Replacing a Test System: Match the Channel Architecture, Not the Kilowatts

Replacing an instrument · 4 min read · 20 cited facts

Every enquiry about a discontinued test system leads with the same number: total power. It sits on the original purchase order, it survives twenty years of staff turnover, and it is the easiest specification for any replacement to match — which is precisely why matching it decides almost nothing. The instrument being replaced was never really an amount of power. It was an arrangement of power: divided at some voltage, into some number of simultaneous circuits, and the arrangement is what a substitution quietly changes.

The recurring failure has one shape

It repeats across instruments that have nothing else in common. The NH Research 9300 went into pack-level programmes on the strength of its channel arrangement rather than its headline power, so a substitution that copies the kilowatts and misses the per-channel current gets found out at commissioning, not on the quote. The NH Research 9200 shows the mirror image at module level: module test is throughput-bound, and a replacement carrying the same total power across half the channels doubles the qualification calendar. The schedule fails before the electronics do — and the purchase order looked correct the whole time.

One power rating, four instruments

Nothing in that table changes except the point on the voltage–current trade where the chassis was ordered. The 7 kW is constant; 80 V at 340 A and 1000 V at 30 A are different instruments for different work, and neither substitutes for the other on any bench. A replacement matched on power alone has matched the one line that carries no information about fit.

Channel count against channel size

Scale the same trade up to system level and it becomes a fork in the catalogue. The CTS4000 divides Max. 21kVA (user-definable) across eight independent channels, each spanning 0V to 110V and 0A to 3000A — a machine for serving a queue of cells and modules in parallel. The Procyon PTS 2100-20 concentrates 5kW – 480kW into one envelope of 40V to 1500V at up to 8,000A — a machine for one large article at a time. Swapping a system from one side of that fork for a system from the other matches plenty of kilowatts and none of the work, because the queue the lab actually runs was designed around the old machine's channel split.

Bidirectional systems add a second envelope to miss

Bidirectional supplies extend the same mistake into a second quadrant. They are specified on their sourcing envelope and then spend their working lives sinking — a drive under regen, a pack under discharge — so the sink-side rating and the behaviour crossing zero are what the test actually exercises, and neither appears in the headline. When a bidirectional series goes end-of-life, get the envelope in both quadrants and the zero-crossing behaviour on paper: they differ between generations of the same product family, let alone between machines that happen to share a power rating.

Demand these in writing before agreeing a swap

• Total system power and, separately, power per channel.
• Channel count, and per-channel current sustained — not peak — with the duty cycle any peak figure assumes.
• The voltage window of each channel at both ends: the top where the rating lives, and the bottom where cell-level and deep-discharge work happen.
• For bidirectional equipment: the full envelope in both quadrants and the behaviour crossing zero.
• Regeneration efficiency and the grid-tie terms your facilities team will be asked to sign.