What is an STS (Static Transfer Switch) in a data center and why it can save you from an outage when your UPS does not
The device that operates where the UPS and ATS no longer reach
An STS (Static Transfer Switch) is an electrical switching device that transfers a critical load between two independent power sources in less than a quarter cycle (typically 4-6 milliseconds), without any mechanical movement. Unlike an ATS (Automatic Transfer Switch) which uses contactors and operates in 1-10 seconds, the STS uses silicon-controlled rectifiers (SCRs, back-to-back thyristors) to make the transition. The result is a transfer so fast that sensitive electronic equipment (servers, storage, network gear) does not register the event as a power failure.
The STS is not a substitute for the UPS. It is a complement that solves a specific problem: protecting the load when one of the two power sources fails, without the transfer time of an ATS being long enough to drop the load. In a data center architecture with redundant UPS, the STS is the device that allows the load to be powered from the secondary UPS or the bypass when the primary UPS fails, without the servers detecting a voltage drop.
The problem the STS solves
In a 2N redundant power architecture, the data center has two independent UPS systems (UPS A and UPS B) feeding each rack from different sources. The ideal scenario is that both operate in parallel and the load is shared. In practice, one of the two paths can fail (UPS failure, breaker open, maintenance), and the load must transfer to the surviving path in a time that the equipment can tolerate.
Without an STS, the transfer between sources takes 1-10 seconds (typical ATS) or longer if the transfer is manual. During that time, the servers lose power and restart. The cost of an unplanned restart in a data center is high: lost transactions, corrupted data, application restarts, and SLA penalties. The STS reduces that transfer time to 4-6 milliseconds, below the ITI (Information Technology Industry Council) curve tolerance threshold that servers can withstand without registering the event as a power failure.
Typical STS architecture
A typical STS for data center applications has:
- Two inputs (Source 1 and Source 2): typically the two outputs of independent UPS systems, or one UPS and one bypass from the switchgear.
- SCRs (silicon-controlled rectifiers) in back-to-back configuration: one SCR per source per phase, allowing current flow in only one direction at a time. The control logic decides which set of SCRs is conducting at any moment.
- Microcontroller-based control: that monitors the voltage and frequency of both sources continuously, and decides when to transfer based on configurable thresholds.
- Fast transfer logic: that detects the failure of the active source and triggers the transfer to the alternate source in less than a quarter cycle.
- Output to the load: the critical load panel that feeds the racks.
- Bypass for maintenance: that allows the STS to be taken out of service for maintenance without cutting power to the load.
STS units for data center applications are typically sized from 100 A to 1600 A per phase, in three-phase configurations. Some manufacturers offer 1-phase units for smaller loads, but the data center application is typically 3-phase.
The operational difference with the ATS
The ATS (Automatic Transfer Switch) and the STS (Static Transfer Switch) solve related but different problems:
- ATS: mechanical, transfer time 1-10 seconds, typically used between utility (CFE) and emergency generator. Designed for loads that can tolerate a brief power interruption (lighting, HVAC, mechanical systems).
- STS: static (no moving parts), transfer time 4-6 ms, designed for sensitive electronic loads that cannot tolerate a power interruption (servers, storage, network gear).
The ATS is used at the input of the data center (between CFE and generator), and also at the output of the UPS (between UPS and bypass). The STS is used at the rack level, between the two outputs of the redundant UPS systems (UPS A and UPS B). They are complementary devices, not substitutes.
The transfer time difference is critical: a 1-10 second transfer (ATS) is too long for servers, which would lose power and restart. A 4-6 ms transfer (STS) is below the ITI curve tolerance, so the servers do not register the event as a power failure.
When an STS is needed
An STS is typically needed in the following scenarios:
- 2N redundant architecture at the rack level: when each rack has two independent power sources from two different UPS systems, and the load must transfer between them in less than 10 ms.
- Single-corded loads: equipment with a single power supply that cannot be connected to two sources simultaneously. The STS provides the redundancy at the input.
- Distributed redundant architecture (N+1 or N+2): when the redundancy is shared between several UPS, but the critical load needs the reliability of a fast transfer between sources.
- Compliance with Tier III or Tier IV of the Uptime Institute: Tier III requires that each load can be transferred to an alternate source in less than 10 seconds (which the STS can do in ms). Tier IV requires that the redundancy work even during maintenance, which the STS supports with its bypass.
- Critical loads that cannot tolerate any interruption: financial transaction processing systems, real-time control systems, medical equipment, industrial control systems.
The standards and classifications that apply
STSs for data center applications must meet several standards:
- IEC 62310-3: static transfer systems, specifically for STS. Defines the transfer time, synchronization criteria, and test methods.
- IEEE 446 (Orange Book): recommended practice for emergency and standby power systems for industrial and commercial applications. Covers the STS as part of the redundant power architecture.
- UL 1008: transfer switch equipment, for the safety of the device.
- IEEE 1100 (Emerald Book):
- NEC (National Electrical Code): for installation in the United States. In Mexico, NOM-001-SEDE applies based on the NEC.
