RETURN LOSS FIGURES EXPLAINED
Return loss figures are being increasingly used in microwave articles and equipment descriptions. Actually I'm not sure they are overtaking VSWR but we hear about return loss much more now that more and more people are finding their way onto microwave frequencies. More use denotes more discussion. Hence we hear the term more often. This sort of makes perfect sense since VSWR is difficult to directly measure at these frequencies. Return Loss (RT) can be calculated using this formula:
RL (dB) = -10 log (P2)
Return loss directly relates to VSWR. In fact, VSWR and be calculated from return loss numbers. The chart below lists just a few examples of return loss numbers converted to VSWR.
|
Return Loss
|
VSWR
|
|
10 dB
|
1.90:1
|
|
14 dB
|
1.50:1
|
|
20 dB
|
1.22:1
|
|
25dB
|
1.12:1
|
|
30 dB
|
1.06:1
|
From this chart you can see that you don't need 40 dB or higher return loss numbers to have a nice VSWR. But stop and think about it, a 10 dB return loss means that only 10% of the forward power is coming back to the source. That is an approximate 2:1 VSWR. Still not all that bad.
How can return loss be measured? Well the simplest method is to use a microwave directional coupler rated for the frequency of interest. If you know the coupling from the sampling port is (for example) 20 dB then all you need do is measure forward power and sampled power; factor in the 20 dB coupling and you are essentially there. Of course the microwave coupler would need to be capable of measuring reflected power. Dual port microwave couplers are fairly common. If the only coupler you have is a single port couple then you could use it by putting it in one way, measuring forward power and then reversing its direction to measure reflected power. Let's do an actual example:
Measured forward power is 1 watt.
20 dB sampling signal would then be .01 watt (-10 dB is 10% power. 20 dB is yet another 10%) We move the decimal point two places.
The power in the measured port was say .001 watts (which is 30 dB down from the main signal).
Then we add the 20 dB back into the measured power and wind up with .1 watt
So we have now determined that our return loss is 10 dB and from the above chart we know we have a 1.90:1 VSWR.
What if you do not have access to a power meter?
Well another method of return loss measurement can be done with a diode detector that is rated for the frequency of interest. This would not be expected to be as accurate as a power meter but it can still provide useful information. Once you know the measured voltage from the detector you can calculate power across the 50 ohm load the detector sees. That formula would be:
E2/R
Or voltage squared divided by 50 ohms.
Here's additional info about return loss provided by Doug McGarrett, WA2SAY that is quite useful:
To calibrate, if you have only a diode detector, you must be able to control the power, whether by a calibrated external attenuator, or by a calibrated knob on the sweeper (or sig gen). You can't really rely on the square-law or linear function, or whatever is in-between, of a diode detector. Obviously, if you have a calibrated spectrum analyzer, this problem goes away. Or sweep really slowly with a power meter. Most power meters have some kind of analog output that can be put on a scope. (You still have to calibrate the scope, unless you know the db-to-voltage function of the power meter.)
Also, the directivity of the directional coupler is _extremely_ important. The directivity, or lack of it, adds directly to the errors in the measurement. If the directivity is 40 dB or so, you probably have the best coupler you can buy. If it's 15 dB or so, it's probably Microstrip, and for this purpose it's junk. Remember that VSWR's multiply, so if you measure a 1.2 VSWR with a coupler that has a directivity of 22 dB, you could have a VSWR of 1.4 or so. It might be better, but you couldn't prove it! I think you should put some emphasis on the quality of the directional coupler. Dual directional couplers were made by Narda specifically for this purpose, and some of them may be available at flea markets. The original price was astronomical, so if the seller knows what he has, the price is still likely to be quite high. And they are not all that broad-band; maybe less than an octave, I haven't actually seen one lately. A lot were actually mounted on a little stand. The big Narda and HP couplers with type N connectors will probably have a directivity of at least 28 dB, maybe a lot better, but probably not worse, and I don't have a feeling for the smaller couplers that work at X-band. Waveguide couplers should be pretty good, but at those frequencies you may have the problem of finding a really good 1:1 load, assuming you need to terminate whatever you're measuring, or even to validate the coupler.
You must calibrate your return loss measurement into an infinite or zero impedance. At frequencies up to 1 GHz or so, you can just leave the coaxial port open. At higher frequencies, I would recommend using a good short, either home-brewed, very carefully, or a commercial one found at a flea market. (There are also commercial "opens.") (You can _never_ calibrate a waveguide circuit into an open- ended guide. A waveguide is a fairly efficient antenna. So you _must_ calibrate waveguide circuits into a shorting plate.)
Also, as I think of it, the source impedance is important, in that the power level should not change due to load. Ideally, you would use a terminated circulator (i.e., an isolator) between the source and the coupler. In the non-ideal case, a 6 dB pad would help.