Troubleshooting old Century autopilots

Quite a few older light aircraft have old analogue autopilots installed, such as the Century III, or the Piper Altimatic IIIc that is installed in my own aircraft. These can be troublesome, so here are some tips on how they can be mended, based on my own experience and talking to others. DISCLAIMERS: (1) There are several variants of these autopilots, so the information here may be plain wrong for the version you have. I accept no liability for any errors in this information. (2) If this information helps you diagnose your autopilot problem, get the necessary remedial work done and signed off by a licensed avionics engineer. (3) Although the electronics in these autopilots are relatively simple and built with 1970s technology, due to over-zealous regulation only a few outfits are actually approved to repair them. Don’t take anything you read here as a suggestion to do it yourself.

First, some general advice:

1. Make sure you know at least 2 ways of turning the autopilot off in a hurry. Hopefully, you will have a quick disconnect button on the yoke. You can also use the switches on the autopilot control console, or the autopilot circuit breaker (it should be pullable), or in an emergency, the master switch.

2. Be wary of using automatic pitch trim. The Century manual says it is important, but it can also be dangerous. If you suffer a reduction in engine power when you have pitch mode or altitude hold mode selected, the autopilot will trim back. The same can also happen if you get a failure in the autopilot system (or it could trim full forward). Then, when the autopilot disconnects, you are in a stall which you may have difficulty recovering from, because the aircraft is so far out of trim. See this AAIB report for a sad example (although the autopilot involved in that case was not made by Century). Some modern autopilots don’t have automatic trim, instead they have LEDs telling you which way they want you to trim.

3. Don’t expect these old autopilots to perform as well as a modern digital autopilot – they are much less sophisticated.

Now, some specific problems:

1. Older variants of the Century III range have a 1D395 amplifier that uses a “gripple board” PCB (not fibreglass, and the vias are made from small rivets, and from soldering the component leads on both side of the board). They also use carbon composition resistors.  They can be very unreliable because of bad joints. If you are tempted to resolder some joints, bear in mind that (a) only a few repair stations are authorised to mend these boards, and (b) the PCB may be varnished – the varnish has to be removed before any soldering (it’s not the modern self-fluxing sort), and replaced afterwards using a suitable PCB coating spray, to protect against condensation.

Newer variants (e.g. Century IIIc and Alimatic IIIc) use a 1C515-1 or 1C515-2 amplifier. This has a fibreglass PCB with proper vias, and carbon film resistors, and is much more reliable than the older type.

2. If the autopilot basically works but tends to over-control, producing small oscillations in pitch or roll, then one possible cause is that the flying controls are sticky or stiff. As a result, the autopilot has to apply too much force to move the controls; and then the controls move too far. Having the flying control systems lubricated may fix the problem. If it’s overcontrolling in pitch, and you have automatic pitch trim turned on, try it with automatic trim turned off, in case it’s over-trimming. If it’s oscillating in roll when set to heading mode but not when  set to roll or nav mode, it could be caused by backlash in the DG or HSI (i.e. the DG or HSI doesn’t respond to small changes in heading).

3. Some of the connections between the different components are made using small round blue connectors with 4, 5, 7 or 9 pins. These connectors can become unreliable after many years, because the gold-plated split receptacles lose their grip on the pins. In my case, this problem occurred in a connection between the glideslope coupler loom and the main loom. When no glideslope coupler is installed, the main loom connects directly to the AI; but if a glideslope coupler in present, the glideslope loom plugs into the AI instead, and the main loom AI connector plugs into a connector that the glideslope loom provides. This connection is constrained by the glideslope loom to be only a few inches away from the AI, so it has to be made behind the instrument panel, where it is subject to heat generated by the avionics.

The original connectors were made by Amphenol, who no longer manufactures them; however there are substitutes made by Cooper Interconnect. One supplier in the UK of these is Farnell (Newark in the USA). The part numbers all begin with 126-, for example 126-214-1000 is the 4-pin inline locking plug, and 126-215 is the 4-pin inline socket.

