Interference in low power frequency measurements, using a frequency prescaler
When using a frequency prescaler, we tend to forget that we are dealing with a very wideband, wide dynamic range receiver, because usually we are measuring high amplitude signals, which are dwarfing all noise and spurious emissions that our test setup might be picking up.
Unlike other receivers (like a spectrum analyzer), a prescaler lacks the ability to tune into a signal, instead, the prescaler “sees” all existing signals in its operating range, at all times, even when being triggered by a larger signal, and because prescalers are literally counting edges, it is easy to understand how additional edges, created by interfering signals, could cause problems with our measurements.
What does interference looks like?
Signals capable of generating continuous interference are those from broadcast stations, mobile telephony towers, as well as any devices that transmit a continuous stream of data, or which transmit packets at a high rate.
These type of devices are usually also responsible for spurious interference, because they normally operate in bursts, to conserve their own battery power or the battery power of the devices they are talking to. For example, Wi-Fi communication happens in bursts, and in many locations today, the signal from the Wi-Fi access point is the most powerful signal that can be received.
When a measurement is being affected by such a signal, the frequency reported by the frequency counter will present some form of instability. In the case of a continuous or very frequent interfering signal, you would see the last couple of digits fluctuating or even a completely wrong reading.
For spurious interference, the same effect can be observed, with the exception that it only happens for one or two measurements at once, followed by a period of correct/stable measurements.
How stray signals get picked up
There are many ways in which an external signal can be injected into the system you are measuring.
One common way to pick up signals from the environment is by using long coaxial cables. The lower quality and the longer the cable is, the higher the odds that it will pick something up. The location of the long cable is also important – for example, if you had it before an amplifier, rather than after, then that could make things worse, but it could also make them better if the amplifier had attenuation at the offending frequency, instead of gain.
Another common way of injecting stray signals into a system is through the power supply cables, which are usually unshielded, and are excellent antennas. The signals that power cables receive may get attenuated by the input filtering network of the DUT, but power supply filtering is rarely wideband, so it’s bound to be inefficient at least at some frequencies.
Obviously, any unshielded part of a system is also a good entry point for such signals, as well as any IO cables, which are almost never followed by any kind of filtering, other than ESD protection perhaps.
One cause many people overlook is sometimes other test instruments that are part of the experiment. Name brand instruments would rarely exhibit this issue, but in the past years, the market has seen an increase in a variety of low cost RF signal generators, which in the majority of the cases, are unfortunately poorly designed despite appearances. Aside from the high harmonic content, they are also known to pick up and amplify stray signals from the environment, which are then made available directly at the output port.
Methods of suppressing stray signals
The first step in stopping this problem is identifying the cause. Randomly unplugging wireless devices, like the closest Wi-Fi access point, is a great way of detecting it. When that fails, you can resort to observing the input to the prescaler, on a spectrum analyzer, however finding the offending signal could prove to be difficult, especially if it happens in very short bursts (like Wi-Fi), which when you don’t know what you’re looking for, would be missed by most analyzers.
Once the cause is identified, depending on the type of measurement you are making, blocking the problem frequency with a band-stop filter is a perfectly reasonable solution, especially if it is in a very popular band, for which band-stop filters are widely available. Problems may arise if you are picking more than one signal, from different bands, because that can easily lead to a tremendous waste of time.
A simpler approach is to shield your DUT, use shorter cables or eliminate them completely, use ferrite beads on power and IO cables, or simply turning off the Wi-Fi access point and/or any other devices you have identified as the cause of the problem, for the duration of the experiment.
The level of attenuation that needs to be achieved, at the problem frequency, should be enough to put the interfering signal at least 10 dB below the signal you are trying to measure, after adjusting for the sensitivity curve of the prescaler.
When the environment is the limiting factor in your measurements, the only things you can do is try to change it or to isolate your setup from it. Sometimes it’s only a temporary problem, other times it always happens in a specific interval of the day.
For those days, I have come to the conclusion that a cheap Faraday cage, that you can improvise from a variety of metal enclosures, does wonders for very low power measurements.