My son found this Exact 301 Function Generator somewhere, and while it's in marvelous shape, it didn't work. First, though, some more pictures:

It turns out that there is next to no documentation for this unit on the internet. Some poking about returns a product line brochure that shows the unit and it's specs, and that's about it.

Ours is the rack mount version; there's also a much more compact version without the rack panel.

With no schematics or any other troubleshooting information, I thought I had five choices of what to do with this unit, in descending order of difficulty:

• Fix it. (On reflection, after I wrote this, I realized that this was easier than the next two choices, and probably easier than all but the last choice.)
• Replace the electronics with op-amps and try to retain the front panel controls.
• Replace the electronics with an analog function generator IC, like the Intersil 8038, and try to retain the front panel controls. The ICL8038, despite being obsolete and out of production, is readily available on eBay.
• Replace the electronics with an Arduino and a modern function generator IC.
• Throw it out and give my son one of my working function generators.

Out of curiosity, I suppose, I started drawing the circuitry on the "main" board, (the "Square Triangle P.B."). This unit is all discrete transistors. It's a late enough design to have silicon transistors, but too early for integrated circuits. Also, it seemed a shame to hack this unit, as it seemed like a very nice unit. It didn't hurt that (Google revealed) Exact Electronics was started by ex-Tektronix employee(s).

Architecture

The generator, perhaps unsurprisingly, uses the canonical approach to making a function generator: an integrator feeds a comparator, whose output is the input to the integrator. Front panel selection of the integrator's capacitor allows range selection, and front panel variable resistor is a vernier on the integrator input current.

There are four boards in the unit, not counting the power supply, which is a straight forward linear supply outputting plus and minus 15 volts. The four boards are:

• The aforementioned Square Triangle board.
• The Gate Lockout board, used to gate/trigger the oscillator for non-free-running applications.
• The Sine Shaper board. This is a standard diode shaper.
• The Main Amplifier board. This allows one of the three waveforms to be output to the front panel BNC with amplitude control. All three waveforms are also available at constant amplitude on three banana jacks.

(Partial) Schematics

I manually traced the schematics for the two most important boards (where "important" means "can prevent it from oscillating").

Studying the schematics revealed some truths:

• The oscillator schematic is the standard "integrator feeds comparator feeds integrator" circuit.
• The Gate Lockout board isn't the problem as long as the front panel trigger select is on "INT" (internal), because then, the "TRIANGLE IN LOCKOUT" and "TRIANGLE OUT LOCKOUT" signals fed to the Square Triangle board are both at -15, so the diodes in series with these inputs are biased off, so those signals are effectively disconnected.
• Comparing the schematic of this unit to the (available on-line) schematic of the tube (valve) version of the same product reveals that the Exact engineers used the same circuit design for the oscillator, but translated to solid-state.

Troubleshooting

I figured I'd need a card extender if I was to do any serious debugging. Luckily, I'm a pack-rat and still have in my junk box some breadboards with matching connectors, which connector is the SAME as the one used in the Exact 301. 18 wires and some cyano-acrylate later, I had an extender card:

I put the Square Triangle board in the extender in the card cage, and started probing voltages. Nothing jumped out at me; all the semiconductors had likely voltages (for example, 0.7v drop from base to emitter) on their leads. I'd previously tested all the transistors using my trusty Chinese/eBay component tester, and they all looked reasonable.

Whilst probing and trying to understand the front panel frequency controls, I noticed that the output of the square circuit would toggle from high to low and vice-versa if I touched the circuitry, or if I power cycled the unit when the controls were set in different ways. This suggested to me that the unit was "almost" oscillating; perhaps some parts had drifted out of spec?

Most of the resistors in this unit are 1/2 Watt carbon composition, some 10%, some 5%. There are a few metal film 1% resistors, but not as many as one would hope. Since carbon composition resistors are known for drifting over time (sometimes radically), I decided to ohm-meter all the carbon comps on the square triangle board.

This process revealed about a four or so resistors that had drifted significantly (like 20%). They almost always drift high, in my experience. I replaced those four resistors, and surprisingly, the unit now oscillates. The waveforms are lop-sided, and it'll need a re-calibration, but it's working.

Why did drifting resistors make the unit stop oscillating? Because the resistor values control the threshold at which the square wave comparator switches, and when the values drifted far enough away from their design values, the comparator voltage was no longer reached by the triangle wave output.

If all the resistors had drifted by the same percentage, the unit would still have worked (for drift value within "reason", like 20%).

Another piece of equipment saved from the landfill.

What exists on the internet

See pages 7 and 8 of this PDF. Note the price of $550!

Calibration

Lacking any proper documentation for the Exact 301, here is a link for the manual for the Exact 240 Function Generator. This unit is tube-based, but the circuit architecture is the same, so *some* of the calibration methods are close enough that they work for the Exact 301 as well. In particular, the calibration of square wave symmetry and sine wave distortion are appropriate. Obviously, calibration steps that suggest measuring voltage at a tube pin can be ignored.

FAQ

Q. Will you fix mine?

A. No, I'm retired and so not interested in a job.

Disclaimer/Warning

This is just my documentation for my repair. I don't claim that doing this is safe or recommended.

Soldering irons are dangerous, be careful. Oh, and don't eat the solder.

There are live mains voltages in this unit. Be careful when poking around in the power supply area.

William Dudley
June 14, 2022