In the last issue of Oscar News, a spreadsheet was described which allows you to calculate how strong the signals should be from AO-40, and it is hoped that the present article will enable you to make practical measurements to see if your system is up to par. The same method can also be applied to making comparisons between equipment such as different dish feeds or downconverters/preamps, or as an aid to optimising the performance of what you have (by measuring S/N ratios before and after each change is made).

Measuring signal to noise ratio using an S-meter alone is very unlikely to yield accurate results. Different manufacturers use different definitions of what an S-point is. My main station receiver (IC735) runs at about 3dB per S-point for example. Few radios seem to use the "accepted" definition of 6dB per S-point. Unless you have access to professional test equipment to enable the S-meter to be calibrated accurately, then clearly not much can be learned from the S-meter. Even then, I have found that receivers may drift, rendering a previous calibration potentially inaccurate after some period of time.

The S-meter is not a completely useless item, however, as it can be put to good use as a signal level indicator! The basis of the proposed method for measuring S/N makes use of a variable attenuator at IF (eg 28 or 144MHz) to reduce the level of the signal to the same as the background noise. The amount of attenuation required to do this is then a measure of the signal to noise ratio. Don’t worry if you don’t own a suitable attenuator – a hopefully reproducible design for homebuild is on the stocks.

In use, measuring signal to noise ratios with an attenuator is very straightforward. The system gain must be sufficient for the S-meter to be reading some finite value above zero, preferably in an area where the indication is as sensitive as possible to small changes in signal strength. This will probably occur in the S1 to S3 region. If the system gain is too high, some external attenuation may be needed to bring the S-meter into this region, and if the system gain is too low some form of amplification may be needed (see below). Off-tune the receiver from the signal to be measured, so that the s-meter is just reading noise, with the attenuator set to 0dB. Then tune in the signal for maximum S-meter reading, and set the attenuator until the S-meter reads exactly the same as it did with just noise. The number of dB's added is then the signal plus noise to noise ratio

Probably the main source of error is the accuracy of the attenuator itself, since the method relies on this to be accurate. Professional attenuators are probably OK, but if obtained second hand it is possible that one or more sections might be burned out. This would be evident by a larger or smaller than expected step in attenuation from one setting to another.

Another quite serious source of error is impedance mismatching. Most attenuators will only attenuate by the expected amount if they are loaded with the correct impedance. A mismatch will cause the designed attenuation to change, sometimes by a dB or more. Figs 1 and 2 show this effect clearly. The insertion of a fixed attenuator (of same working impedance as the attenuator) of 10dB or so should largely eliminate this source of error. Figs 1 and 2 show this effect clearly, in a real case. I have not (yet) measured the input VSWRs of my receivers, so I err on the safe side and use a fixed attenuator at all times on the output of the variable attenuator. Most S-Band downconverters should have enough conversion gain to still give a useable S-meter reading even with this extra 10dB attenuator in place.

The last source of significant error is largely in the hands of the operator. Obviously the "reference" S-meter reading on noise alone needs to be made carefully, and the signal needs to be then tuned in properly and the antenna adjusted for absolute maximum signal. If the S-meter moves to a region where it is not very sensitive to small changes when the signal is tuned in, making the exact maximisation of signal more difficult, the easiest thing to do is to use the attenuator to bring the signal down close the noise reference reading, and then do the final tuning/antenna tweaking. Finally the attenuator can be set to bring the fully optimised level to the reference point.

As with an S-meter, the attenuator will measure signal plus noise to noise ratios. This is because when a signal is tuned in, the S-meter indicates the total power coming into the receiver, ie signal and noise. If the signal is very strong, the difference between S/N and (S+N)/N is small, but at low signal levels the difference can be quite significant. A good example of this is when the incoming signal is at the same level as the noise, when the S/N ratio is 0dB, as they are at the same level. The (S+N)/N ratio, as measured, is 3dB.

