I thought I would respond to both Carol and Daniel here in one post rather than do two separate posts. Such worries are prevelant when someone who has used analog communications for some time switches to digital. Much of the concern comes in the differences in the assumptions made while calculating link budgets. They use very different paradigms, so when looking at converting to digital from the paradigm used for analog, these concerns arise.

Digital communications are actually very stable, robust systems which can provide better performance that analog communications. In fact, there is a large gain in how much bandwidth is required between the two. Yes, there is a gentle roll of in analog communications, whereas digital simply ends. However, this analog roll off is accompanied by a degredation in quality. The real trick here is making sure the design of the system (not just the radio) takes into account the quality required by the user. These parameters can be quantified and measured - it goes beyond a user complaining the quality is low.

Now, I mentioned the "system" design needs to take this into account. A simple analog radio is not much more than an a modulator and amp which takes the input from the mic, modulates it using AM or FM, amplifies it and transmits that amplified signal. The input to the radio is analog, so it the modulation and transmitted signal. A digital radio requires a little bit more. It requires a digital input, it only recognizes bits. Granted, modern digital radios are able to convert several bits into symbols (using QPSK or other higher order modulation schemes) vs. processing a single bit at a time (such as BPSK). But regardless, it requires bits. Depending on the modulation scheme and datarate, it will take these bits, assign them to a symbol based on its "constellation" and transmit those symbols. It requires far less power and far less bandwidth to do so.

But in voice communications, this does not work directly. It requires an additional step not found in analog voice communications. The voice must first be "encoded" before it is sent to the radio. This encoder is what converts it from an analog signal to a digital one. This is where the quality of the digital radio is failing. In order to keep the bandwidth extremely low, they use a low encoder rate, meaning there are less samples of the analog signal to transmit. For example, a CD is created using a sampling rate of around 44 KHz. This provides a pretty good sound quality. This frequency is selected because it is twice the frequency of the highest frequency the human ear can hear. If the sampling rate is lowered, the quality will suffer. If you take that CD and "rip" a song from it to the MP3 format, you have the ability to re-sample the audio. If you select the lowest quality rate, it will sound funny

This same thing is going on in these radios. If they select a very low quality encoder in order to decrease the required bandwidth (thereby gaining more channels), the audio quality will be low. They need to find an encoder which will provide good quality, but also reduce the bandwidth. From what I have heard, they are trying to reduce the bandwidth required by between 3 and 6 dB. That might be a bit of a stretch.

The other thing which needs to be addressed from a systems perspective are link budgets, or more specifically how these radios are going to be used. The range/data-rate requirements of a disadvantaged user (ie hand-held radio) are going to be very different from that of an advantaged user (locomotive or dispatcher radio). They need to look at what kind of ranges they can get at these data rates to make sure they are not creating a safety issue. Ground to Ground communication is not the easiest thing to accomplish, since terrain and other issues are so different they are nearly impossible to model accurately.

Either way, the switch to digital is a good thing, so long as it is done correctly, with the railroads and the radio manufacturers looking at it from the systems, or big picture, perspective.