Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
IEC TR 61321-1:1994 — Transient Data Recording — Guidance and Test Methods
Digitizer Performance Characterisation, Sampling Techniques and Measurement Uncertainty for High-Speed Transient Recording
Scope: IEC TR 61321-1:1994 provides guidance on the selection, calibration, and use of transient data recording systems for capturing fast electrical transients such as lightning impulses, switching surges, and fast fault transients. It covers digitizer performance parameters, sampling methods, record length requirements, and measurement uncertainty assessment for high-voltage and high-speed testing applications.
1. Transient Recording System Characteristics
Transient data recording (TDR) systems capture single-shot or infrequent events with extremely high timing precision. Unlike repetitive sampling oscilloscopes, TDR systems must record all necessary information in a single acquisition. IEC TR 61321-1 addresses the key performance parameters that determine a system’s suitability for transient capture: sampling rate (samples per second), resolution (bits), record length (samples), analogue bandwidth, and trigger performance.
The standard emphasises that the sampling rate must be at least 5-10 times the highest frequency component of the transient for accurate amplitude reconstruction. For lightning impulse measurements (1.2/50 microseconds waveform), this requires sampling rates of at least 20 MS/s, ideally 100 MS/s or higher. For very fast transients such as partial discharge pulses (nanosecond rise times), sampling rates exceeding 1 GS/s may be necessary.
1.1 Digitizer Performance Parameters
The standard defines and specifies methods for measuring effective bits (ENOB), which is the most meaningful figure of merit for transient recording quality. ENOB accounts for quantisation noise, differential nonlinearity, aperture jitter, and thermal noise. A 10-bit nominal ADC may deliver only 7.5 effective bits at high sampling rates due to these error sources. The standard also addresses DC offset error, gain error, and integral nonlinearity — parameters that directly affect measurement accuracy.
| Transient Type | Rise Time | Minimum Sampling Rate | Required Record Length | Min. Effective Bits |
|---|---|---|---|---|
| Lightning impulse (1.2/50 microsec) | 1.2 microseconds | 20 MS/s | 5,000 samples | 8 bits |
| Switching impulse (250/2500 microsec) | 250 microseconds | 200 kS/s | 2,000 samples | 10 bits |
| Fast transient (IEC 61000-4-4) | 5 ns | 1 GS/s | 10,000 samples | 7 bits |
| Partial discharge pulse | 1-3 ns | 2-5 GS/s | 20,000 samples | 6 bits |
| VFTO (GIS disconnector) | 5-20 ns | 500 MS/s | 10,000 samples | 8 bits |
2. Sampling Techniques and Triggering
2.1 Real-Time vs. Equivalent-Time Sampling
The standard distinguishes between real-time sampling (single-shot acquisition at full sampling rate) and equivalent-time sampling (repetitive sampling of periodic signals). For transient recording, real-time sampling is mandatory because transients are non-repetitive events. The standard warns against using equivalent-time sampling for transient measurements, as it assumes signal periodicity that does not exist for true transients.
2.2 Pre-Trigger and Post-Trigger Recording
A critical feature for transient capture is the ability to record data before the trigger event. IEC TR 61321-1 recommends that recording systems provide at least 10-25% pre-trigger recording capability to capture the signal baseline and the leading edge of the transient. For lightning impulse tests, this ensures that the zero-reference level and the impulse onset are properly recorded, which is essential for accurate time parameter measurement (T1, T2 for impulse waveforms).
Measurement Pitfall: Insufficient pre-trigger recording is a common source of error in transient analysis. Without adequate pre-trigger data, the zero-reference level must be estimated from the recorded waveform itself, introducing uncertainty. For lightning impulse analysis per IEC 60060-1, an incorrect zero-reference can shift the measured peak value by 2-5% and significantly affect the calculated front time (T1), potentially causing false compliance decisions.
3. Calibration and Performance Verification
3.1 Step Response and Bandwidth Verification
The standard describes the use of a fast step generator to measure the digitizer’s step response, from which the analogue bandwidth, rise time, and settling behaviour can be determined. The step generator must have a rise time at least 3-5 times faster than the digitizer’s specified rise time to avoid measurement system interaction. The measured 10-90% rise time of the digitizer should be within 20% of the manufacturer’s specification.
