A 52-cow tie-stall, an old 100A service, and a quiet stray-voltage problem.

The call

The farmer called in mid-August. The complaint was the bulk-tank chiller had tripped its breaker three times in four weeks; he reset it each time and milk had not spoiled, but he didn't trust the panel. The panel was a Federal Pacific Stab-Lok 100A, installed (according to the inspection sticker) in October 1962. We told him we would not work on a Stab-Lok panel¹ as a service-call repair; we would price a panel changeout and a service upgrade, and we would do that work next.

We were on site for the site visit on August 28th. The meter base was the matching FPE bracket; the main service entrance was 2/0 aluminum that had not been re-torqued in our lifetime; the equipotential plane in the milking parlor (installed 1981, re-bonded once around 1999 according to the previous owner's notes) was cosmetically intact. We did a stray-voltage baseline:

Measurement point	Reference	V AC, baseline
Cow-stand stainless rail	Equipotential plane	0.18 V
Bulk-tank stainless body	Earth (drive rod)	0.34 V
Wash-tank galv. frame	Earth (drive rod)	0.41 V
Milk pipeline (stainless)	Equipotential plane	0.22 V

The cow-stand rail measurement was within the < 0.3 V threshold we'd want for a happy herd. The bulk-tank reading was the surprise: 0.34 V from the tank body to a driven reference rod meant the tank was, electrically, slightly hot — small enough not to bother the chiller's contactor, large enough to suggest the bonding from the tank to the building grounding electrode system had degraded somewhere between the tank and the panel. The chiller breaker tripping was a downstream symptom; the loose bond was the cause.

What we found inside the panel

Day 1 of work, with the customer's signoff to interrupt power for an hour, we opened the FPE 100A. The grounding electrode conductor was a #6 stranded copper, terminated in a split-bolt to the meter-base ground bar. The split-bolt was visibly corroded. We megger-tested the GEC: 0.4 MΩ to a driven reference rod, when it should have been < 1 Ω. The bond from the panel to the equipotential plane bus was also a split-bolt; same condition.

This is the failure that explained the bulk-tank measurement. The grounding electrode system was carrying load currents because the neutral wasn't getting back to the utility transformer cleanly; load currents through the bonding system raise local potentials on bonded equipment; the bulk tank was reading 0.34 V because of it.

The plan

1. Schedule the cutover for a Saturday morning between milkings (window: 06:30 to 11:30).
2. Pre-stage the new 320A combo meter-and-main next to the existing FPE.
3. Pre-pull #4/0 SE cable to the new feeder to the dairy sub-panel and to the dwelling.
4. On the day: kill power, swap the meter base, set the new panel, restore power. Re-feed circuits over the following ten days, one per evening.
5. Re-bond the equipotential plane bus directly to the new panel's grounding electrode bus with a #2 unbroken copper, irreversible-compression terminations on both ends.
6. Drive a third ground rod (the existing two were a single-rod-with-jumper config, marginal); place the third 6 ft from the second, brought to the panel as a separate #6 GEC.
7. Coordinate with Green Mountain Power on the drop swap. Their crew came that Saturday at 09:00; the new drop was on at 09:35.

What we wired

Service

320A combination meter-and-main, single-phase 240 V, two 200A main breakers feeding two 200A sub-panels (dairy / dwelling).

Dairy sub-panel

40-space Square D QO Plug-on Neutral, 200A main lug. AFCI/GFCI on dwelling-equivalent circuits; GFCI on every milkhouse and parlor receptacle.

Vacuum pump branch

240V single-phase 30A, with a soft-start retrofitted on the existing 5 HP DeLaval pump. Soft-start cut inrush from 96A to 24A.

Bulk-tank chiller branch

240V single-phase 30A, dedicated, with a Type-2 SPD on the breaker.

Milk-pipeline pulsator

120V on a separate breaker from lighting, GFCI.

Equipotential plane re-bond

#2 bare copper from plane bus to panel ground bus; 4 ferrules replaced; existing #4 grid retained and continuity-tested at every cow stand.

Type-1 SPD

at the meter-and-main, 10 kA per mode, UL 1449.

GEC

#4 to three driven 8-ft Cu-clad rods spaced 6 ft. Soil resistivity measured 320 Ω·m (granite-rich till); third rod brought us to 12 Ω earth resistance, well within target.

Stray voltage, after

Measurement point	Reference	V AC, after	Δ
Cow-stand stainless rail	Equipotential plane	0.09 V	−0.09
Bulk-tank stainless body	Earth (drive rod)	0.06 V	−0.28
Wash-tank galv. frame	Earth (drive rod)	0.08 V	−0.33
Milk pipeline (stainless)	Equipotential plane	0.05 V	−0.17

The bulk-tank reading dropped from 0.34 V to 0.06 V — well inside the < 0.1 V we like to see. The chiller breaker has not tripped in the 32 weeks since. The herd's average milk-letdown was up 4.1% in the next two milking weeks per the farmer's records, though we won't claim credit; weather, feed, and a herd-management change (a calf weaning shift) all happened in the same window.

Surprises

• The 1962 FPE bracket had been wet on the back side for some indeterminate number of years. We discovered it on removal: the back of the bracket was a layered green-and-white corrosion that suggested chronic moisture infiltration through the riser conduit. We re-flashed the new install with EPDM sleeves and a drip loop.
• The original 1981 equipotential grid was still intact and bonded — credit to the previous electrician. Only the bus terminations had degraded.
• The customer's wife's piano (in the farmhouse 200 ft away) had been making a faint hum on a particular F-sharp for two years. After the GEC and SPD work, the hum stopped. She mentioned it. We told her it was probably a coincidence, but we don't know.

Final billing

Labor (master + journey + apprentice)

96 hrs · $9,648

Materials (panel, SE cable, SPD, ground rods, EPDM, ferrules)

$3,210 + 12% handling = $3,595

VT DFS permit

$185

Working-farm discount (8% labor, ≥ 8 hr job)

− $772

Total

$12,656

Customer signed off, paid net-15, and asked us to come back in October 2026 for the annual re-bond on the equipotential plane. That will be a small J-02 follow-up and is on the calendar.