What Is a DNP3 Frozen Counter?

Back to Blog

Why Freeze a Counter

A counter is a monotonically increasing total - meter pulses, a volume totalizer, accumulated run time. If a master simply reads the live counter, the value is a moving target: it changed between when the outstation sampled it and when the master received it, and two counters read on separate polls do not represent the same instant. For interval totals and any billing-grade reading, that inconsistency is unacceptable.

The freeze operation fixes it. On a freeze command the outstation copies each live counter into a companion frozen counter at one instant. The master then reads the frozen values, which are stable and represent a common point in time. A broadcast or multi-point freeze captures a whole set of totals simultaneously, so the numbers add up correctly across a site - the DNP3 analog to how a flow computer latches a totalizer at an interval boundary.

Freeze, Freeze-and-Clear, and Counter Rollover

DNP3 offers freeze on its own and freeze-and-clear, which snapshots the counter and then resets the running counter to zero in the same operation. Freeze-and-clear is how per-interval totals are built: each interval starts from zero and the frozen value at the end is that interval's quantity. Plain freeze leaves the running total intact, giving a cumulative reading.

One thing to plan for is rollover. A counter of finite width eventually wraps back to zero, which is a normal event, not an error - the same behavior as a counter rollover in a totalizer. When totals are computed from frozen readings across a rollover, the arithmetic must account for the wrap. Understanding freeze also means understanding that the frozen counter and the running counter are two separate objects the master can read independently, so a mismatch between them at a given moment is expected, not a fault.

Fitting Freezes Into the Polling Plan

A freeze does not happen unless something issues it, so the freeze schedule is part of the master's polling design, not an outstation default. The usual pattern is a freeze at each accounting boundary - top of the hour, end of the contract day - followed by a read of the group 21 points, all sequenced so the frozen set is safely collected before the next boundary arrives. On a congested channel that sequence competes with everything else the master does, so it pays to plan it alongside your integrity and event poll rates rather than bolting it on afterwards.

Broadcast freezes capture every outstation at once but are unacknowledged, so a careful master follows the broadcast with per-station reads and treats a missing response as a flagged interval. The freeze-at-time variants shift the trigger into the outstation itself: the device freezes on its own clock at the configured boundary, which removes the dependence on the channel being up at that exact instant - valuable on cellular or satellite links. That only works if outstation clocks are right, which is why time sync and event timestamps deserve verification before anyone trusts time-scheduled freezes.

A Worked Interval, in Symbols

Call the frozen reading collected at boundary k F(k). With plain freeze the interval quantity is the difference of successive snapshots: Q = F(k) - F(k-1). If the counter wrapped once during the interval the difference goes negative, and the correction is Q = M - F(k-1) + F(k), where M is the rollover modulus for the counter's width - 2 to the 32nd power for a 32-bit point. With freeze-and-clear there is no subtraction at all: the frozen value is the interval quantity, because the running counter restarted from zero at the previous boundary.

A reconciliation habit worth keeping: trend the live group 20 counter as well as the frozen values. The sum of your interval quantities should track the growth of the cumulative counter, and any divergence between the two is an early, cheap signal that a freeze was missed, a wrap went uncorrected, or a device restarted and reset its count.

Plain Freeze or Freeze-and-Clear: Picking One

The table contrasts how the two behave when things go wrong, which is the real basis for choosing:

Aspect	Plain freeze	Freeze-and-clear
Frozen value means	Cumulative total at the freeze instant	Quantity accumulated since the last clear
Interval arithmetic	Subtract successive frozen readings	Use the frozen value directly
A missed freeze	Next reading spans two intervals; the total is still recoverable	Two intervals merge and cannot be separated afterwards
Device restart	Visible as a cumulative total that stepped backwards	Harder to distinguish from a genuinely small interval

Plain freeze is the forgiving choice for anything with audit weight, because the cumulative total survives every mishap and the intervals can be re-derived from it. Freeze-and-clear buys simpler arithmetic at the cost of being destructive: once the counter is cleared, the history it held exists only in the reading you did or did not manage to collect.

Frequently Asked Questions