How does a BERT measure bit error rate?

A BERT measurement has two synchronized halves: Pattern Generator (PG): generates a deterministic pseudo-random binary sequence using a linear feedback shift register (LFSR). PRBS-N generates a sequence of length 2^N - 1 bits before repeating. Common patterns: PRBS-7 (127 bits): short, used for quick functional checks and clock recovery verification. PRBS-15 (32,767 bits): standard for SONET/SDH testing. PRBS-23 (8.4M bits): ITU-T O.150 standard for telecom. PRBS-31 (2.1G bits): longest standard pattern, stresses CDR bandwidth and baseline wander. The PG drives the DUT input at the target data rate with calibrated output amplitude, pre-emphasis, and de-emphasis. Error Detector (ED): receives the DUT output, recovers the clock (CDR or external reference), synchronizes to the PRBS pattern (typically within 2^N bits), and performs bit-by-bit comparison. Errors are counted and accumulated over the measurement interval. BER = bit_errors / total_bits_compared. Statistical requirements: to state BER < X with confidence C, need N = -ln(1-C)/X bits error-free. For BER < 10⁻¹² at 95% confidence: N = -ln(0.05)/10⁻¹² = 3×10¹² bits. At 10 Gbps: 300 seconds. At 100 Gbps: 30 seconds.

What PRBS pattern should be used for testing?

Pattern selection depends on the DUT and what stress conditions are needed: PRBS-7 (2⁷-1 = 127 bits): shortest repeating pattern. Used for: initial link bring-up, quick pass/fail screening, devices with narrow CDR bandwidth. Limitation: does not stress baseline wander or long run lengths. PRBS-9 (2⁹-1 = 511 bits): used for 10GBASE-T Ethernet PHY testing per IEEE 802.3. PRBS-15 (2¹⁵-1 = 32,767 bits): standard for SONET/SDH (ITU-T O.150). Adequate for most telecom equipment. Maximum run length = 15 consecutive identical digits (CID). PRBS-23 (2²³-1 = 8,388,607 bits): standard for high-speed testing. Maximum run = 23 CID. Stresses AC coupling and CDR tracking bandwidth. Required by many 25G/50G/100G standards. PRBS-31 (2³¹-1 = 2,147,483,647 bits): most demanding standard pattern. Maximum run = 31 CID. Stresses baseline wander in AC-coupled links, CDR loop bandwidth (requires <100 kHz BW), and equalizer adaptation. Required by IEEE 802.3 for 100GBASE-SR4 and higher. SSPRQ (Short Stress Pattern Random Quaternary): PAM4-specific pattern for 400G/800G testing. 2^13 - 1 symbols, designed to stress all PAM4 symbol transitions equally. QPRBS-13: PAM4 variant used in OIF CEI-56G-VSR and IEEE 802.3bs.

What are the key specifications when selecting a BERT?

Critical BERT specifications for RF/high-speed digital applications: 1) Maximum data rate: determines which standards can be tested. Current tiers: 32 Gbaud (for 25/28G NRZ), 64 Gbaud (for 50G PAM4, 400G-DR4), 120+ Gbaud (for 100G per lane, 800G). 2) Pattern generator output: amplitude range (50-1000 mVpp typical, 1500 mVpp for some), rise/fall time (5-15 ps for 10-20%/80-90%), output jitter (<500 fs RMS, <2 ps peak-peak), pre-emphasis/de-emphasis (2-4 tap TX FIR), common mode voltage range. 3) Error detector sensitivity: minimum signal amplitude for reliable detection (10-50 mVpp for state-of-the-art), input bandwidth (1.5× data rate minimum), clock recovery bandwidth (adjustable, 100 kHz to 10 MHz), equalization (CTLE + DFE, 1-15 taps). 4) Jitter injection: sinusoidal jitter (SJ) for jitter tolerance testing, random jitter (RJ) calibrated to target RMS, bounded uncorrelated jitter (BUJ) per IEEE 802.3. 5) Eye diagram/bathtub curve: integrated sampling oscilloscope for eye opening measurement. Eye height, eye width, jitter components (DJ, RJ, TJ). 6) PAM4 support: for 400G/800G. Gray-coded PAM4 PRBS generation, per-level BER measurement, TDECQ (Transmitter Dispersion Eye Closure Quaternary). Price range: $50K-$500K depending on data rate and features.

Test & Measurement

BERT

Q: What PRBS pattern should be used for testing?

Pattern selection depends on the DUT and what stress conditions are needed: PRBS-7 (2⁷-1 = 127 bits): shortest repeating pattern. Used for: initial link bring-up, quick pass/fail screening, devices with narrow CDR bandwidth. Limitation: does not stress baseline wander or long run lengths. PRBS-9 (2⁹-1 = 511 bits): used for 10GBASE-T Ethernet PHY testing per IEEE 802.3. PRBS-15 (2¹⁵-1 = 32,767 bits): standard for SONET/SDH (ITU-T O.150). Adequate for most telecom equipment. Maximum run length = 15 consecutive identical digits (CID). PRBS-23 (2²³-1 = 8,388,607 bits): standard for high-speed testing. Maximum run = 23 CID. Stresses AC coupling and CDR tracking bandwidth. Required by many 25G/50G/100G standards. PRBS-31 (2³¹-1 = 2,147,483,647 bits): most demanding standard pattern. Maximum run = 31 CID. Stresses baseline wander in AC-coupled links, CDR loop bandwidth (requires <100 kHz BW), and equalizer adaptation. Required by IEEE 802.3 for 100GBASE-SR4 and higher. SSPRQ (Short Stress Pattern Random Quaternary): PAM4-specific pattern for 400G/800G testing. 2^13 - 1 symbols, designed to stress all PAM4 symbol transitions equally. QPRBS-13: PAM4 variant used in OIF CEI-56G-VSR and IEEE 802.3bs.