3.2 Sinusoidal Testing for ENOB and Distortion
Effective bits and harmonic distortion are measured using a low-distortion sine wave source. The standard recommends testing at multiple frequencies spanning the digitizer’s bandwidth, typically at 10%, 50%, and 90% of the rated bandwidth. The sine wave amplitude should be set to 90% of the digitizer’s full-scale range to exercise most of the ADC codes. Total harmonic distortion (THD) and spurious-free dynamic range (SFDR) are reported alongside ENOB.
| Test | Input Signal | Measured Parameter | Acceptance Criterion |
|---|---|---|---|
| Step response | Fast step (< 1 ns rise time) | Rise time, overshoot, settling time | Rise time within 20% of specification |
| Sinusoidal | Low-distortion sine wave | ENOB, THD, SFDR | ENOB ≥ specified value – 1 bit |
| DC accuracy | Calibrated DC voltage source | Offset error, gain error | Offset < 1% FSR, gain error < 1% |
| Noise floor | Input shorted | RMS noise, peak-to-peak noise | < 0.5 LSB RMS |
| Trigger jitter | Synchronised pulse | Trigger timing uncertainty | < 0.2 × sample interval |
4. Engineering Best Practices for Transient Recording
Successful transient recording depends on more than just the digitizer specification. Practical engineering considerations include:
- Signal conditioning: Attenuators and probes must have bandwidth exceeding the transient’s frequency content. Voltage dividers for high-voltage transients must be compensated for frequency response and must not introduce significant phase shift in the frequency range of interest.
- Cable and connection quality: Coaxial cables must be properly terminated at both ends to avoid reflections. A 1-metre mismatched cable can produce reflections that distort the recorded transient by 5-10% for rise times below 10 ns.
- Electromagnetic shielding: Transient recording systems are often used in electrically noisy environments (high-voltage laboratories, converter stations). Proper shielding, common-mode rejection, and isolated measurement front-ends are essential to prevent noise from corrupting the recorded signal.
- Data analysis: Post-processing algorithms for parameter extraction (peak detection, time parameter calculation, curve fitting) must be validated against known reference waveforms. The standard recommends using algorithm testing with synthetic waveforms of known parameters to verify analysis accuracy.
Design Insight: When specifying a transient recording system for high-voltage laboratory use, prioritise vertical resolution (ENOB) over sampling rate for most applications. A system with 12 effective bits at 100 MS/s will provide more accurate impulse parameter measurement than a system with 7 effective bits at 1 GS/s. The additional vertical resolution enables better detection of small waveform features (e.g., partial discharge superimposed on the impulse) and reduces the uncertainty of time parameter calculations. However, for very fast transients (rise time < 10 ns), sampling rate becomes the dominant factor.
5. Frequently Asked Questions
Q: Can I use a standard digital storage oscilloscope (DSO) for IEC 61321-compliant transient recording?
A: Yes, many modern DSOs meet or exceed the requirements of IEC TR 61321-1, particularly those with ≥ 8-bit vertical resolution, deep memory (> 10 Mpts/channel), and comprehensive trigger capabilities. However, for highest accuracy in high-voltage impulse testing, dedicated transient recorders with 12-16 bit resolution and calibrated measurement channels are preferred. Verify that the DSO’s ENOB meets the application’s requirements at the frequencies of interest.
Q: What is the recommended record length for lightning impulse testing?
A: For standard 1.2/50 microseconds lightning impulses, a minimum record length of 5,000 samples is recommended, but 10,000-50,000 samples provides more margin for accurate parameter extraction. The record should include at least 25% pre-trigger data and cover the full impulse duration including any oscillations on the tail. For chopped impulses (where the impulse is abruptly terminated by flashover), the record length must be sufficient to capture 50-100 microseconds after chopping.
Q: How do I determine the required analogue bandwidth for a transient recording application?
A> As a rule of thumb, the system bandwidth (including probes and cabling) should be at least 5 times the reciprocal of the signal rise time (BW ≥ 0.35 / rise time for Gaussian response, BW ≥ 0.5 / rise time for flat response). For a 1.2 microsecond rise time impulse, this gives BW of only 300-400 kHz — easily met by any modern digitizer. However, the bandwidth requirement is driven by the fastest transient component of interest, not the overall waveform rise time. For detecting superimposed PD on an impulse, you may need 20-50 MHz of bandwidth.
Q: What causes aperture jitter, and why does it matter?
A: Aperture jitter is the uncertainty in the exact sampling instant of each ADC conversion. It arises from clock phase noise, ADC internal timing uncertainty, and trigger circuit noise. For high-frequency signals, aperture jitter causes voltage measurement uncertainty proportional to the signal’s slew rate. A digitizer with 10 ps RMS aperture jitter measuring a 1 MHz signal at full scale has negligible jitter error, but the same jitter on a 100 MHz signal produces approximately 0.5% RMS amplitude error, equivalent to losing 1-2 bits of resolution.
