Most engineers are probably familiar with the basic idea of time-domain reflectometry (TDR): The time-domain reflectometer sends a signal, frequently a square wave or stepped pulse, into the medium under investigation (typically a cable) with the far end of the cable shunted, open or left connected at the receiver. The instrument measures the time elapsed for the reflected signal to return to the instrument, in the manner resembling that of radar. The amplitude of the reflected signal is a function of the impedance of the fault–in other words, the magnitude of reflected signal characterizes the impedance at the fault. The distance to the discontinuity can also be determined from the time taken for the pulse to complete the round trip. Because the time-domain reflectometer is sensitive to impedance variations, it can measure and display cable impedance qualities, splice and connector locations and cable lengths.
But there is one problem with TDR: The circuit under investigation typically must be deenergized before the TDR can be connected. That’s not a problem if the line is already dead, but it eliminates TDR from being applied on lines that have issues but must stay in use. However, there are more exotic types of TDRs that can handle lines that remain in use during the test. Where traditional TDR uses stepped or pulsed voltages, frequency domain reflectometry (FDR) and mixed-signal reflectometry (MSR) use sinusoidal signals. Chaos or noise domain reflectometry use existing or added noise in the system. Spectral or binary time domain reflectometry use pseudo-noise (PN) codes, and spread spectrum time domain reflectometry (SSTDR) uses modulated PN codes. The magnitudes of the reflections will be essentially the same for each reflectometry approach, if the bandwidth is the same.
Of these, SSTDR can detect and locate faults within operating (energized) electrical systems. Originally developed for locating faults on live aircraft wires, SSTDR can detect and locate problems such as open and short faults, arc faults, and conductor damage. For example, DoE engineers used SSTDR in PV systems to detect, and in some cases locate ground faults, arc faults, disconnection faults, and broken panels.
SSTDR sends a square-wave-modulated PN code signal into an electrical system. The returned signal is cross-correlated with a delayed version of the incident signal to create the reflectometry signature. In any reflectometry system, the reflection coefficient (Γ) tells how much signal will be reflected, and is a function of the impedances of the transmission line Zline and fault Zfault:
in which reflected and incident are the reflected and incident voltages. The reflection coefficient is frequency-dependent. However, when the fault is not frequency-dependent (as for open and short circuits), Γ is equal at all frequencies and can be applied across the time domain reflection signature. For an open circuit, Γ= 1. For a short circuit, Γ= -1. For open and short circuits, all of the SSTDR signal reflects either in phase (open circuit) or out of phase (short). The short circuit is the largest reflection for any electrical system. Thus the ability to locate all other types of faults relates to this maximum reflection coefficient (Γ= 1). The location of the reflection comes from multiplying the time delay by the velocity of propagation on the cable.
The point of using a spread of frequencies in SSTDR is that the higher frequencies (shorter wavelength) provide accuracy at shorter distances while the low frequencies (longer wavelength) allow monitoring of longer cables. SSTDR can run on energized wires because the spread spectrum signals can be isolated from the host system noise and activity. Small impedance changes caused by known conditions (connectors, loads, etc.) as well as inherent system noise can be ignored. Because SSTDR doesn’t require monitored items to be taken off line, it can be used for live monitoring during normal use so events and their location can be logged as they happen, and collected data can be used to build a profile or signature of a system.
In a typical SSTDR setup, a sine wave generator serves as a master system clock. Its output is converted to a square wave which drives a PN sequence generator. The of PN generator output is multiplied with the initial sine wave, generating a direct sequence spread-spectrum (DSSS) binary phase shift keyed (BPSK) signal. This signal gets sent down the wire to the DUT. The signal is reflected back by any impedance discontinuity against the characteristic impedance of the propagation path. The reflected signal is cross-correlated with the delayed copies of the incident signal, and a lobe is generated at a time delay in the autocorrelation plot that corresponds to the distance from the source terminal to the impedance mismatch at the load terminal. A lobe with a positive peak indicates a positive reflection coefficient (implying load impedance exceeds the characteristic impedance) and a negative peak indicates a negative reflection coefficient.
Systems having a wide variation in impedance can be challenging to analyze via SSTDR for anything more than open and short circuits. Examples include PV arrays which contain different materials and kinds of interconnections. So it’s easy to find research papers still being written about SSTDR correlation techniques and advanced statistical methods applied to SSTDR.