Two Times I Chose the Wrong GE Transformer Protection (and one I almost did)

I learned this the hard way. Twice.

In my first year working with substation upgrades—2017—I made a mistake that cost about $3,200 in rework and a week of delays. I'd just spec'd a GE Multilin 850 protection relay for a 15 MVA transformer, but I skimped on the auxiliary power module. Saved maybe $180. The relay dropped out during a brief voltage sag. Net loss: $3,200 for the replacement module, emergency field service, and extended downtime. My boss still brings that one up.

That's the problem with transformer protection: it's easy to focus on the relay's fancy features and forget that the foundation matters more than the bells. If the power supply fails, none of those advanced protection algorithms do a thing.

The surface problem: picking the wrong relay model

Most engineers I talk to think the biggest risk is choosing a relay that doesn't have enough protection functions. They compare fault detection speeds, harmonic filtering specs, and communication protocols. And sure—those matter. But in my experience, that's not where most failures happen.

The real problem is usually something far more basic: voltage rating mismatch, incorrect CT/VT wiring, or—like my 2017 mistake—an underspecced power supply.

I went back and forth between the GE Multilin 850 and a simpler model for two weeks. The 850 offered advanced arc-flash detection and better event logging. The simpler model saved about $400 per unit. On paper, the simpler model made sense for a standard distribution transformer. But my gut said to future-proof the installation. I compromised by buying the 850 but cheaping out on the power supply. Classic penny-wise, pound-foolish.

What I missed: the protection system is a chain

A transformer protection system isn't just the relay. It's the relay plus:

• Current transformers (CTs) and potential transformers (PTs)—everything must match the relay's nominal input ratings.
• Power supply—must handle voltage sags, transients, and ambient temperature extremes.
• Wiring and termination—one loose connection can disable the entire scheme.
• Settings and logic—incorrect pickup values or time delays can make the relay useless.

My 2017 failure was a broken link in that chain. The relay's power supply was spec'd for 120 VAC, but the station had frequent dips to 95 V. The relay would brown-out and reboot mid-fault. Three events logged as 'loss of communication' when the real issue was the relay wasn't even powered up.

The deeper problem: protection coordination isn't just about the relay

If you've been in this industry long enough, you've probably seen a mis-coordination event: a downstream fault takes out an upstream transformer because the time-current curves weren't properly matched. That's what happened on a project in September 2022.

We'd installed a Prolec GE transformer at a small industrial facility. The primary protection used a GE Multilin 845 relay. We'd set the overcurrent elements based on the transformer's nameplate data and a standard coordination study. Looked fine on paper. But—here's the catch—the downstream switchgear had old circuit breakers with significantly slower trip characteristics than we assumed.

The old circuit breaker types (a mix of thermal-magnetic and older molded-case breakers) had trip curves that shifted with temperature. On a hot day, the breaker took 30–40% longer to clear a fault. That meant our relay's backup protection operated before the breaker cleared—taking the whole transformer offline for a downstream feeder fault. We fixed the coordination after the fact, but not before the facility lost production for a 6-hour evening shift.

What you should actually worry about

Based on the mistakes I've documented, here are the gotchas I see most often:

• CT saturation under fault conditions — especially with older CTs or those that are too small for the relay's burden. Check the CT's knee-point voltage against the maximum fault current.
• Incorrect VT connection — wye/wye vs. wye/delta makes a huge difference in voltage-based protection functions. I've seen relays that operated incorrectly because the VT polarity was reversed on one phase.
• Relay logic complexity — the GE Multilin 850 allows incredibly flexible logic. But that flexibility means you can easily create unintended interactions. I've seen schemes where the undervoltage element inadvertently disabled the overcurrent element during a motor start.

One more that I almost missed: the transformer's own inrush current. A standard overcurrent setting that works for normal loading may trip on transformer energization. The solution is to use harmonic restraint or second-harmonic blocking—both available in the Multilin 850—but only if you enable them. I forgot to do that on a commissioning in 2020. The relay tripped on the first energization attempt. Embarrassing. Cost us a half-day of commissioning time.

The cost of getting it wrong

Let's put dollar figures on these mistakes, because that's what ultimately gets management's attention:

• My 2017 power supply error: $180 saved, $3,200 spent on replacement + labor + lost time. Net loss: ~$3,000.
• The mis-coordination event (2022): One unplanned outage for 6 hours. The facility estimated lost production at $15,000. Plus our service call to re-evaluate settings: $1,200. Total: ~$16,200.
• The missed inrush flag (2020): Half-day delay for the commissioning team—roughly $600 in wages, plus the frustration of re-energizing under scrutiny. Hard to quantify, but not trivial.

And here's what I've learned: none of these were about the relay being bad. The GE Multilin 850 and 845 are excellent products. The problem was me—or, more precisely, the way I spec'd and configured the complete system. The relay is just one component. The protection scheme is a system, and every part has to work together.

What I do differently now (short version)

I'm not going to give you a lengthy checklist here—that's a whole other article. But if you take away three things from my mistakes, make them these:

1. Test the whole system, not just the relay. Simulate faults that stress the CTs, check power supply voltage under sag conditions, verify that every connection is tight and correctly wired. A functional test of the relay alone doesn't prove the protection works.
2. Pay attention to the 'boring' parts. The power supply, the wiring terminal block, the CT wiring polarity—these cause more failures than a relay's internal logic error ever will.
3. Don't assume the old stuff is compatible. Old circuit breakers may have much slower trip times than you expect. Verify coordination against actual curves—not what you think they should be.

I recommend the GE Multilin 850 for any transformer application where you need advanced protection functions and robust event recording. But—here's the honest part—if you're not willing to invest the time to properly engineer the support system around it, you might be better off with a simpler relay that has fewer options. A well-configured basic relay beats a misconfigured advanced one every time.

The bottom line: transformer protection isn't about picking the fanciest relay. It's about making sure every part of the protection chain works, every time. I learned that lesson the expensive way—twice.