Manufacturers of STS for data center applications include ABB, Eaton, Schneider Electric (ASCO), Vertiv (Liebert), and Siemens. The choice depends on the size, transfer time required, and integration with the existing power architecture.
STS sizing
STS sizing for a data center depends on:
- Total load current: the sum of the currents of all the racks that the STS will feed. Typically calculated at 80% of the nominal capacity of the STS for continuous operation.
- Voltage and phases: typically 220 V or 480 V in three-phase for data center applications in Mexico.
- Source configuration: the two sources can be in phase (synchronized, ideal for fast transfer) or out of phase (requiring a brief open transition, slightly longer).
- Short-circuit capacity: the STS must withstand the short-circuit current of the sources without damage, which depends on the source impedance and the available fault current.
- Required transfer time: typically 4-6 ms for data center applications. Some applications (medical, industrial) require 2-4 ms.
An STS of 800 A in three-phase 480 V can feed approximately 665 kW of critical load (800 A × 480 V × √3 × 0.80 = 531 kVA, ~665 kW at PF 0.8). For a 1 MW data center, two STS of 800 A in parallel are needed, each feeding half the load.
Losses and heat
The SCRs of the STS generate heat during operation. The losses are typically 0.5-1.5% of the load power, which for an 800 A STS at full load means 3-6 kW of heat that must be removed from the equipment room. The STS manufacturer specifies the cooling method (forced air or water) and the heat dissipation in the datasheet.
The heat generated by the STS is additional to the heat of the UPS and the IT load, and must be considered in the data center cooling design. An STS room with two 800 A units requires approximately 6-12 kW of additional cooling, which can be significant in a small data center.
Maintenance bypass
A critical feature of the STS for data center applications is the maintenance bypass, which allows the STS to be taken out of service for preventive or corrective maintenance without cutting power to the load. The bypass is typically a manual make-before-break switch that routes the load directly to one of the two sources while the STS is out of service.
The maintenance bypass must be sized for the full load of the STS, and must include mechanical interlocks that prevent the operator from creating a parallel between the two sources (which would cause a short circuit). The bypass is typically a separate panel adjacent to the STS, with a clear mechanical indicator of the position (BYPASS / NORMAL).
The risks that the STS does not mitigate
The STS is a critical device, but it does not mitigate all risks:
- Voltage sag on both sources simultaneously: if both power sources experience a sag at the same time (e.g., a CFE failure that propagates to both UPS), the STS cannot transfer to a source that is also in failure. The UPS downstream (battery) is the last line of defense.
- Frequency drift on both sources: the STS transfers based on voltage and synchronization, not on frequency. If the two sources have different frequency (e.g., 60 Hz and 59.5 Hz), the transfer can cause a load step.
- Overload of the alternate source: if the alternate source does not have enough capacity to take the full load, the transfer can cause an overload on that source and trigger its protections.
- Failure of the STS itself: like any device, the STS can fail. The maintenance bypass and the redundancy of the two sources are the mitigation.
- Harmonic distortion: the SCRs of the STS can inject harmonic distortion into the load, affecting the power quality. The STS manufacturer specifies the THDi (Total Harmonic Distortion of current) in the datasheet.
What your next electrical RFP should ask
If the data center needs STS at the rack level, the questions that must appear in the RFP:
- What is the transfer time of the STS at full load? Is it less than 6 ms for the critical load? Is it less than 4 ms for sensitive loads?
- What is the synchronization criterion between the two sources? Is a fast open transition required or a closed transition?
- What is the short-circuit capacity of the STS? Is it sized for the available fault current of the sources?
- What is the heat dissipation of the STS at full load? Is forced air or water cooling required?
- Does the STS have a maintenance bypass? Is the bypass mechanical or static? Is it make-before-break or break-before-make?
- What certifications does the STS have? UL 1008, IEC 62310-3, IEEE 446?
- What is the MTBF (Mean Time Between Failures) of the STS? What is the expected useful life?
- What is the O&M cost of the STS? Are there preventive maintenance kits? What is their frequency?
- What is the integration with the data center’s BMS? What communication protocols does it support? Modbus, SNMP, BACnet?
The answers to these questions determine whether the STS is sized correctly for the data center’s architecture, and whether it integrates properly with the existing power system.
Sources
- IEEE (2012). IEEE 446 Recommended Practice for Emergency and Standby Power Systems for Industrial and Commercial Applications. https://standards.ieee.org/ieee/446/5676
- IEC (2006). IEC 62310-3 Static transfer systems (STS).
- ABB (2024). Static Transfer Switches for Data Centers — Technical Guide. https://library.abb.com/d/9AKK108467A5159
- Eaton (2024). STS Static Transfer Switch Product Manual. https://www.eaton.com/content/dam/eaton/products/backup-power-ups-surge-it-power-distribution/static-transfer-switches/eaton-sts-product-manual.pdf
- Vertiv (2024). Liebert STS2 Static Transfer Switch — Data Center Power Distribution. https://www.vertiv.com/en-us/products-catalog/critical-power/static-transfer-switches/liebert-sts2/
Want to master this?
Noxtel Academy →
