To detect muscle signals, conductive metal electrodes are placed in contact with the skin. EMG signals at the skin surface fall in the microvolt to millivolt range, and with such small signals the contact between the skin and electrode must be very intimate. Most commercially available devices employ bipolar measurement between two electrode contacts (and sometimes a third for grounding reference) (Figure 1). These electrodes are connected to a processor/amplifier (Figure 2). The processor examines the signal and filters out what may likely be interference. This interference may be caused by a variety of influences in the environment. Fluorescent lights, microwave ovens, and cell phones are three of a plethora of common items encountered in daily life that can cause interference. Failure to filter this could easily cause the unintentional motion of a device, crushing a can, or dropping a delicate object. The amplifier increases the signal from microvolts to volts that can be more easily processed by the main controller of the prosthetic device.

Most modern commercially available electrode amplifiers provide a clean, filtered signal. Many also have an accessible gain potentiometer to allow adjustment of the amplification.

The signal is sent from the electrode amplifier to the device controller. This microprocessor provides adjustment of signal amplification, threshold adjustment, control schemes, and use of alternative input devices. With multiple degree of freedom systems, input switching from one device to another is accomplished by the controller.

With prosthesis controllers, a digital user interface provides access to make adjustments. Each prosthetic manufacturer designs their own user interface to communicate with their devices. The platform may be a laptop computer, an iOS, or Android handheld device.

There are many advantages with myoelectric control, one being the coordinated nature of using muscle contractions to control a closely
associated movement in the prosthesis. In dual-site control it is common to assign control movements to antagonistic muscles, matching control signals to native biologic movements to the extent possible. For example, in the case of a transradial level amputation, remnant wrist extensors can control opening a prehensile device, while remnant wrist flexors close the device, much the same as with an intact limb. At the more proximal amputation level, dual-site control of multiple joint systems may be required through inputs that will be described shortly. For example, at the transhumeral amputation level remnant elbow extensors are commonly assigned control of both elbow extension and opening of a prehensile device, while remnant elbow flexors are commonly assigned control of both elbow flexion and the closing of the prehensile device. Through sophisticated systems, these two sites can also provide wrist pronation and supination. Control systems strive to allow the user to control three degrees of freedom as intuitively as possible.

In some cases, only one useable muscle site can be found. This may be because of a degloving injury, severe burns, or neurological trauma. The challenge is creating a system where the wearer can control two directions of motion, for example, a hand open/close or elbow flexion/extension with only one signal. For situations like this, several software algorithms have been developed to assign an action to a single muscle contraction.