−48V grounding in a lithium retrofit: what changes and what doesn't

Nothing about lithium requires touching the ground point. What actually changes is that a second monitor — the BMS — is now watching the same wiring.

VOLT48 engineering desk · Published 24 Aug 2026

Key figures at a glance

Parameter	Value	Where it comes from
Grounding convention on −48V DC plants	Positive terminal grounded, single point; −48V is the distributed conductor	Industry convention
What a battery swap changes about it	Nothing — the ground point is set by the plant's distribution, not the battery	Site design
What a lithium retrofit adds	A BMS running its own isolation / ground-fault check on the string	Manufacturer datasheet
What to test at commissioning	Insulation resistance between the new string and chassis ground	Site commissioning practice
Where a false alarm usually comes from	Two independent ground-fault monitors with different thresholds, not an actual fault	VOLT48 compatibility audit

Why −48V plants ground the positive terminal, not the negative

Telecom DC plants conventionally ground the positive rail, which leaves the distributed −48V conductor as the one carried around the site. The main reason is corrosion, not electronics: grounding the positive terminal makes the exposed copper cathodic relative to earth, which slows the electrolytic corrosion that a grounded negative rail would accelerate on any buried or lightly protected run. This is a property of the plant's rectifiers and distribution — decided once, when the site was built — and it has nothing to do with which battery chemistry happens to be hanging off the bus that week.

The ground point doesn't move — but a second monitor shows up

A lithium pack brings its own BMS, and most BMS units run their own isolation or ground-fault check across the string's DC terminals, independent of whatever ground-fault alarm the plant controller already watches at the distribution level. Adding that second monitor is not a wiring change and it does not touch where the plant is actually grounded — but it does mean two independent devices are now watching overlapping wiring, each with its own sensitivity.

What actually needs re-checking on commissioning

The step worth doing on any retrofit, lithium or otherwise, is an insulation resistance (megohm) test between the new string's terminals and chassis/frame ground before the string is energised — the same step a competent installer already runs when commissioning a new VRLA bank. It is what catches a pinched cable, a BMS harness chafing against the rack, or a control lead routed across a sharp edge: none of these are lithium-specific failures, but a new install is exactly when they get introduced.

When the plant, not the battery, is the actual variable

A handful of older or non-standard sites were built with a different scheme — floating/isolated rather than grounded, or in rare cases grounded the other way. That is a fact about the specific shelf and its distribution, established the same way a controller's firmware version is: by looking, not by assuming. It belongs on the same short list as the other things a compatibility audit records before a battery is ordered, because a plant-level fact like this doesn't change with the battery — it has to be known before the order, not discovered after.

Equipment this guide covers

Common questions about this family

Do I need an electrician to re-ground anything when installing lithium?

Not for the ground point itself — it stays exactly where the plant's own distribution already put it, and a battery swap doesn't move it. What is worth having checked is the insulation resistance test after the new string goes in, the same step any competent installer already runs on a new VRLA bank.

My BMS shows a ground fault the plant controller doesn't. What's wrong?

Often nothing is actually faulted. The BMS and the plant controller are two independent monitors, each with its own threshold, watching overlapping wiring — a difference in sensitivity between them can show up as one alarming while the other stays quiet. Confirm with an insulation resistance test before assuming a wiring defect.