Circuit Details

Having seen inside the unit, the following diagram should make sense.

Wiring diagram (13K)

As you saw earlier, the internal wiring is quite simple, and removing the front or rear panels for access is very easy. That's the payback for taking your time at the planning stage - it would be a total mess if it wasn't for the ribbon cables. It really is worth the effort, and doesn't really require that much extra work at the design stage. Just keep a piece of paper with all the signals and pin allocations...

There's probably little value in looking at the Logic board, as that is standard CMOS stuff. The same applies to the Meter board, as that is just an implication of the ICL7107 data sheet - I used an external reference because the internal references are affected by the dissipation of the LED drivers - a trap for the unwary (in other words, read the datasheet!). The Logic PSU is simply a pair of 7805/7905 regulators. When I finish building the desk in the office, I'll be able to get back to the drawing board (quite literally!) and finish the schematics - I'll perhaps post scans of them here if anyone is interested.

The Control circuit is worth a look:

Block diagram (9K)

It's a symmetrical design, using NPN pass transistors for the positive supply, and PNP for the negative. The two halves are identical, so only the positive supply is discussed.

The pass transistors consist of a Darlington driver feeding a parallel pair of power bipolars, meaning that the voltage at their input is roughly 1.8 to 2.3 volts higher than the output rail. IC1a controls the output voltage by sinking current via the CV LED. The current source has two benefits - as well as providing consistent operation over the whole range of output voltage, it ensures the LED's are always of uniform brightness.

The loop formed by IC1a and the pass transistors have 100% voltage feedback, meaning that point 'X' is exactly equal to the output voltage in CV mode. This approach ensures minimum output impedance (i.e., good load regulation) at all output voltages - it's dubbed 'Regulator with a regulator' by my 1984 Motorola Linear/Switchmode Voltage Regulator Manual.

IC1b is a summing amplifier, meaning that the voltages supplied by both course and fine controls are added together. The value of Rf is pre-set to ensure the maximum output voltage is exactly 18V, and Ri and Ri1 are different values to provide the different control ranges. This configuration makes remote programming an option.

The tracking switching is on the inverting inputs of IC1b and IC2b. When selected, the potentiometers are switched away from the virtual earth, and Rt provides the input. Because each summing-amp is inverting, the negative voltage at the output of IC2b is inverted and appears at point 'X'. The value of Rt is adjustable to ensure extremely accurate operation. This is so much easier than the normal approach, and has the benefit of independent current-limiting on each rail. The commercial designs that I have seen are slightly odd - an overload on the master rail will bring down both rails, whereas an overload on the slave only affects the slave voltage.

My tracking scheme is inherently simple - especially as it only requires a single-pole switch to engage. However, to facilitate this, the overall supply is slightly more complicated because the supplies have to be complimentary. The main drawback is that you can't provide remote-sensing for the 0V terminal while maintaining tracking accuracy, but as I mentioned before, this isn't a problem with the low current output of the unit (2 amps).