Data center transformers

The four families, and what actually separates them

Transformer selection for a data center is rarely one decision. Power arrives at utility medium voltage and has to reach IT equipment at utilisation voltage, and different parts of that path call for different machines. Picking on kVA alone — the most common shortcut — ignores the two things that usually cause trouble later: impedance, and where the unit is physically allowed to sit.

Specifications that decide the selection

Four numbers do most of the work. Getting any of them wrong tends to surface late, when the fix is expensive.

Selection criteria in practice

• Where can it physically sit? Liquid-filled units outdoors need containment and clearances; indoors they bring fire-separation requirements that often make dry-type the only workable answer regardless of lead time.
• What is the available fault current downstream? Settle this before ordering switchgear, not after. It is set by the transformer's impedance and the utility's contribution, and it decides the interrupting rating you must buy.
• Is the load harmonic-heavy? Rectifier front-ends and older UPS topologies produce harmonic currents that heat a standard transformer beyond its rating. That is what K-rating exists to address.
• Does efficiency regulation apply? DOE efficiency rules constrain what can legally be sold for many distribution ratings, which narrows the field more than buyers expect.
• What is the redundancy target? N+1 at the transformer level changes both the count and, often, the individual unit rating.

Typical applications

A conventional build uses a pad-mount transformer at the utility interface, feeding main switchgear at 480V. Inside the building, dry-type units step down again for mechanical and house loads. K-rated isolation units appear where harmonic-producing equipment is concentrated. Large power transformers only enter the picture on campuses large enough to interconnect at transmission voltage — at which point the lead time becomes the single longest pole in the project schedule.

Work the numbers before you specify

Related reading

Common questions

Which transformer type do data centers actually use?

Usually more than one. A pad-mount steps utility medium voltage down to the voltage the building distributes; dry-type or cast-resin units sit indoors closer to the load; large power transformers appear only where the campus takes power at transmission voltage; K-rated isolation units handle harmonic-heavy loads.

Why does impedance matter so much?

Because it sets the available fault current downstream, and that decides the interrupting rating on your protective devices. A lower %Z gives a stiffer source and a higher fault current. Selecting a transformer on kVA alone and discovering the impedance implication after the switchgear is ordered is a recurring and expensive sequence.

How far ahead should orders be placed?

Against a preliminary specification rather than a finished design. With pad-mount units indicatively around 85 weeks and large power transformers toward 160 weeks, waiting for a locked design can add a year or more to the schedule.

Can a dry-type unit substitute for a pad-mount to save time?

Rarely. They solve different problems — indoor air-cooled versus outdoor liquid-filled, and different voltage classes. The much shorter dry-type lead time makes the substitution tempting, but the voltage class and installation location usually rule it out.

Family reference pages

Each family below has its own page: how it is specified, the standards it is built to, its indicative lead time and market price band.