Sunday, May 12, 2019

#7 – So Many Radios – What Determines a Good One?

With so many types of radios, so many radios at different price points, how do we know what's a good one?

If there's one thing I've said a lot of times in the history of this blog, the answer to that question is another question:  “good for what?”  There are many reasons for having a radio system, and the definition of what's needed isn't always the same.  A radio that's intended to be battery powered and carried for long periods is different from one intended for a fixed installation with AC power; that's not even mentioning that radios for different radio services are different from each other.

I had a conversation with a guy who claimed to be a salesman for Motorola radios in the “old days”, when their HT series of radios was new.  These radios eventually became the defacto standard radio for police, fire and public service, but that outcome wasn't guaranteed when they were starting out.  He said a police radio buyer told him they needed utmost toughness and reliability from their radios because the radios had a hard life.  The salesman took the radio, not knowing if it would pass the test or not, and threw it against a wall.  He picked up the radio and it still worked.  The sale followed quickly.  While my guess is that all of us would like a tough radio when the time came that we dropped it or something made it go flying but that we might not be willing to pay two or three times the price of other radios that aren't as tough.

As a contrast to that, one of the reasons I bought my Yaesu VX-6R handie talkie is that it's rated water resistant to an industry specification, JIS7, that rates it for submersion of up to 30 minutes at a depth of up to three feet.  At ham shows, however, they'll submerse the radio in a fish tank less than a foot deep with wires to an external speaker, and leave it underwater all day.  No, I don't plan to use the radio underwater, but getting caught in the rain is pretty routine, and having a radio I don't need to worry about is a Good Thing.

Very early on in this series, I talked about the big three factors in receiver design: sensitivity, the sheer ability to hear any signal above the noise by enough margin; selectivity, the ability to receive that signal while rejecting much stronger signals off the desired frequency and potentially very close to it; and signal strength handling, often called dynamic range, the ability to handle multiple nearby strong signals without the receiver's performance degrading.  Let's go through each of these. 

Sensitivity can be stated as the required signal level to produce a required signal to noise ratio to the listener.  An immediate problem with this definition is that the signal level that's available to the receiver varies with a handful of factors.  The biggest variable is the frequency we're tuning to.  The effect of noise from the sun and the “universe” is to have higher noise levels at lower frequencies than at higher frequencies.  That means radios that work at lower frequencies can't sense signals as weak as radios working at VHF or higher frequencies can detect them.  Not from a signal generator, but from the environment.

We need to make a short excursion into the idea of noise and noise figure – two things that are different but related.  Noise is pretty simple.  All components generate electrical noise at a level determined by their resistance and bandwidth.  The noise voltage is given by:

E= ktBR

Where k is Boltzmann’s constant as known in Physics classes, t the temperature in degrees Kelvin, R is the resistance, and B the bandwidth in Hz.

For general receiver design, that tells us the noise in a 50 ohm resistor at room temperature.  That’s -174 dBm per Hz of bandwidth (this is slightly off the real value, but we all use it for rough calculations).  This says a 50 ohm resistor lying on your workbench puts out this much noise power, and a 1000 ohm resistor puts out 20 log (1000/50) or 26 dB more power. 

It’s easy to turn this number into the noise in any receiver bandwidth by adding 10 log (BW).  The noise bandwidth is approximately the passband or 3 dB BW of the circuit you’re considering.  The noise power in a 1 MHz BW is then -174 dBm + 10 log(1,000,000), equal to -174+60 or -114 dBm.

Noise figure is ratio of the Signal to Noise ratios (SNRs) of the output to the input of a network.

NF = SNRout/SNRin

A different way of saying that is that NF is the degradation of the SNR caused by going through the circuit.  A pad, for example has a lower SNR at its output than its input because of its loss.  The NF of a pad is its attenuation, the reciprocal of its gain.  A 3dB pad, for example, has a gain of -3dB and a noise figure of +3dB.  This is generally true for a passive circuit.

What this means is that we can determine what signal level gives us a required SNR by doing some arithmetic.  This is usually called the Minimum Discernible Signal or MDS.

MDS = -174 dBm/Hz + 10 Log (BW in Hz) + Desired SNR + NF

For example, in that previously mentioned example of a 1 MHz bandwidth, if we want a 10 dB SNR, we know it's at least 10 dB higher than that -114 dBm noise in that BW, or -104 dBm.  We're not done yet; -104 isn't our “final answer”; that answer depends on how much that -104 dBm is degraded by the receiver Noise Figure.  We'll call it 4 dB for a convenient number:

MDS = -174 dBm/Hz + 10 Log (1,000,000 Hz) + 10 dB SNR + 4 dB NF = -100 dBm

The problem is that this calculated number isn't very meaningful for lower frequencies.  This chart shows some typical noise levels at lower frequencies (below 1000 MHz), in dB above the ktBR as we calculate above.

