Open Architecture ATE: The Test Program Should Outlive the Instrument

Test system integration · 4 min read · 9 cited facts

A test system rarely dies of failure. It dies when one instrument inside it is discontinued and the test program cannot be separated from it — the sequences call one vendor's driver, the fixture assumes one vendor's chassis, and the only quoted path forward is a new platform and a rewrite. The instrument was a component. The rewrite is the loss.

Intepro builds the other way round. Systems are assembled on universal buses — VXI, PXI, LXI, CAN, Ethernet, GPIB and Serial — rather than inside one vendor's ecosystem, so off-the-shelf instrumentation sits alongside Intepro and customer hardware in the same rack. That is the whole of the claim. What follows is why it matters, and what it costs.

The program outlives the instrument

Because the executive addresses instruments over open buses rather than through one vendor's stack, a test program is not hostage to any single box. The same program runs across testers, and a rack can be re-equipped without rewriting the test. PowerStar has been the test executive for power electronics work for over twenty years, and programs written fifteen years ago have an upgrade path to current hardware.

This is not a hypothetical risk. This site carries a whole migration cluster because instruments do get discontinued — platforms get absorbed, product lines get orphaned, and the depot still has to run the test. That argument is made at length in /insights/obsolete-instrument-replacement/ rather than repeated here. The short version: when the instrument is replaceable, discontinuation is a purchasing event. When it is not, it is an engineering programme.

What re-equipping looks like

Take a load stage. An ELR 9000 chassis sinks 3.5 kW to 10.5 kW in 3U, at up to 1500 V and up to 510 A per chassis, and joins a master-slave bus expandable to 240 kW. Control is analog and USB as standard, with GPIB, CAN, Ethernet, Modbus TCP and EtherCAT among the options — buses, not a proprietary socket. If that chassis is ever unavailable, the rack accepts another load on the same buses and the sequences above it keep running, because they were written against the test rather than the part number.

The same property is what makes scaling work. A Procyon battery test platform spans 5 kW to 480 kW and 40 V to 1500 V at up to 8,000 A, precisely so hardware is reused as programmes grow instead of being replaced when they do. Re-equipping and scaling are the same idea: the test content is separated from the iron that executes it.

What open architecture costs you

Integration work a single-vendor stack would have done for you. When everything comes from one catalogue, the vendor has pre-qualified the drivers, the triggering, the cabling and the error behaviour between boxes. On an open rack, someone has to do that per instrument — verify the driver, prove the timing, document the failure modes. Intepro does this work when it delivers a system; if you re-equip on your own later, the work is yours. Openness moves that cost from impossible-by-design to merely real, and it is honest to budget for it.