Designing polymorphic subclasses so that new ones can be easily added without redeploying

I am redesigning an application that is used for data acquisition. We have a base class called Sensor which is more or less as follows:

public abstract class Sensor
{
    public virtual String Name { get; }
    public virtual UnitOfMeasurement Unit { get; }
    public virtual int SamplingRate { get; }
    public virtual CalibrationModel Calibration { get; }
}

Each of its subclasses has a "hardcoded" value for each property. For example EmgSensor has "EMG" as Name, "UnitOfMeasurement.Volt" as Unit, etc.

Our current system has this problem I am trying to solve: Each time that a new actual sensor hardware is developed and added to our portfolio, we need to do the following "shotgun surgery":

1. Create a new Sensor subclass;


2. Perform some scattered, minor changes on the rest of the system (due to depencencies, etc.);


3. Release an update to our client base.

I understand that each physical sensor (the hardware) is a conceptual unit, and as such should be able to be "plugged" to a running system so that it doesn't need to be recompiled.

On the other hand, if each subtype of Sensor is to be considered a class, how could I create a "dynamic" class from, say, configuration files? Should I be using the Factory pattern? Is there any other, more suitable way of achieving what I need?

I am using C# (although I think this is incidental to the question).