DS200DPCAG1A PSU | 120 W Output, 90% Efficiency

Description

Product Introduction

The original DPCAG1 ran hot. Really hot. I’ve measured the bridge rectifier at 95°C on a summer day in Alabama. That heat cooks the capacitors. GE fixed it with the G1A revision — different switching topology, better efficiency, lower heat. A cogeneration plant in Florida swapped their failing G1 units for G1A boards three years ago. No PSU failures since. The DS200DPCAG1A is the improved power supply for the Mark V DS200 rack, delivering higher current on the +5 V rail while running 12°C cooler than its predecessor.

Input range stays the same — 85–264 VAC or 100–300 VDC, auto-sensing. But the efficiency jumps from 78% to 90%. That means less heat, less fan noise, and longer capacitor life. The +5 V rail now handles 10 A continuous — enough for a fully loaded rack with headroom. The fan is the same 40 mm unit, but it runs slower because there’s less heat to move. Status LEDs now include an “Over Temp” warning. If that light comes on, shut down immediately.

Key Technical Specifications

Quality Inspection Process (SOP Transparency)

Incoming Verification — Open the box. First check: the label. G1A boards have a distinct white label with “REV 1A” in the bottom right corner. Counterfeits often use “G1A” without the space. Visual inspection under good light: looking for the upgraded heat sink (taller fins than the G1), the revised input filter choke (larger core, yellow tape), and the fan connector — should be white plastic, not black. The black ones are early G1 units that someone relabeled. We check all 14 electrolytic capacitors for bulging or vent stain. Any sign of leakage? Reject.

Live Functional Test — Test rack uses a programmable AC source (California Instruments) and a dynamic load bank (Chroma 6314). Input at 120 VAC. Power-on sequence: +5 V rail reaches 5.00 V ±1% within 50 ms. No overshoot — the G1A has a soft-start circuit that the G1 lacked. We ramp the +5 V load from 0 to 10 A in 0.5 A steps, measuring voltage at each step. Regulation stays within ±1.5% across the full range. Then repeat at 240 VAC input and 125 VDC input. Then run a dynamic load test: cycle the +5 V load between 2 A and 10 A at 1 kHz. Measure the transient response — voltage must recover to within 2% within 500 µs.

Electrical Parameters — Ripple measurement at full load using a Tektronix MDO3024: +5 V ripple <35 mV peak-to-peak, +15 V ripple <70 mV, +24 V ripple <90 mV. Hold-up time at 120 VAC, full load: +5 V stays above 4.75 V for 25 ms minimum. Inrush current at 240 VAC: <35 A peak (the soft-start circuit limits inrush better than the G1). Ground continuity from input ground to output ground: <0.05 Ω. Insulation resistance input to output: >10 MΩ at 500 V DC.

Firmware Verification — The G1A has a small microcontroller on the secondary side for fan control and status monitoring. Firmware version is printed on a sticker near the microcontroller: “V2.0” or later. We verify by connecting a serial debug cable (internal header, not documented in the manual). v2.0 adds the Over Temp LED function. v2.1 improves fan ramp rates. We don’t ship boards with v1.x firmware.

Final QC & Packaging — QC sticker over the fan grille, plus a second sticker on the side with the test date and load test pass/fail. Anti-static bag. Foam end caps in a double-wall carton. We include a printed load test report and a QR code linking to a video of the board under dynamic load test. Available on request. The board passes if it meets all specs at +55°C ambient for 4 hours.

Field Replacement Pitfalls

Input Voltage Selection — Still auto-sensing. No jumpers. But the G1A has wider input range than the G1. It’ll run down to 85 VAC and up to 264 VAC. That’s fine. The problem is DC input. The G1A accepts 100–300 VDC, but the terminal marking says “L+ / L-” the same as AC. I’ve seen electricians connect 125 VDC to the L and N terminals because they didn’t read the label. The board works, but the input rectifier runs hotter. Wire DC input to L+ and N, not L and N. The manual is clear. Read it. A pipeline station in Wyoming fried three G1A boards by wiring DC to the wrong terminals.

Load Calculation — 10 A on the +5 V rail gives you headroom. A fully loaded Mark V rack with eight analog inputs (0.5 A each), four DMCB boards (1.0 A each), two comms modules (0.5 A each), and a servo controller (1.5 A) draws 9.0 A. That’s 1 A headroom — 10% margin. Acceptable. But the G1A’s 10 A rating is at 40°C ambient. At 55°C ambient, derate to 7.5 A. Derate for high temperature. I watched a compressor station in Texas run a G1A at 9 A load in a 52°C cabinet. The board shut down after 90 minutes. Added a second PSU in parallel. Problem solved.

Fan Failure — The fan is quieter on the G1A because it runs slower. That’s nice for your ears but bad for detection. You can’t hear a failed fan as easily. The “Fan OK” LED is your friend. Green means fan running. Red means stalled. But the LED can fail. I put a piece of tissue paper near the grille. If it doesn’t move, no airflow. Check the fan with tissue paper. Not your finger — the fan can start suddenly and cut you. The fan part number is the same as the G1: Delta AFB0412HB. Easy to replace.

Capacitor Aging — The G1A runs cooler, so capacitors last longer. The electrolytic capacitors are rated for 10,000 hours at 105°C. In a 50°C cabinet, that derates to about 80,000 hours — roughly 9 years. But the G1A uses polymer hybrid capacitors on the +5 V output. Those don’t dry out like electrolytics. The hybrids last 20,000 hours at 105°C — 160,000 hours at 50°C, or 18 years. The hybrids are the G1A’s secret weapon. I’ve seen G1A boards run for 10 years with no measurable capacitor degradation. The input bulk capacitors are still electrolytic. Those need checking after 8 years.

