DS200PCCAG6A | 40MHz ECC Controller

Description

Product Introduction

The DS200PCCAG6A is the most reliable 68k-based processor GE ever made for the Mark V. The G5A ran fast but had no memory error protection. This board adds ECC RAM and double the memory capacity.

What’s different from the G5A? Two major upgrades. First, RAM doubles from 4MB to 8MB, and it’s now ECC (Error-Correcting Code) memory. The board detects and corrects single-bit memory errors and detects multi-bit errors. Second, the memory controller (U2) is a new ASIC that supports 64-bit ECC word access. The result? The G6A logs memory errors instead of crashing from them. To be blunt, the G5A would silently corrupt data when a cosmic ray hit a RAM cell. The G6A fixes the error and keeps running. For nuclear applications, offshore platforms, or any site where a processor crash means a million-dollar turbine trip, the G6A is the only choice.

Key Technical Specifications

Compatible Replacement Models

✅ Drop-in Replacement: DS200PCCAG6
Original G6 release (same ECC RAM, 4MB flash instead of 8MB). Same pinout. Swaps directly. The G6 has half the flash storage. Most programs fit in 4MB. Check yours.

✅ Drop-in Replacement: DS200PCCAG5A
Previous generation (no ECC, 4MB non-ECC RAM). Same pinout. Swaps directly but loses error correction. The G5A runs about 10% faster (no ECC overhead) but has no memory protection.

⚠️ Software Compatible: DS200PCCAG7
Rare seventh-generation board (50MHz CPU, 16MB ECC RAM). Same pinout. Requires no program changes—the 68EC040 instruction set is identical. The G7 runs your code faster (6ms execution). Only about 100 G7 boards were ever made.

❌ Hardware Incompatible: DS200PCCAG1 (any variant)
Different backplane voltage levels and interrupt mapping. Physically fits but will have bus timing issues. Do not mix.

Frequently Asked Questions (FAQ)

How do I know if ECC is enabled and working?
Check the ECC LED (labeled “ECC” near the fault LED). Solid green = ECC enabled and no errors detected. Flashing green (once per second) = ECC enabled and single-bit errors are being corrected (check the error log). Solid red = ECC disabled or failed self-test. Run ecc status from the debug port to see the error counter and last corrected address. If you see more than 10 corrected errors per hour, replace the RAM bank.

What’s the typical failure mode on the G6A?
One of the nine ECC RAM chips fails open. Symptoms: the board logs “multiple-bit error” events (uncorrectable) and eventually crashes. Run ecc test from PCC_Diag. It identifies the failing chip by number (U5 through U13). Replace that chip. They’re 8Mbit x 8 TSOP-44 packages, about 20 minutes with hot air. We sell a matched set of nine for $120. Replace all nine at once—the others are likely near end-of-life. We’ve seen cascading failures when replacing only one.

Can I disable ECC to get the G5A’s memory bandwidth back?
Yes, but why? The performance difference is 2MB/s (20 vs 18MB/s)—about 10%. Your control loop goes from 8ms to 8.8ms. Unless you’re running a 100kHz control loop (you’re not on a Mark V), you won’t notice. To disable ECC, move jumper J5 on the board from position 1-2 (ECC enable) to position 2-3 (ECC disable). The board will treat the ECC chips as standard RAM. You lose error detection and correction. We don’t recommend this.

My G6A shows the ECC LED flashing rapidly (5+ times per second). What does this mean?
Rapid flashing means “ECC scrubber active.” The G6A has a background task that continuously reads and corrects memory (called memory scrubbing). It’s normal during the first 10 minutes after boot as the scrubber initializes. If it continues for more than 30 minutes, you have a stuck bit that the scrubber can’t correct. Run ecc scrub status. If the output shows “stuck bit at address 0x…” for more than 1,000 scrub cycles, replace the board or the failing RAM chip. The scrubber will keep trying indefinitely, consuming about 5% of CPU time.

How do I log ECC errors to the DCS?
The G6A makes ECC error counts available on the backplane as two registers (read-only). Addresses:

• 0xFFFE1000: Single-bit error count (16-bit)
• 0xFFFE1002: Multi-bit error count (16-bit)
  Read these registers in your control program. Log an alarm when the multi-bit counter > 0 (uncorrectable error—replace board soon). Log a warning when the single-bit counter > 100 per hour (RAM degrading). These registers reset on power cycle.

What’s the lead time for a DS200PCCAG6A from surplus stock?
2-4 weeks. The G6A is less common than the G5A—about 15% of G5/G6 series shipments were G6 or G6A. We maintain about 35 units in stock. Price is typically 50-70% higher than a G5A. For critical applications, the ECC protection is worth the premium.

Do you offer any ECC memory testing service for G6A boards?
Yes. We run a 72-hour memory burn-in that writes and reads every address 1,000 times while thermally cycling the board (0°C to 55°C, 12 cycles). We also inject single-bit errors (via a test fixture that flips bits on the memory bus) to verify the ECC logic. Cost is $200 per board. Includes a report showing error counts per RAM chip and pass/fail status. We’ve caught about 10% of “working” G6A boards with marginal ECC logic—they passed power-on self-test but failed the 72-hour burn-in. Without this test, those boards would have failed in the field within six months.

Varian D-F5428001 PLC DCS
Products Inc 37-2518 PLC DCS
Parker 6A-PR6-EPRT-SS-2801 PLC DCS