Transient solver

4) Designer allows you to have a dynamic link and calculate automatically the actual s11. Otherwise you could export each s-matrix and do it manually as post processing but you need anyway to save manually the s-matrices as s1p...

5) Design a broadband balun or; use the following approaches:

- **broken link removed**
- http://publik.tuwien.ac.at/files/PubDat_187024.pdf
- Another technique that I saw is to divide the dipole in two and connect its half orthogonally to a metal ground plane. In this way, you apply the symmetry condition of E-type. Then connect a coax connector to the dipole feed and the coax ground to the ground plane and measure. I was trying to locate the reference paper but I could not manage to find it. Maybe you can search for it.

Regards,
I.

10dB return loss is common for defining antenna bandwidth, like dipole-type tag antennas. However, for patch-type tag antennas (like those used for metallic tags), 3dB bandwidth definitation does exist and this definitation makes sense for two reasons. 1) Patch antennas own a significantly larger Q than dipoles. 2) Patch antennas deliver a larger gain which compensates for the poorer return loss to some extent.

Obviously 10dB return loss provides a better power transfer than the 3dB case. However, the practical power transfer coefficient depends on the environment a lot. Thus extremely good matching like 10dB return loss may be so sensitive that it is easily to be lost in real-world applications. It is found that 6dB return loss may be more easily and consistent to be expected in practice, while still maintains most of the read range.

BTW, return loss with respect to the chip impedance or power transfer coefficient can be calculated by HFSS automatically after defining variables and inputting the relevant formulas.

rf1008 said:

I'm in industry field,and i think which bandwidth is correct between 3dB and 10dB mainly based on the actual tag performance that you could accept. 3dB bandwidth may be ok for some special applications,such as metalli tag in which the read distance maybe is not the most important.
For biggest energy transfer,10dB return loss is of course the better which means the longer read distance. All must be based on your requirements.

Click to expand...

Dear Ivan,
Thanks for your advice on impedance measurement,but your papers which attached I've read already, and tried and tried to do the measurement step by step according to the article.Maybe sth. wrong i made,or sth. undisclosed in the paper which result to the measurement has much difference to simulation.I found the calibration has some issue for two-port calibration,S21 and S12 have much phase variations.

Also,the senond paper advised one method for tag chip impedance measurement ONLY for single-end tag chip, is it suitable for differential tag chip, i think it's not.

Glad to discuss with you.

- - - Updated - - -

I measured my antenna in lab circumstance NOT the anechoic chamber, does is affect the measurement accuration very much. I doubt.

Ivan,

Can you explain what is the different between return loss bandwidth and axial ratio bandwidth?

- - - Updated - - -

Ivan,

Can you explain what is the different between return loss bandwidth and axial ratio bandwidth?

sriza85,
axial ratio AR gives a description of the polarization of the antenna, in particular between the two orthogonal field components in the far field. Now, for some antennas it is important that it is very low or below some thresholds within a relative large bandwidth. For example take a a circularly polarized antenna that must mantain its field polarization (circular and namely ideally Ex~=Ey or Etheta~=Ephi -->AR ~= 1). In this case you define the bandwidth as the range in wich the ratio of the two field components is close to 1 and the antenna mantains a good circular polarization.

The return loss bandwidth gives a measure of the frequency range in which the reflections due to impedance discuntinuity at the antenna input (or output) port are sufficently low to assume that a relevant amount of power is delivered to the antenna itself.

Whats the cost of this tag