EDIT: as a temporary measure, you may be able to fix the problem of a loose connector like this. The female connector has two prongs that are supposed to grip the pin of the male connector. Insert a needle between the outside of one prong and the housing, and use it to bend the prong towards the centre so that it will grip the pin of the male connector better.

4. You can do a number of autopilot tests on the ground. The AI needs to be steady, so you’ll probably need to start the engine, assuming a vacuum driven AI. If you engage Roll mode, then you should be able to get the yoke to turn left and right by adjusting the roll knob, and there should be a position of the roll knob somewhere near the centre where the yoke doesn’t turn. Likewise, if you engage Pitch mode, you should be able to get the yoke to move forwards and backwards by adjusting the pitch wheel, and stop when the pitch wheel is near the centre position.

5. If there is no yoke movement at all in one plane, the fault probably lies in the amplifier. I’ve heard of blown output transistors (which are easily replaced), and a blown transformer. If there’s no movement in either plane, then either there is no power to the amplifier, or the oscillator in the amplifier probably isn’t working – this could be due to a fault in the amplifier or in the wiring to the DG (the coil in the DG is part of the oscillator, unless you have a Bendix-King HSI, in which case the autopilot adapter takes its place).

6. What those tests won’t tell you is whether the autopilot is receiving and processing the pitch and roll signals from the AI. To do this normally requires a flight test. However, it may be possible to remove the screws securing the AI so that the AI can be rotated and pitched a little (get your engineer to advise whether this is safe to do). If so, then with the roll knob and pitch wheel set to produce no yoke movement, you should be able to make the yoke move appropriately if you rotate or pitch the AI.

7. If the problem is lack of pitch feedback, then the cause could be a faulty amplifier or a faulty connection in the wiring (it’s unlikely to be the AI, the sensor is just a coil). You can check this with a multimeter. I’ve measured the resistance of the sensor in the AI at 120 ohms in one case, and 80 ohms in another. [EDIT: 80 ohms is correct. The 120 ohms measurement was due to a bad Amphenol connector in the circuit.] You should be able to measure this between pins 14 and 15 of the 24-way Centronics plug that connects the 1C493 glide slope coupler (if present), or between pins 8 and 12 of the amplifier connector. [Note: (1) the Century documentation suggests that some models with older amplifiers may have a pitch filter unit installed between the AI and the amplifier, in which case this won’t apply. (2) Pin numbers on the amplifier connector, looking into its open end, are A thru S from bottom to top on the left side, and 1 thru 15 from bottom to top on the right hand side.] The other possibility is lack of pitch excitation signal to the AI.

8. If the problem is with pitch mode or altitude hold mode, could the fault lie in the 1C493 glideslope coupler? There is a relay in the coupler that passes the pitch signal from the AI and the altitude sensors straight through when the glideslope has not been captured, so there’s not much to go wrong. However, from the circuit diagram, it looks that the designers made the elementary mistake of not putting a diode in parallel with the relay to catch the back EMF when the transistor controlling the relay switches off. Unless there’s some sort of protection not shown on the circuit diagram, that transistor is liable to blow. If it goes short-circuit then the relay will be permanently on – although in that case, the glideslope engaged lamp should light. If it goes open-circuit, the glideslope will never engage. If you want to eliminate the glideslope coupler from the equation, you can replace it with a 24-pin Centronics socket – you need to link pins 13&15, 19&21, and 20&22.

9. There are 6 helical potentiometers in the control console to adjust autopilot performance (left bank angle, roll centering, right bank angle, down limit, pitch centering, up limit). These can be accessed by removing the front panel. See the maintenance manual for the in-flight adjustment procedures, and be sure to take a safety pilot or lookout with you. The pitch centering adjustment is also the one used to get the autopilot to hold the altitude you had when you selected altitude hold, rather than an altitude a little higher or lower.

9. If you’re planning on keeping the aircraft for a while, and your analogue autopilot isn’t working, you may be better off having a modern digital autopilot fitted rather than paying money to fix the old one.

That’s all for today. If you are interested in the sort of faults that occur in Century II and III autopilot amplifiers, you may be interested in my later post. Please post a comment if you wish to share any experience of getting these old autopilots working.