Управление скоростью деформации: метод A с обратной связью и метод B
Метод A замыкает контур по экстензометру и даёт минимальный разброс; метод B — по скорости напряжения.

Two control paths
Method A of GB/T 228.1 closes the loop on the extensometer signal: the controller adjusts crosshead speed so the true strain rate within the gauge length stays in the specified band, such as 0.00025 s⁻¹ or the slower 0.00007 s⁻¹ option at yield, with ±20 percent tolerance. Method B runs open loop, using an estimated parallel-section strain rate or a stress rate — the actual strain rate then depends on specimen modulus and machine compliance and can miss the set point by a factor of two to three.
Effect on results
Rate sensitivity shows up most strongly in yield strength: at room temperature, each decade of strain rate raises the yield of carbon steel by about 10-20 MPa. Mixing methods A and B is a leading cause of interlaboratory disagreement. Closed-loop method A also suppresses crosshead overshoot during the yield plateau — lower-yield determination is extremely sensitive to rate fluctuations, and overshoot depresses ReL by 5 MPa or more.
Implementation essentials
Method A needs three things: extensometer update rate of at least 100 Hz, preferably 500 Hz; a controller with enough bandwidth to survive the stiffness collapse of one to two orders of magnitude at the elastic-to-yield transition without oscillation; and correct switching logic that runs stress rate in the elastic range and hands over smoothly to strain rate before yield. High-end servohydraulic and electromechanical machines include this by default; retrofitting old machines means upgrading both extensometer and controller.
Verification and records
Verify closed-loop performance with a known rate-sensitive specimen such as 6061 aluminium: run at the specified rate, replay the strain-time curve, and confirm the yield-region rate stays within the band without overshoot. Software should log the full curve and rate history automatically — audits look at the process, not just the numbers. All specimens of one batch must run at identical rates, otherwise within-batch comparability is meaningless.



