Rat Race coupler: Improving Isolation by mismatching?

[arefeidi]

rat-race coupler

Hi,

I've got a rat race coupler like this one for ~900MHz on microstrip line which I'm using it to isolate the transmitter signal from the received signal.

**broken link removed**

Connections are as following:
A - Transmitter
B - Antenna
C - Receiver
D - Idle Port, Matching impedance 50Ω to keep the reflection coefficient of D = 0.

Isolation between transmitter and receiver is about 40dB - though, I'd like to have a better isolation, as the received signal is very small compared to the transmitted signal (RFID).


Now I've got an idea: Wouldn't it be possible to use another load instead of 50Ω, to manually change the reflection coefficient of port D in its magnitude & phase in such a way, that the leaked signal from transmitter to the receiver input gets cancelled? In other words: Can I improve the isolation by mismatching the idle port D?

(It doesn't have to be a broadband coupler, in fact it's just used for a bandwith of less than 1 MHz)


I've already done some research in the internet, but wasn't able to find anything which said something about if it's possible (or not).

Maybe some of the experts here have an idea if I could do it this way (or even better: give me a clue how to calculate the whole thing ).

 

[Azulykit]

rat race power divider pdf

Have you considered a T/R switch or a circulator as an alternate to the hybrid?

Is the -3 dB insertion loss in the hybrid a problem? You could probably also use a Wilkinson but there would still be a -3 dB insertion loss . Have you considered the noise figure and transmitter power performance hits with the hybrid?

 

[arefeidi]

rat race coupler ads

Currently there's no circulator or switch here which I could use - also (as usual )it has to be as cheap as possible. Though, isolation of standard circulators wouldn't be much better anyway I think.

Wilkinson might be an idea, but can they really get a better isolation compared with a rat race coupler?

The 3dB loss is no problem. Noise figure etc. also seem to be no problem with the hybrid I'm using. The only prob is the isolation, which should be a bit better then the 40dB it currently has (there's a difference of 80dB between transmitter power and received power).

And this led to my idea of changing the reflection coefficient of port D (mismatching), to make the reflected signal cancel the leaked signal which goes from transmitter to receiver.

Though - I just don't know if this really is possible and woul improve the isolation. Any ideas?

 

[biff44]

rat racer rx

Is this a practical application or a student's game? As soon as you hook a real world antenna, with a 15 dB return loss of any phase to the antenna port your carefully tuned 40 dB null will degrade. So playing around further with some port D mismatch is similar folly.

Most people just put a spst switch in front of the receiver and put the switch into isolation mode while they are transmitting. If the transmit and receive go on simultaneously, then you need to filter out the transmit frequencies with a bandpass, or design the system to live with the transmitter interference. One trick is to make the transmitter a relatively "low noise" amplifier so that its broadband noise does not pollute the receive bandwidth.

 

[arefeidi]

hybid rat race coupler

Well, it's more a student's game, but with some practical background. I've now played a bit with some antennas: Didn't think they'd have such a big effect on the results - but you're right, seems it doesn't make sense to further try improving isolation that way, as it will just be impossible to get the best isolation exactly at the desired frequency.

A switch sounds like a good idea, will have a look into that. Baseband processing (transmitting/received data) is done with a DSP, so I could use the DSP to operate the switch with something like "IF (transmitter = on) THEN (switch=open)" ...but that's just an idea yet.

 

[biff44]

wilkinson divider 40db isolation

Just for yuchs, what do you think your insertion loss between transmitter and antenna, and between antenna and receiver are for your structure?

 

[arefeidi]

rat race broadband

Just for yuchs, what do you think your insertion loss between transmitter and antenna, and between antenna and receiver are for your structure?

Just a guess, but with the 3.something dB loss of the coupler I'd say about 4 dB?

 

[Azulykit]

rat-race isolation

40 dB of isolation in a hybrid is really good. Trying to go deeper would be close to pointless.

Biff44's question and your response makes this thread more understandable.

Typical good practice is to use either a circulator or a switch for this situation and I am still surprised about your comment that a 3 dB insertion loss is inconsequential.

Without knowing the details, I would anticipate that a T/R switch would be the best way to couple a transmitter and receiver to the same antenna. Mechanical switches are slow but have low loss. An electronic switch can be much faster, but with higher loss.

 

[arefeidi]

improving isolation in antenna

Typical good practice is to use either a circulator or a switch for this situation and I am still surprised about your comment that a 3 dB insertion loss is inconsequential.

Well, for the transmitter in this case it's not such a problem to double its output power. For the receiver it's more critical, ok, but the Tx/Rx isolation seems to be the bigger problem to me.

Without knowing the details, I would anticipate that a T/R switch would be the best way to couple a transmitter and receiver to the same antenna. Mechanical switches are slow but have low loss. An electronic switch can be much faster, but with higher loss.

OK, so I have two more ideas to look into now as alternatives:

- T/R switch
- "isolation switch" (as biff44 suggested)

Thanks, will do some research now first.
Any help appreciated if someone by chance knows a source where to find infos on this topic!

