2008-08-02
Return Loss Bridge
I've built many return loss bridges over the years, this is but another. I've actually been using it for a while, but only just today got around to adding banana plugs for easy attachment to a multimeter and of course writing it up on the website. Unlike the VHF Impedance Bridge, output to a multimeter allows fairly precise measurement of return loss. The meter scale of the older bridge is better for tuning for best return-loss, or you might use a VOM instead of a digital meter with this bridge. The older meter still allows estimation of the reflection coefficient and comparisons between the instruments suggests the previous readings were quite valid.
The circuit is extremely simple and can be thrown together in a few minutes. I don't really use the external detector output much, it would be better terminated in a BNC as well, but if you are going to go to that trouble for use with a SA and tracking generator you might want to build the unit nicely in a screened box with a good layout to improve the directivity.
Initially it was constructed using all SMD 100 Ohm resistors, but abuse with use fractured them several times (point-to-point with SMDs soldered directly to BNC plugs is a bad idea!), so I replaced the two splitter resistors with leaded ones and kept the SMD ones for the reference load. The original configuration was quite good to high VHF, the leaded resistors only slightly worsening it.
The extra flexibility of the leaded resistors means they will absorb the flexing of the device as leads pull on the connectors without breaking as easily. The splitter resistors are less critical than the reference load anyway, due to symmetry all that really matters is that they are fairly well matched. 1% devices would have been a better idea, but even with the 5% ones the directivity of the unit exceeds 53 dB to beyond 50 Mhz.
The unit has been characterised from 100 kHz to 100 MHz and offers better than 43 dB all the way. Directivity peaks around 3 MHz at over 69 dB, by 10 MHz it has dropped to 60 dB. At 100 MHz it is at least 43 dB but may be slightly better as I am suspicious about the calibration load I was using.
Usage
To use the RLB apply a signal source to the generator input large enough to give several volts of DC out when the X port is open. Larger drive is better as it keeps you away from the diode non-linearity, but don't burn out the resistors. Note the reading on the multimeter, then terminate the X port with a good dummy load, the reading should drop enormously. Note the new reading and divide the original reading by it, take the log and multiply by 20. The result is the bridge directivity in dB. The directivity should exceed the return losses you wish to measure by a reasonable margin for sensible results, ideally make sure it exceeds at least 40 dB and the 0-30 dB RL region is of most practical interest.
Similarly in use, note the unterminated reading, then divide it by the reading once connected to the load under test, take the log and multiply by 20. The return-loss through an attenuator is twice the attenuator loss (through and back, attenuated twice). Return loss is specified as a -ve number by convention, +ve numbers imply gain not loss, but it is common to see RL specified loosely as a +ve figure.
Return loss, Reflection Coefficient (magnitude), and Standing Wave Ratio are all related mathematically and measuring one allows to infer the other two. Here is a table of some common data points:
| RL (dB) | |Ρ| | SWR |
|---|---|---|
| 0 | 1 | ∞ |
| 1.743 | 0.818 | 10 |
| 3.522 | 0.667 | 5 |
| 6.021 | 0.5 | 3 |
| 5.543 | 0.333 | 2 |
| 13.979 | 0.2 | 1.5 |
| 17.692 | 0.130 | 1.3 |
| 20.828 | 0.091 | 1.2 |
| 26.444 | 0.048 | 1.1 |
| ∞ | 0 | 1 |
| RL (dB) | |Ρ| | SWR |
|---|---|---|
| 0 | 1 | ∞ |
| 2.499 | 0.75 | 7 |
| 6.021 | 0.5 | 3 |
| 10.458 | 0.3 | 1.857 |
| 12.042 | 0.25 | 1.667 |
| 13.979 | 0.2 | 1.5 |
| 20.0 | 0.1 | 1.222 |
| ∞ | 0 | 1 |
| RL (dB) | |Ρ| | SWR |
|---|---|---|
| 0 | 1 | ∞ |
| 1 | 0.891 | 17.391 |
| 2 | 0.794 | 8.724 |
| 3 | 0.708 | 5.848 |
| 5 | 0.562 | 3.570 |
| 10 | 0.326 | 1.925 |
| 20.0 | 0.1 | 1.222 |
| 30.00 | 0.032 | 1.065 |
| 40.00 | 0.010 | 1.020 |
| 50.00 | 0.003 | 1.006 |
| 60.00 | 0.001 | 1.002 |
| ∞ | 0 | 1 |
Alternatively you can use my return loss calculator.
Practical Application
One of the first tests I did with it was to measure the line loss of my new coax runs. I spent most of last Saturday running 3 RG-58 coax lines from the shack to the balcony. The job is fairly neat, terminated at each end using BNC bulkhead female-to-female connectors and standard wall plates. The wall plates come from Jaycar and originally had four RCA sockets, which I removed and drilled out the holes to fit the BNC bulkheads. Using bulkheads means good terminations can be made at each end. I had to remove a two bricks from the cavity wall at the shack-end to feed the cable through, a process which took much time and cost some bruises. The real hard part of the job however was at the other end. The eaves are extremely difficult to access from the plenum. I was fortunate enough that a previous device had left a circular hole in the soffit cladding which I reused for this project, but I had to improvise a cable-snake/fish from some galvanised wire to pull the cables through. The access space was barely sufficient for a feline, let alone a larger human, so many more bruises and Oregon splinters were accumulated in the process. Tanya assisted greatly by feeding the snake from the balcony side, and I complete with knee pads and a head-mounted light source did the fishing job in the plenum.
Anyway, a week after that effort the cables were measured for loss. Leaving the balcony-end open circuited the return loss of the cable approximates twice the loss of the line. Of course this is at essentially infinite VSWR which increases the line loss, however this simple test gives a good indication of cable health. The run is only about 30 metres so a homebrew TDR is unlikely to be useful. One of the three runs is from a different batch of cable (I would have run four to utilise all the BNC bulkhead adapters, but ran out of RG-58). This different batch had a better loss compared to the other two runs, but only slightly. Runs 1 and 2 measured almost exactly the same at 1.40 dB at 3.581 MHz (I used the 80 metre beacon TX as a signal source). Run number 3 measured 1.26 dB. Considering these figures where measured at infinite VSWR the lines and terminations seem to be in fairly good condition.
The RLB was then used to measure the match of the 80 metre beacon's antenna after recent improvements in its water proofing. Once tuned up the return-loss is 26.5 dB, which is a ρ of 0.047 or a VSWR of 1.099. It drifts around a little in the wind as the capacitance of the whip varies, this is unavoidable because of the high-Q nature of the matching network to get reasonable efficiency, but the variation is quite small and acceptable.
7 comments.
Attachments
| title | type | size |
|---|---|---|
| Circuit Diagram Source | application/postscript | 12.880 kbytes |