FAQ: HBM I-V Curves

I-V curves are a plot of device (DUT) current that results from applied voltages. The original Transistor Curve Tracers made I-V curve traces by sweeping a voltage while measuring the current and displaying the results in real time on a CRT screen. Today this is done in a very similar way with Source-Meters or Semiconductor Parameter Analyzers. An HBM system that has high speed voltage and current measurement capabilities can make measurements just like the old Transistor Curve Tracers. The results show the operation of DUTs during the HBM pulse.

TLP makes a quasi-static pulsed I-V measurement. Each TLP rectangular pulse produces a voltage across, and a current through, the DUT and these values are measured during the pulse. It is called quasi-static because the measurement is made after the initial transient has settled and the values become semi-stable. HBM pulses do not have stable values like TLP, but if DUT current is measured at the same time as DUT voltage then a plot similar to a TLP I-V curve can be made. This is more like the older transistor curve tracers with measurements being made while the values are continuously changing. While data can be taken during both the HBM pulse rising and falling current, the rise time is much shorter than the fall time and therefore the fall time provides more data points and is closer to TLP because the data is taken after the initial pulse edge.

TLP provides a low noise measurement because the DUT current and voltage are reasonably stable during the measurement time. The raw measurement data is averaged over a significant time period of the TLP pulse to remove noise. HBM in contrast is taking data while the parameter values are changing and may do little or no averaging over time. While the HBM I-V curve has significantly more noise, the entire I-V curve is taken in one HBM pulse so it is a much faster test than TLP.

While there is more noise in HBM I-V curves, they can both be calibrated to provide accurate information. Similar techniques for measurement, corrections, and calibration can be used in both cases.

This is one way to differentiate HBM I-V curves from the quasi-static TLP measurements. Since the HBM measurements are made while the DUT currents and voltages are changing, they show the dynamic response to the HBM stress. Measuring the clamping voltage during the HBM pulse provides dynamic characterization information.

Because they show different information and together they present a more complete picture of the DUT being tested. ​ - TLP is often used to determine the details of a DUT that fails to reach its target voltage during an HBM test. An HBM I-V is a direct measurement of the DUT’s response to the HBM stress and can provide the designer with the needed information to understand the failure. - The test impedance of HBM is 1500 ohms while TLP is usually 50 ohms, which means that an HBM I-V curve can be 30 times more current sensitive. - The holding current information and turn-off characteristics of a snapback device can clearly be seen in a HBM I-V but only inferred from a TLP I-V. - Many devices have a negative resistance R-On at low currents that can only be seen with high impedance TLP or HBM I-V. - Heating effects are seen in TLP I-V curves at high current levels where the straight R-On curve sags. This implies that at low current there are no heating effects. - However, the HBM pulse produces the most heat at the beginning of the pulse and as the HBM pulse decays higher clamping voltages are measured at the low currents with HBM than with TLP. - HBM I-V curves provide a clearer picture of the actual behavior during an HBM pulse while TLP provides the best pulsed I-V information.