Look at 10 MHz, for example, right at the middle of the horizontal axis and read going up.  It offers many different possible noise levels; the “quiet rural” noise floor is never reached because the atmospheric noise in the day or night are considerably higher.  If you're in that “quiet rural” place, during the day, the noise could be 35 dB higher than the calculated -114 dBm or -79 dBm.  It gets worse from there.  A rural environment will be 40 dB higher than ktBR, a residential neighborhood close to 50 dB noisier and a business district could be 60 dB noisier than ktBR.

A result of this is that if the design is for the lowest points of these curves, it's likely the environment will never be as good as the receiver and won't degrade the receiver.  For an HF radio, for example, it's virtually never necessary to have a noise figure below about 10 to 12 dB and therefore virtually never necessary to choose a radio based on Sensitivity claims.  For a VHF/UHF HT, it's worth looking at those numbers, and sensitivity can be verified fairly easily with test equipment that isn't all that expensive or hard to find.

As an aside, last year I did a couple of posts on a small antenna project I was building specifically for 10 MHz and below.  These antennas are viable at 10 MHz and below because, while they receive less signal and less noise, their directivity might allow the user to position the antenna to improve the SNR slightly.  It's only possible because the strength of the signal required to overcome the noise at these frequencies means we have signal to throw out in an effort to improve the SNR.

Unfortunately, selectivity doesn't offer us any fundamental relationships from the hard world of physics that make our work easier.  The requirements are really determined by the type of signals we want to receive, and a multi-mode radio can have large sections devoted to holding all the filters it might need.  You might recall that I said selectivity is bought by the cubic inch, and that's nowhere as evident as in a multi-band, multi-mode receiver.

Selectivity is entirely determined by the filters – largely either in the IF of an analog radio or in the DSP of a Software radio.  It's measurement of the amount that off-channel signals are attenuated (reduced).

Selectivity is often specified by not just the width of the passband (the lowest loss part of the filter) but by the ratio of the 6 dB loss bandwidth to the 60 dB loss bandwidth.  A theoretical filter (and the digital filters can approach this the closest) has a 60/6 shape factor of 1 – they're exactly the same bandwidth, something referred to as a “brick wall” filter – the attenuation shape is the rectangular outline of a brick.  Practically, filters with shape factors more like 2:1 or 3:1 are more common in analog radios.  A 2:1 shape factor means the 60 dB rejection part of the filter is in the adjacent channel, so if there aren't strong signals within that offset and nothing there to reject, 2:1 is fine.  Amateur use isn't as strictly controlled as commercial use and therefore might expose the user to interference in the adjacent channel.

Radios for commercial service will have requirements they must meet for the IF BW that will specify how far from the desired channel an interfering signal will be applied and how much they should be rejected.  For example, it may say to set a signal generator on the desired channel, set a 10 dB SNR, the move the generator frequency to desired offsets and increase the generator output strength to verify that you don't exceed the same SNR unless the signal is more than 60 dB higher than when on channel.  Most services will have a wide bandwidth set of tests, such as rejection of the image frequency, or any services that the radio is expected to encounter in service. 

The strong signal handling, or dynamic range, properties of the design are very much under the control of the designer.  In commercial services, there are requirements that new designs will be tested for compliance to, while in amateur and more casual services, there may be little or no mention of it.

While selectivity is bought by the cubic inch, dynamic range is bought by the milliamp.  The basic idea is that the radio distorts less if the signal being amplified in any stage is a small portion of the “idling” or bias conditions in the stage. 

When two signals are applied to an amplifier, say two strong signals in the band being monitored, the amplifier can multiply them by each other if that amplifier becomes nonlinear.  Since multiplication is the same as amplitude modulation, the two signals are said to intermodulate each other, producing Intermodulation Distortion, commonly called IMD.  The most troublesome products are generally closest to the two signals.  They're called “third order IMD” because of the terms in the equations below the graphic (two times the first frequency minus the second frequency).  Third order IMD is seen in the left picture here, the two smaller signals on either side of the two bigger signals.  You can imagine that if you're trying to listen to the something on the frequency of the smaller signal that when the IMD is present, it might wipe out whatever you're listening to.

One of the reasons these are troublesome is that they increase at twice the rate of the desired signals.  This leads to a condition, seen on the right, where the increasing intermodulation products (IM Power) will equal the power of the desired signals.  This is called the Intercept Point (IP) or, more specifically, the third order Intercept Point, IP3.

The general rule for this number is “the higher it is, the better it is”.  The cost, as briefly mentioned is more current consumption, which means lower battery life in a portable radio, and more heat in any radio.

The dynamic range of a receiver is rarely specified except in high end receivers, and there are many ways to measure it.  Since those signals might be attenuated by filters in the radio, typical tests will set different amplitudes for both off frequency channels and it is tricky to measure in the lab. 


  1. The common advice among serious gun owners is that in regards to optics you get what you pay in buy once cry once. Attempts to get by with cheap glass usually end up with a failure when it's needed most. With consumer electronics the same issue is true. For TV's and other such devices that come home and just SIT quality of construction is less an issue. For electronics that get dragged around, dropped and exposed to the elements rugged construction and attention to detail is far more important. Quality costs money. And if you need it to work you need to spend the money.

  2. This comment has been removed by a blog administrator.