Genset voltage sag is the short dip the bus takes when a block load or a motor start pulls harder than the alternator can hold at rated voltage, and it decides whether a site rides through or drops out. A sag is not a fault code and it rarely damages the set itself; the risk lands on the equipment the bus feeds, so the limits come from the load side, not the engine side. Yangzhou Tesla Power Equipment Co., Ltd. matches the excitation option to the load schedule at the quotation stage, so the dip figure is calculated before the set ships rather than argued about after it lands. This guide explains what counts as a sag, which numbers matter for IT loads and contactor coils, and which fixes hold up in service. Send the load schedule and ask the export desk for the dip calculation beside the offer. Four points carry the decision.
- Depth: the ITIC envelope gives IT loads 70% of nominal for 500 milliseconds, and contactor coils release lower.
- First dip: set by the subtransient reactance; a 250kVA frame with 12% facing a 100kVA block dips 4.8%.
- Recovery: the AVR ceiling brings the bus back inside 1.5 seconds when the excitation is independent.
- Fix: PMG or AREP excitation, a soft starter on the largest motor, and staged steps before a larger frame.
What Counts as a Voltage Sag on a Genset Bus?
A sag is a fall in voltage to between 10% and 90% of nominal that recovers on its own within one minute, and on a genset bus it usually lasts a few cycles to a few seconds. It is not an interruption, which drops the bus to near zero, and it is not a swell, which raises it. The genset keeps running through a sag; the question is what the connected equipment does while the bus sits low.
On a standby set the first dip is predicted arithmetic rather than a mystery. The depth follows from the starting block and the alternator design, and the recovery follows from the excitation system. That makes the sag one of the few site risks you can price at the quotation stage, which is why the numbers below belong in the offer, not in the commissioning report.
Genset Voltage Sag on a Block Load: The First Cycle
The size of the first dip is arithmetic, not luck. The subtransient reactance of the alternator stands between the bus and the block. A 250kVA set with a subtransient reactance of 12% facing a 100kVA motor block dips 4.8% on the first cycle before the AVR answers: 12% x 100kVA / 250kVA = 4.8%. The engine barely notices at that instant; the dip is an electrical event inside the alternator, and it recovers ahead of any change in fuel delivery.
Cables and transformers add their own impedance in front of the set, so the dip at the motor terminals runs deeper than the dip at the alternator terminals. On long runs the difference reaches several percent, and the fix is a bigger cable, not a bigger set. Tesla Power Equipment publishes the subtransient reactance for every alternator frame it ships, so the calculation belongs in the offer, not after the start. A dip deeper than about 15% on a single motor start usually means the alternator is undersized for the starting current, so check the frame before touching the excitation settings.
Why Does the AVR Take Time to Recover the Bus?
The AVR sees the dip within a quarter of a cycle and responds by pushing field current, but the field cannot change instantly. What matters is the excitation ceiling: how much spare voltage the exciter can force onto the field winding. Shunt excitation takes its supply from the sagging bus, so a deep dip starves the very circuit that should correct it. A PMG or an AREP winding powers the AVR independently and holds the ceiling while the bus is down.
With a healthy ceiling the bus recovers inside 1.5 seconds on most export builds, and Tesla Power publishes the recovery figure beside the dip figure on the test sheet. Watch the underfrequency roll-off as well: once engine speed falls below the UFRO setpoint, often set near 45Hz on a 50Hz machine, the AVR folds voltage back on purpose to protect the load. That reads as a long sag and behaves like one.
Which Loads Ride Through and Which Drop Out?
Set the sag limits from the load, because every class of equipment lets go at a different depth. IT loads ride to the ITIC envelope, which allows 70% of nominal for up to 500 milliseconds. Control contactors release once the coil sits near 65% to 75% of rating, and a released contactor drops a running motor even though the genset was recovering. Drives trip on a DC-bus low fault inside tens of milliseconds. Review genset voltage sag in hospital projects the same way, because our healthcare clients carry both IT loads and motor loads on one bus: list each load class and its ride-through before the rating is chosen.
The table gathers the bands that matter on a mixed site. Use the tightest figure as the design limit, not the average, because the first motor to drop out defines the outcome for the whole process. The same reading settles genset voltage sag in a data room, where the ITIC line rather than the genset sets the depth limit. Where a class is missing from the datasheet, measure it during commissioning and write the number into the record.
| Load class | Band | What bounds it |
|---|---|---|
| IT loads | 70% of nominal for 500 milliseconds | ITIC envelope |
| Contactor coils | Release near 65% to 75% of rating | Coil hold-in band |
| VFD inputs | DC-bus low trip in tens of milliseconds | Drive firmware |
| Genset bus | First-cycle dip bounded by the nameplate class | ISO 8528-5 class |
How to Measure and Record a Sag
Measure at the load, not at the generator terminals, because the two ends of the cable disagree during a start. A power quality meter set to trigger at 90% of nominal catches the event and stores the depth, the duration and a timestamp. Log the load state alongside: which motors were running, which one started, and at what step. Three repeats separate a marginal start from a healthy one.
