Matching PCB loop antenna to Transmitter

[col_implant]

Hi,

Im working with a PCB loop antenna which has an impedance of about 0.1+j50 at 433.92 MHz. I need to match this to the MELEXIS TH72015 transmitter device.

The manufacturer does not give a definite impedance to match to, instead they mention an equation for an optimum load resistance. The equation is as follows:
Ropt = ((Vcc - Vsat)^2)/2*Pout

In my particular case, operating at 3V, and transmitting at -12dBm the optimum load resistance computes to 16200. Even using Q degrading to broaden the antennas bandwidth im finding it impossible to match anywhere close to this value (using real component values).

Im wondering is there another way to match my antenna to the device? Or is there some mistake in the method i have used?

Thanks

C

 

[mneary]

The data sheet is very confusing. I think those matching numbers assume maximum power output. Maybe if you calculate the match for "full power" but set the chip for power step 1 you would have -12 dbm yet properly matched.

 

[RadioRon]

The manufacturer provides additional application information including this note:

https://www.electro-tech-online.com/custompdfs/2008/07/0004972_AN720xx-PA-Output-Matching_rev001.pdf

It is brief, but makes clear that the equation you mention determines the best load to achieve the maximum output power from the device. If you plug in a power of 0.0063 watts (8 dBm) I'm sure that the resulting match design will also work fine when you select lower power outputs. The example given seems to match your application and they get 256 ohms.

Since the IC is rated to deliver about 8 dBm at your voltage, and you only want -8 dBm you have some leeway in how precise your match is. That is, unless you are fighting for power efficiency, in which case you should be fussy.

If you calculate a fairly high resistance (like more than 80 ohms) to present to the output transistor, then perhaps have a look at example matches from the past. Vacuum tube amplifiers typically had relatively high impedance outputs and usually used a simple PI network (shunt C, series L, shunt C) to match to low values like 50 ohms. The test circuit in part 6 of the data sheet uses just such a configuration. I'm not sure if this is appropriate when feeding a pure inductance, but give it a try. Perhaps you will end up with an L network.

 

[col_implant]

Thanks for the info. I have been through that apps note on PA Output Matching.

Are you saying that I should match to a value calculated by the equation give in that note for the highest power output. And then simply opperate at the lowest power and neglect any resulting mismatch? I thought that since the loop resistance is so small that even the slightest mismatch would result it very poor radiation?

On the transmitter side im fighting with current consumption (for the implant). However communication is very short range (1-2m) so radiation efficiency is not a huge concern.

I would sleep much easier if I could develop a matching circuit which is reasoned in theory and demonstrated in practice. It seems that this particular type of matching is something of a black art.

 

[RadioRon]

You are still thinking like an RF power amp (large signal) designer and trying to apply S parameter transmission line principles to achieve maximum power transfer. Melexis is suggesting that this approach is not appropriate. They are saying that maximum power transfer is not their goal. Instead, they are applying the basic principles of "incremental circuit analysis of a common emitter amplifier at mid frequencies" or in other words, they are simply applying the basic concepts of Norton or Thevenin equivalents, AC load line and so on, much the same as you would with a low frequency CE transistor amplifier.

The result will indeed be a large impedance mismatch between the transistor and its load, but that does not mean it won't be able to deliver 8 dBm of power into the load.

Recall that the AC load line endpoints are positioned by finding the DC quiescent operating point and passing a line through that point with a slope of -1/R where R is the AC load resistance. As they point out, to get certain amount of power to flow into a specific load resistance, you need only swing the output voltage over a sufficient range. Since the AC voltage swing range is limited by Vce and Vsat, you have to adjust the load resistance seen by the transistor to get a specific maximum power without voltage clipping occurring. In our case, that is 256 ohms for 8 dBm and 3V VCC. If you establish a load resistance of 256 ohms and then swing the voltage over a lower range than that, you will get less power into the load. That's all they are doing when they program the output power to lower values than 8 dBm using the power selection capability of the IC. The mismatch remains high in all cases.

Of course, as they point out, you can't ignore the reactance part of the output but it is relatively small and can be incorporated into the matching circuit.

Edit: a side note for those interested in such topics, many power amplifiers, especially those designed to deliver powers in the range of 0.25 watts to 5 watts, are not designed to have an output impedance of 50 ohms. Instead, they are designed to deliver their rated power into a 50 ohm load. Because the designer of the amp is usually forced to compromise between power output (gain), efficiency, and distortion the actual output impedance (S22, or the Z looking back into the amp) is rarely 50 ohms and often not even close. Of course there is an impedance mismatch seen by power travelling from the load mismatch reflection back to the PA, but ideally power travelling in this direction should be minimized by good matching of the load to the line.