 

[arefeidi]

rat race isolator

Aye, don't know where my mind was yesterday - Tx and Rx happen at the same time in UHF RFID (backscattering), so a Tx/Rx switch won't help that much in this case. The isolation switch sounds promising though.

 

[Azulykit]

Can you consider separate antennas for transmit and receive? In passing, more than -40 dB isolation between closely spaced elements can also be a good trick. With care greater values can ge achieved but it is not automatic. No one likes this solution because there are now TWO antennas but if you are full duplex and need high isolation it is probably your best chance of success.

If the transmitter and receiver are not on the same transmission line then the chances of high isolation are better.

 

[arefeidi]

So far we've done it with two antennas (not really optimized though, just a few tests), but if possible, a one-antenna solution should be used. As you said, two antennas are not a prefered solution. In this case it's especially the size that matters (two antennas, space between them). It's still an option though, but only if everything else fails.

 

[yendori]

What you are considering sound similar to variable attenuators which use a coupler and PIN diodes. Basically the attenuation is the isolation of coulper which degrades or improves depending on the resistance of the PIN diodes.

Rod

 

[Azulykit]

How much isolation transmit to receive do you really need? Describe the situation in more detail and maybe we can offer reasonable suggestions.

 

[arefeidi]

PIN diode - don't know much about them yet but this might be another idea, will have a look into that.

@Azulykit:
With an antenna connected (which has about 14dB insertion loss at the desired frequency) the isolation is about 22dB (the 40dB posted before is the insertion with a 50Ω dummy load instead of the antenna).

There a two situations:
a) Transmitter sends modulated data with +33dBm
b) Transmitter sends a continous wave signal with +33dBm. The RFID-tag now "backscatters" it's data (reflecting the incoming carrier power). As it's just reflecting, the tag answer is on the same frequency as the transmitted signal by the transmitter.

The reflected signal, which is the RFID-tag-data, can (depending of the distance) be as low as -60 or even -70dB. So there's a difference of up to 100dB between transmitted and received signal.

To be able to amplify the received signal and not drive the amplifier into saturation, the isolation should be at least 40dB (better 50-60dB).

 

[Azulykit]

It sounds to me that you need on the order of 100 dB isolation. It is unlikely that you will achieve that much with a single antenna system. Get the antennas as far appart as practical and use directive beams (if You can) so the transmitter does not illumninate the receive antenna (and vice versa).

 

[biff44]

Just for yuchs, what do you think your insertion loss between transmitter and antenna, and between antenna and receiver are for your structure?

Just a guess, but with the 3.something dB loss of the coupler I'd say about 4 dB?

Well I did analyze it, and it is what we call in engineering terms a "P.O.S.". Try getting it to work as a simple power divider first, before worrying about isolation at some port!

 

[ecapso]

arefeidi,

It is possible to enhance the isolation in TX/RX applications eg RFID fMCW etc. a few examples are:-

**broken link removed**
**broken link removed**

However cancellation can add an additional 30-40 dB of isolation to a circulator or passive isolation network (eg -70dB- to -80dB depending upon tuning)

https://ieeexplore.ieee.org/Xplore/...2/04057781.pdf?tp=&isnumber=&arnumber=4057781

(Feedforward technique)
https://ieeexplore.ieee.org/Xplore/...74/00788833.pdf?tp=&isnumber=&arnumber=788833
https://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel4/75/14194/00650980.pdf?arnumber=650980

(Passive technique)
https://duepublico.uni-duisburg-ess...vlet/Derivate-14694/Final_Papers/GM0058-F.pdf (wideband)
https://v3.espacenet.com/textdoc?DB=EPODOC&IDX=GB2372902&F=0
An advantage of the passive technique is that no extra components are required.

In the case of narrow band operation very high isolation can be achieved which can extend your RFID system range considerably.

 

[arefeidi]

Hi ecapso,

thanks for your reply and all those links.

In the meantime I have tried a similar approach to the third link you posted, though with lumped elements (SMD resistor + capacitor trimmer) instead of microstrip lines to realize the mismatch.

The result looks promising with a maximum isolation of > 60 dB, though it's a very narrow solution (40dB isolation within a bandwidth of just 1.5 MHz), which means that when the antenna is moved, the capacitor trimmer has to be trimmed again. Though, as the antenna(s) probably will be fixed in the final implementation (tracking via RFID), this probably won't be such a big problem - have to do some more tests first though.

Separated antennas would give a better isolation, but if I can get it to work with one antenna, this is the preferable solution. In industrial systems they also work with one antenna systems using circulators etc. which give 30 or 40dB isolation so I guess it should be possible (Surely I won't reach similar reading ranges though).

 

[ecapso]

arefeidi,

I agree at 900Mhz bandwidth could be limited. But at higher frequencies i.e 2.45GHz this technique works well. As for range I have found that if a passive RFID chip can be powered by the radiated field it can be read. Range then becomes limited to the RFID system power output. The range of some commercial systems can be almost doubled using minor modifications.