Power Repair & Protection

WangPS (opens in new tab)

Volt Clamps for High-Risk Power Rails

The +5 and -12.5 mentioned above are High-Risk Power Rails because overvoltage on either could be catastrophic:

• the +5v rail supplies around 250 TTL ICs. DoPECC has another TTL-based machine that has suffered a catastrophic supply overvoltage and many chips have actually ruptured, leaving only the metal legs and some plastic fragments. No doubt many others will be dead even if they are physically intact.
• the -12.5v rail supplies rare first generation MK4008 1k x 1 MOS DRAMs. These hard-to-replace ICs are at risk from over(?under)voltage.

Checking the schematics shows that both supplies are regulated with 723 first-generation regulator ICs and series-pass transistors to perform the actual voltage dropping. Series pass transistors are under some stress as they must dissipate the I²R involved in dropping the input voltage to regulated level. If these transistors fail in short circuit mode, full unregulated voltage will be applied to the downstream components.

Such a failure mode can be mitigated by fitting active voltage clamps to the relevant rails. If overvoltage occurs the clamp will shunt the rail to ground, protecting the downstream components and dissipating the overvolt energy until the protective fuse blows. DoPECC prefers clamps of the style discussed in AoE Chapter 9x - see the Equipment section of this site for a more detailed discussion of these clamps.

Voltage clamps were fitted to the +5v and -12.5v supplies.

RAM Negative Suppply Rail Value

While researching the supply rail limits to assist with design of the voltage clamps it was noted that the Wang schematics specify -11.5v for the MK4008 RAM ICs. Referring to the Mostek datasheet it was found that the MK4006 has maximum supply voltages of +4.75/-11.4 while the MK4008 has maximums of +5.25/-12.5.

With this in mind the RAM negative supply was adjusted to -12v with the clamp set to -12.5v. These close tolerances are possible with a voltage-reference IC based active clamp, simple zener diode clamps have a very soft knee and cannot achieve this level of precision.

-12.5v RAM Failure

Shortly after the regulated supplies were brought back up, the -12.5v RAM supply failed. Testing revealed that the driver side of the output Darlington pair had failed. This is marked as a GT544 and schematics show a Ge PNP device but this type code could not be found in any datasheet or substitution search. A 2N465 was substituted as a generic sort of Ge PNP device. The 2k2 bias resistor was tweaked but ultimately the same 2k2 value seemed to put the transistor into its active range.

Restoring the output stage connection to the 723 regulator did not restore the RAM suppply. Measurements around the 723 showed reasonable values for the voltage reference and the error amplifier output but the internal driver transistor appeared to be defective. The 723 regulator IC was first introduced nearly sixty years ago but surprisingly it is not too difficult to find replacements. Some devices that appeared to be legitimate Motorola NOS were obtained and fitting a replacement brought the RAM supply back to life and on-specification.

+12v Regulator Upgrade

The Wang schematic for the 341 Regulator Board has an ECN that upgrades the output ampacity of the +12v regulator by adding a series pass transistor to the original very basic dropper/limiter resistor and zener reference. DoPECC’s machine is an early number and the 341 did not have the active elements to increase the ampacity of the +12v supply.

The simplest solution was to disable the simple zener regulator and replace with a 7812, which fitted pretty neatly on the rear of the 341 board.

Power Supplies Restored

Finally, all supplies were restored and stable at appropriate voltages.