The only reason to be concerned about this is consistency. Some link budget calculations may provide S/N predictions rather than (S+N)/N. The table below should assist conversion, if needed.

The method described above is generally applicable for any signal, but when measuring signals from AO-40, several things need to be borne in mind. Firstly, when the transponder is on, this itself generates noise, and so what will be measured using the attenuator will vary according to the transponder schedule. If the transponder is on, a beacon measurement would be made relative to the sum of the transponder noise added to the receiver noise. I have been advised that the transponder noise may vary according to configuration and transponder loading, so such a measurement will have limited value in checking out your system. It would be better to wait until the transponder is off. At present, the schedule includes a short slot on every orbit, near optimum squint where signals are at their strongest, where the beacon is on without the transponder. This is an ideal time for making measurements, and hopefully this slot can be retained in the schedule over the coming months.

If measurements can only be made when the transponder is on, the most accurate method is to tune off the beacon to a clear frequency nearby ( I usually go a little LF as this half of the transponder is usually empty), and then move the antenna off the satellite towards clear sky until the S-meter has reached a minimum reading. Note this as the reference value, and then tune back to the beacon and realign the antenna for maximum signal. Then use the attenuator to bring the S-meter reading back to the reference. Since the transponder noise is at a much lower level than the beacon, the resulting error will be very small.

A variable attenuator can also be used to measure sun noise, with some limitations. The main problem is that if the sun noise increase is only a few dB, as is the case with a 60cm dish at 2.4GHz for example, a switched attenuator with 1dB steps will be rather coarse to permit accurate measurements. It could be used to calibrate the receiver S-meter, allowing some degree of interpolation. With larger increases (say 6db or more) the direct use of an attenuator would be more practical.

Using sun noise to check out a 60cm class AO-40 system is perhaps less useful than measuring the AO-40 beacon, where much larger differences are available.

With the ability to quantify improvements that measuring signal to noise ratios brings, it is possible to embark on system modifications or adjustments with greater confidence. I have used this kind of optimisation recently to find the optimum position for dish feeds in part of a patch v helix performance study.

A common mistake when adjusting antennas is to go for maximum signal. Unfortunately moving the feed position in a dish usually has a significant effect on the VSWR of the feed, which in turn can affect the performance of the preamp or downconverter. The main manifestation of this is a changing background receiver noise level as the adjustment is made. What is of interest in making the adjustment is an improvement in signal to noise ratio, and making S/N measurements before and after making a change removes the effect of the interaction between antenna and the receiver. Any changing noise levels are accounted for, since the actual background noise level is used as the reference each time.

The attenuator can be set to any attenuation value between 0 and 31dB in 1dB steps, with a maximum error in any state of about +/- 0.2dB, based on the performance of the prototype.

A photograph of a prototype is shown below. Each "bit" uses a sub-miniature DPDT toggle switch to route the signal either directly through, or via a "pi" attenuator constructed out of precision 1206 size chip resistors. Standard resistor values were used from the E24 preferred value range, resulting in only very small deviations from the ideal attenuation values. Care was also taken to keep the VSWR of each "bit" as low as possible, to ensure that errors caused by interacting internal VSWRs were kept to a minimum.

As an option, the attenuator pcb also includes provision for a 10dB fixed attenuator on the output side. The function of this is to "buffer" the switched attenuator bits from the effects of non-50ohm loads, as explained above. Without this fixed pad, errors of up to 1dB have been measured operating the attenuator into a 3:1 VSWR load. With the fixed pad, the errors are reduced to less than 0.2dB.  The pcb also has provision for a simple input preamplifier, in case the system gain is not sufficient for the S-meter to read on background noise alone.

The attenuator was designed to operate at 144MHz and below. However, it is capable of useful performance at higher frequencies. One of the G3WDG prototypes is in use at 1269MHz between the up-converter and the PA, to adjust transmit power on the AO-40 uplink. As the frequency increases, however, the 1dB attenuation per step is lost.

.

© G3WDG March 2002