Q: What are the key specifications when selecting a BERT?

Critical BERT specifications for RF/high-speed digital applications: 1) Maximum data rate: determines which standards can be tested. Current tiers: 32 Gbaud (for 25/28G NRZ), 64 Gbaud (for 50G PAM4, 400G-DR4), 120+ Gbaud (for 100G per lane, 800G). 2) Pattern generator output: amplitude range (50-1000 mVpp typical, 1500 mVpp for some), rise/fall time (5-15 ps for 10-20%/80-90%), output jitter (<500 fs RMS, <2 ps peak-peak), pre-emphasis/de-emphasis (2-4 tap TX FIR), common mode voltage range. 3) Error detector sensitivity: minimum signal amplitude for reliable detection (10-50 mVpp for state-of-the-art), input bandwidth (1.5× data rate minimum), clock recovery bandwidth (adjustable, 100 kHz to 10 MHz), equalization (CTLE + DFE, 1-15 taps). 4) Jitter injection: sinusoidal jitter (SJ) for jitter tolerance testing, random jitter (RJ) calibrated to target RMS, bounded uncorrelated jitter (BUJ) per IEEE 802.3. 5) Eye diagram/bathtub curve: integrated sampling oscilloscope for eye opening measurement. Eye height, eye width, jitter components (DJ, RJ, TJ). 6) PAM4 support: for 400G/800G. Gray-coded PAM4 PRBS generation, per-level BER measurement, TDECQ (Transmitter Dispersion Eye Closure Quaternary). Price range: $50K-$500K depending on data rate and features.

/bert/ — Bit Error Rate Tester

Precision instrument combining a pattern generator (PRBS-7 to PRBS-31) and error detector for measuring BER of digital links. BER = errors/total bits. Measurement time for 10⁻¹² at 95% confidence: T = 3×10¹²/data_rate. Modern BERTs: 32 to 120+ Gbaud, PAM4 support, integrated jitter injection (<500 fs RMS), eye diagram analysis, CTLE/DFE equalization. Applications: SerDes, optical transceivers, 5G baseband, satellite modems.

Rate: 120+ Gbaud

Jitter: <500 fs

Sensitivity: 10–50 mVpp

Understanding BERT Instruments

The BERT is the definitive pass/fail instrument for digital communications. Whether qualifying a 400G optical transceiver, validating a satellite modem's demodulator, or characterizing a SerDes PHY on a next-generation SoC, the BERT provides the objective measurement: how many bits arrive incorrectly? Every digital RF link has a BER specification, and the BERT is how you prove compliance.

Modern BERTs have evolved far beyond simple error counting. Integrated jitter injection allows stress testing against specifications like IEEE 802.3 jitter tolerance masks. Eye diagram analysis reveals signal integrity issues (ISI, crosstalk, noise) that explain why errors occur. PAM4 capability supports the 400G/800G generation where each symbol carries 2 bits across 4 amplitude levels, requiring per-level BER measurement and TDECQ analysis.

Measurement Time Calculation

Minimum Bits for Confidence:
N = −ln(1 − C) / BER_target
C = confidence level (0.95 for 95%)

Example: BER < 10⁻¹², 95% confidence
N = −ln(0.05) / 10⁻¹² = 3×10¹² bits
At 10 Gbps: T = 300 seconds
At 100 Gbps: T = 30 seconds
At 400 Gbps: T = 7.5 seconds

BER Calculation:
BER = E_errors / N_total
Confidence interval: Poisson (low error count)
or Normal approximation (high count)

PRBS Pattern Comparison

Pattern	Length	Max Run	Application
PRBS-7	127	7 CID	Quick functional check
PRBS-9	511	9 CID	10GBASE-T
PRBS-15	32,767	15 CID	SONET/SDH, ITU-T O.150
PRBS-23	8.4M	23 CID	25G/50G/100G standard
PRBS-31	2.1G	31 CID	100GBASE-SR4+, max stress
SSPRQ	8,191 sym	N/A	PAM4 400G/800G

Common Questions

Frequently Asked Questions

How does BER measurement work?

PG generates PRBS via LFSR, drives DUT. ED recovers clock, syncs to pattern, counts mismatched bits. BER = errors/total. For 10⁻¹² at 95%: need 3×10¹² bits (300 s at 10 Gbps, 30 s at 100 Gbps).

Which PRBS pattern?

PRBS-7 for quick checks. PRBS-15 for SONET/SDH. PRBS-23 for 25G to 100G. PRBS-31 for max CDR/baseline stress (31 CID). SSPRQ for PAM4 400G/800G testing. Longer patterns stress more failure modes.

Key BERT specs?

Data rate (32 to 120+ Gbaud), PG jitter (<500 fs RMS), ED sensitivity (10 to 50 mVpp), rise time (5 to 15 ps), PAM4 support, jitter injection, CTLE/DFE equalization. Price: $50K to $500K.

Related Terms

← Berlekamp-Massey Beryllia Substrate →

Test Solutions

Precision RF Components

RF Essentials provides precision terminations and custom waveguide assemblies for BERT test fixtures, high-speed channel characterization, and digital communications link qualification.

Request a Quote