 

[col_implant]

Thanks again... we are playing with Q degrading and antenna matching independantly from the transmitter at the moment. almost ready to integrate the antenna. I find that i need to match with two capacitors at the antenna input to get near a resonating 256 Ohm impedance, since our loop is so inductive.

I would like to clear up one other issue. Well two little issues... called Cm2 and Lm2, on the opposite end of the antenna. Am I correct in thinking that all the matching occurs at the input to the antenna? That is Cm1, Lm1 (Cm3 in our case)? And that the components on the other end are just to provide a bias (through Lm2) and an RF ground (through Cm2). I suppose to leave Lm2 at the same value as in the EVB, and just insure Cm2 is 2 or 3 times bigger that Cm1?

 

[RadioRon]

Thanks again... we are playing with Q degrading and antenna matching independantly from the transmitter at the moment. almost ready to integrate the antenna. I find that i need to match with two capacitors at the antenna input to get near a resonating 256 Ohm impedance, since our loop is so inductive.

I would like to clear up one other issue. Well two little issues... called Cm2 and Lm2, on the opposite end of the antenna. Am I correct in thinking that all the matching occurs at the input to the antenna? That is Cm1, Lm1 (Cm3 in our case)? And that the components on the other end are just to provide a bias (through Lm2) and an RF ground (through Cm2). I suppose to leave Lm2 at the same value as in the EVB, and just insure Cm2 is 2 or 3 times bigger that Cm1?

Well, Colm, I am going to be a bit coy here and play with words a bit, but with a point to it. Yes, in principle, you affect an impedance match by adjusting the components that are attached to the input of an antenna. However, this loop antenna is a balanced structure, with two conductors arranged in parallel to feed it. So, to be precise, we can say that the input to the antenna includes both of those conductors. So it would be logical to say that CM1, LM1, CM2 and LM2 are all attached to the input of the antenna.

It would be a lot less confusing if Melexis had designed a balanced output from their transmitter because then you would have two pins, labelled Out1 and Out2, each attached to alternate sides of the loop through an L pad of an inductor and capacitor. The schematic would probably make more sense to you then. In fact, their output is unbalanced, having only one output pin, and so they have used a matching configuration that compromises and allows an unbalanced transmitter to feed a balanced antenna structure with reasonable efficiency.

I note that the Melexis EVB, which has a loop antenna about double the size that you are working with, is matched with very different values for CM1 vs CM2 and for LM1 vs LM2. I am a bit confused by these differences and not entirely clear on what their matching procedure is. I also note that Mr. GrosJean advises to do fine tuning using CM2 and LM2. The instruction sheet for the EVB also describes all four of these parts as "impedance matching" components. So it is pretty clear that we should also consider CM2 and LM2 as useful in achieving the desired input impedance. I think that answers your original question fairly clearly.

Since your loop is so much smaller than theirs, I wonder if you can get away without adjusting the values of CM2 and LM2 as you suggest. Are these components merely to provide bias and blocking? Well, no, as we discovered above. To get more insight though, consider the impedance of each part. The capacitor, CM2 is 3.3 pF which is roughly -j111 ohms at 433 MHz. The inductor LM2 is about +j185 ohms. These values are marginally "high", especially when compared to the loop inductive reactance (I think you said it was about +j50) or CM1's impedance (about -j20 ohms) or LM1, +j 73 ohms. If LM2 were meant to be simply an RF choke for bias, its impedance would be better put at around 400 ohms or so while I would expect the CM2 impedance to be an awful lot lower if it was only an RF ground component, like close to 1 ohm. So it is safe to say that these are not simply bias and ground coupling parts as you suggest, they are relatively involved in establishing the impedance seen by pin 8 of the IC.

What to do with your smaller loop then? Well, I am not an expert at loop matching, nor am I willing to spend a few hours to analyze your configuration on a simulator, so it might be wise to consult further with someone who has been through this already with this transmitter. Short of that, I might guess that it might be necessary to lower the value of LM2 by perhaps one half, and raise the value of CM2 by about 2 as a new starting point. But then, I am only guessing.

edit: on further reflection, it seems that my final advice above is not sound. If I were truly matching a balanced very low impedance loop to a balanced fairly high impedance source, I would expect the shunt C on both sides to be somewhat large in value, and therefore low in reactance, while the series L would want to be relatively high in reactance. As you decrease the size of your loop, you are lowering the reactance, so it would logical to maintain the higher reactance of LM2 and decrease the reactance of CM2 (by raising the capacitance). Perhaps begin with a significant adjustment of CM2 as suggested followed by some pretty meaty changes to CM1 and LM1. Safe to say that a large change in the loop will require something other than "fine tuning" and therefore you must give major attention to changing LM1 and CM1. That might be the better path to take if you are going to tune by trial and error. I prefer to do much of my trial and error on a simulator as it is much more controllable.