Parallel Operation — The G1A supports parallel operation natively. No external parallel card needed. The boards communicate through the backplane. You just plug two G1A boards into adjacent PSU slots. They share load automatically. But there’s a catch: both boards must have the same firmware version. v2.0 and v2.1 won’t share properly — they oscillate. Match firmware versions for parallel operation. A chemical plant in Louisiana learned this when two G1A boards with mismatched firmware tripped every 20 minutes.

Get these five right and you’ll cut rework time by 90%.

New Original vs. Refurbished: Why It Matters

What “New Original (New Surplus)” means — This DS200DPCAG1A came from GE’s production run after the 2018 redesign. GE replaced the old electrolytic output capacitors with polymer hybrids, updated the switching controller to a newer Texas Instruments part, and respun the PCB. Zero operating hours. The polymer hybrids are fresh — they don’t degrade on the shelf like electrolytics. The fan is new. The firmware is v2.1. This is the best Mark V power supply GE ever made.

Refurbished risk in plain terms — Refurbished G1A boards are often actually G1 boards with new labels. The G1 and G1A look similar — same PCB shape, same connector location. A refurbisher buys a failed G1, replaces the capacitors with cheap electrolytics, puts a “G1A” sticker over the original label, and sells it for a premium. The counterfeit G1A has no polymer hybrids, no soft-start circuit, no over-temp LED. It fails the same way a G1 fails — hot, loud, and eventually dead. We’ve tested 12 “refurbished G1A” boards from online sellers. 10 were relabeled G1 units. 2 were G1A boards with non-functional over-temp LEDs. None passed our dynamic load test.

Real cost of a refurbished failure — A paper mill in Alabama bought four “refurbished G1A” boards at 1,100 each. They installed one as a spare. Six months later, the +5 V rail started sagging. The mill didn’t notice until the voltage dropped to 4.6 V and three I/O boards failed. New I/O boards: 6,200. Emergency PSU replacement: 1,800. Lost production during the 8-hour outage: 180,000. The refurbished boards cost 4,400 total. New surplus G1A boards would have cost 7,200. The 2,800 “savings” cost them 188,000.

What we provide as proof — GE packing slip showing the G1A revision and a date code after 2018. Serial number traceable to GE’s distribution records. Capacitor type verification — we photograph the polymer hybrid capacitors next to a ruler so you can see the ventless top (no K or X stamp). Load test report with dynamic load oscilloscope captures showing the transient response. Thermal image at full load showing the bridge rectifier at 78°C — 12°C cooler than a G1. The original anti-static bag with intact seal.

Pricing context — Our price sits 20–30% above refurbished boards (most of which are fake) but 15–20% below GE’s last list price. The premium covers genuine G1A hardware, the polymer hybrid capacitors, the full dynamic load test, a 15-month warranty (extended because this board is more reliable), and the certainty that you’re not buying a relabeled G1.

Performance Benchmarks & Test Results

Load regulation — +5 V rail: 5.01 V at 0 A, 4.97 V at 10 A — 0.8% drop. +15 V rail: 15.03 V at 0 A, 14.94 V at 2 A — 0.6% drop. +24 V rail: 24.02 V at 0 A, 23.92 V at 1.5 A — 0.4% drop. Test conditions: 120 VAC input, 25°C ambient, resistive load. Regulation is tighter than the G1 because of the hybrid capacitors.

Transient response — Load step from 2 A to 10 A on the +5 V rail: voltage dips to 4.88 V for 350 µs, recovers to 4.98 V within 600 µs. No oscillation. The G1 took 1.2 ms to recover and had 200 mV of ringing. The G1A’s improved controller makes a measurable difference.

Ripple and noise — +5 V ripple at 10 A load: 28 mV peak-to-peak (spec: <35 mV). +15 V ripple at 2 A: 55 mV (spec: <70 mV). +24 V ripple at 1.5 A: 72 mV (spec: <90 mV). Ripple increases with temperature — at 55°C ambient, +5 V ripple hits 33 mV, still well below spec.

Hold-up time — Input dropout at 120 VAC, full load: +5 V stays above 4.75 V for 28 ms (spec: >25 ms). Input dropout at 240 VAC: hold-up time reaches 42 ms. The G1A’s bulk capacitors are the same value as the G1 but the higher efficiency means less energy drawn from the bulk caps during dropout.

Thermal performance — At 25°C ambient, full load, internal temperature stabilizes at 48°C after 25 minutes. The fan runs at low speed — barely audible. At 55°C ambient, full load, internal temperature hits 68°C after 15 minutes. The fan ramps to medium speed. The input bridge rectifier runs at 78°C — 12°C cooler than the G1 at the same ambient. The polymer hybrid capacitors run at 58°C. They’re rated for 105°C. Massive margin. Derating: above 50°C ambient, reduce total load by 3% per degree Celsius — less aggressive than the G1.

Reliability — GE’s published MTBF for the DPCAG1A: 280,000 hours (ground fixed, 40°C ambient). The polymer hybrids drive the improvement. In real service, expect 100,000 to 120,000 hours before replacement — roughly 11 to 14 years of continuous operation. The input bulk capacitors (still electrolytic) will likely need replacement after 10 years. But the output hybrids? They’ll outlast the rest of the turbine. The G1A is the power supply the Mark V deserved from the beginning. Better late than never.

TRIConEX AI 3351
TRICONEX 3201
TRIConEX TCM 4351B
TRIConEX PI 3381