Keep the traces in the handover pack with the meter model and its calibration date. Tesla Power stores the trace pack for every factory test, so a site figure that reads odd can be compared with the bench record the same day. Write the trigger threshold into the record too, because a 90% trigger counts events that an 85% trigger hides, and read the load step limits note for the step sequence behind the traces.
Specification Fixes That Hold Under a Real Load
Fix the sag at the specification stage while the choices are still cheap. Ask for a PMG or AREP excitation option so the ceiling holds while the bus is down, and we price both options on the same quotation line so the comparison stays honest. Choose the alternator frame from the starting block, not from the running load, because the running load rarely sets the dip.
Then attack the starting block itself. A soft starter on the largest motor cuts the starting kVA roughly in half, and a staged start sequence keeps two blocks from landing together. Step the load on in bands and keep the biggest motor for the last band. A larger frame is the last resort, chosen only after our load audit says the block cannot be arranged.
Genset Voltage Sag for Export Sites: The 500kVA Frame
The 500kVA frame is the export workhorse, rated 400kW at 0.8 pf on a 400V bus, and its first-cycle dip reads the same arithmetic as any other size. A 150kVA motor block against 12% subtransient reactance dips 3.6% before the AVR answers: 12% x 150kVA / 500kVA = 3.6%. The 500kVA silent build with the ATS option ships with the dip table in the handover pack, so the site electrician sees the limit before the first start.
China genset voltage sag suppliers quote the shunt build cheap and the PMG build late, so ask for the dip table before the deposit leaves. Tesla Power lists the excitation option, the ceiling figure and the predicted first-cycle dip on one sheet, and our sales desk returns a revision if the load schedule changes. The same reading covers genset voltage sag for export sites as for home markets; the ambient, the fuel and the cable run move the numbers, not the rule.
Maintenance Causes That Mimic a Sizing Fault
Some sags are not sizing faults at all. A loose sensing lead at the AVR makes the regulator chase noise, aged diodes on a rotating exciter cut the ceiling, and an engine losing fuel pressure drags frequency down until the UFRO folds the voltage back. All three read like an undersized frame on a trend plot, and all three are cheaper to fix than a new alternator.
Make the checks routine: torque the sensing terminals, log the no-load voltage monthly and compare it with the commissioning sheet, and keep the AVR settings in the same file as the trace pack. The routine maintenance list covers the interval for each check. Tesla Power prints the AVR reference sheet into every handover pack, so a set that drifts on no-load voltage is reporting the regulator long before the load ever lands.
Records and the Acceptance Test
The acceptance record turns the arithmetic into evidence. Carry the predicted first-cycle dip, the measured dip at each load step, the recovery time and the meter trigger threshold, and name the load state for every reading. We keep the genset voltage sag documentation for the acceptance test beside the load steps, because the ISO 8528-5 load step rules are the first document a third party reads the dip figures against.
Our handover pack carries the genset voltage sag quality records for the export file in the same structure on every project, so a service technician finds the traces without a phone call. The ISO 8528-1 rating groups explain which rating the dip figures belong to, and the 1000kW Weichai container build ships its trace pack in that same structure. Tesla Power keeps one folder template for every export order, from the 120kW compact frames upward.
Frequently Asked Questions About Genset Voltage Sag
Use the tightest ride-through on the bus as the limit. As a rule of thumb, a dip deeper than about 15% on a single motor start points to an alternator undersized for the starting current, and the ISO 8528-5 class on the nameplate bounds what the set side may show.
No. The first dip is set by the subtransient reactance of the alternator, which the AVR cannot change. What the AVR controls is recovery, and a high excitation ceiling brings the bus back inside a second or two.
A second motor usually landed while the first was still drawing inrush, or the engine lost speed and the underfrequency roll-off folded the voltage back. Stage the starts and check the fuel system before blaming the regulator.
It cuts the starting kVA roughly in half, which halves the first-cycle dip, but the contactors and drives on the bus still set the limit. Check their ride-through against the new figure before signing off the design.
Inside 1.5 seconds is typical where the excitation ceiling is healthy and the block is not stacked. A recovery that stretches past 3 seconds on the same block points at the ceiling, so test the PMG or AREP supply.
Ask each bidder for three things in writing: the predicted first-cycle dip against your load schedule, the excitation option and its ceiling, and the factory trace pack. Buyers comparing genset voltage sag manufacturers on price alone inherit the shunt build and find out at the first motor start. Tesla Power quotes the dip figure beside the rating on every offer, and the export desk returns a revision when the load schedule changes.