The Smiths Industries Limited H6 Director Horizon is an avionics instrument used in a Boeing 737 air plane for indicating the roll and pitch of the air plane.
The H6 Director Horizon is the cockpit instrument for the indication of the horizon. The pitch and roll of the air plane can therefore be seen even is there's no visual on the actual horizon. An artificial horizon consists of a gyroscope (spinning wheel) and a graphical representation of the horizon linked to the gyroscope. 'Simple' artificial horizons are equipped with a (rather small) gyroscope inside the instrument. The gyroscope can be powered by vacuum or electrical power. A larger gyroscope is more accurate, but doesn't fit in a relative small cockpit instrument. Therefore a rather big (and more accurate) gyroscope is electrically connected to the cockpit instrument. The H6 Director Horizon is one of these cockpit instrument. The instrument is fed with electrical signals from the gyroscope that represents the horizon position. There's a closed loop control system so that the 'gyroscope horizon' matches the instrument indication. The H6 Director Horizon uses three synchros for the desires position indication on the instrument. In unpowered state the 'horizon is vertical'. The yellow 'cross hairs' are placed at the bottom right in unpowered state. The pitch indication can be in any 'random' position ans has no 'home position'. The default unpowered state of the roll is 90 degrees off. This is likely for safety so it's very clear that is instrument is not working. For the image below I placed the instrument in it's usual horizon position by manually rotating and fixating the horizon in the 'normal' position. When unpowered the 'horizon is vertical' displayed.
I got inspired by the video's of Michel Waleczek from Le labo de Michel. Michel has a great video channel about spectacular reverse engineering of avionics instruments. I got inspired by video #103 about this H6 instrument. The hardest part of reverse engineering is therefore already done by Michel as shown in the video. The video is Michel is shown below.
Michel reverse engineered also the gyroscope device! In video #132 the H6 indicator is connected to the gyroscope and is therefore functional! Unfortunately I don't have the gyroscope, so it's great to see the (semi) original setup of the device.
Why buy one is you can get two for only double the price? ;-) The instruments are sold 'as is' so it's likely that there are some defects. These instruments become usually available when they're not serviceable any more of are worn. Especially these H6 instruments are rather well priced so a spare device is very convenient toe have. I bought two devices at the beginning of the year 2021 in the UK, but due to the Brexit shipping abroad was not possible. Luckily a fellow ham radio operator in the UK was so kind to buy and ship the instruments for me! Thanks Dave! Due to the Brexit situation in the UK the package was sent back due to uncertain situations of customs. But after a while the customs situation was more clear so the package arrived as planed. Yey!
Both instruments were sealed with lock wire and also marked unserviceable... So hopefully the devices are still functional/usable in some way.
The good news is that the first device has a broken case. Well, not that the broken case itself is good, but this is a serious reason for discarding for original use ans I prefer this cosmetic failures instead of broken components inside. So far so good...
Afer opening the device some parts fell out. These parts are clearly pieces of bearings. That's less good news. The strange thing is that all the mechanics is functioning properly and there is no sign of a broken or missing bearing. Strange, but still so far so good.
I also found a verly loose bolt that should hold the wiring in place. The thread of the housing is broken and could result in loose parts in the gears. Since I don't want to fly with this instrument, I guess that the device is in good condition for me. But it's a good idea for discarding the device for it's original purpose.
There are six actuators in the instrument. Two (on/off) error flag solenoids, two yellow lines which acts as a cross hair, one pitch actuator and one roll actuator. The error flags can be switches on and of by 24 V, so that's quite simple. There's no feedback loop for checking the position. The other four actuators are controlled by a closed loop system. The input signal [CX] for each loop is a desired set point (which comes from the gyroscope). The output is a AC controlled motor which moves the physical parts. And there's a sensor [CT] for detecting the actual physical position. If the desired input set point is the same as the actual position, the motor don't have to move since the desired position as the actual position. If the set point is different than the actual position, the is an 'error' between de desired position and the actual position. This error signal is handled by the error amplifier printed circuit board which results in a AC voltage to the motor and therefore the motor moves until the error signal is gone and the physical position is the same as de the desired set point. This is a continuous process and results that the gyroscope position matches the visual on the cockpit instrument.
An AC control loop is quite well known and 'simple'. An AC motor can be controlled by a AC voltage. The desired set point can be set by a (AC fed) variable resistor which results in some AC control signal. And a second (AC fed) variable resistor can be used to measure the actual physical position which results in some AC voltage. Again, if the voltages are the same, the desires position is obtained. If the AC voltages don't match, the error voltage is used to control the AC motor to obtain the desires position. But is there's some resistance in the wiring or if there's some non-linearity in the resistors, that could result in a wrong reading. And most of all, resistors wear. But the main reason is that the rotation angle is limited to the maximum rotation angle of the resistor. And in theory an air plane should be able to make en infinite roll angle. Therefore is much smarter to use a more complex control system like which is used for the H6 instrument.
I'm definitely not an expert on control systems, but I'll try to explain the system based on the theoretical image below. (Therefore the number of gears not representative for the actual H6 indicator.) The control loops for the four actuators are for the most part the same. The exception is that the control system of the pitch and roll have each two extra position synchros. But you may forget this for now, this will be explained later on... On the image below, the load is the physical part which should be positioned like the pitch indicator, roll indicator of one of the two yellow cross hair lines. This images is mainly the same for all four loops, so I'll explain one loop. In this case I'll mention the roll loop...
The roll indicator can be moved my the drive motor at the right. The motor reference winding is powered with 115 V 400 Hz. Even powered, the motor doesn't move. The motor shaft will start turning when a AC voltage is applied. The direction of rotation is determined of the AC control voltage is positive or negative polarised. The motor speed is related to the actual voltage. 0 VAC will result in a motor that doesn't move. -20 VAC control voltage will result in a maximum backwards speed and +20 VAC will result in a maximum forward voltage.
The desired position of the load is controlled by input synchro [CX]. The synchro is fed with a reference voltage of 115 VAC 400 Hz. The output signal of the synchro is a sine wave and cosine wave in some amplitude. The output signal of the synchro corresponds with the actual angle between 0 and 360 degrees. In real life the reference voltage to the H6 instrument and the gyroscope unit is the same. The input synchro is connected to a gyroscope axis. So the three wires from the gyroscope input synchro is the input signal of the H6 indicator.
(Here's an intermediate step in the explanation... A synchro is a device which can rotate to an infinite angle. Except for the axis there are no moving parts like moving electrical contacts. This makes synchro very durable! The synchro is fed with some reference signal like a 115 VAC 400 Hz signal of a common 26 VAC 400 Hz signal. In the synchro are coils that pick up the reference signal. Based on the position of the axel, the output signal changes in amplitude and phase. This can be done by two or three coils for example. When two resynchro solvers are connected to the same reference signal and the three coil wires are connected, the two synchro are linked. When one synchro is rotated, the other synchro will follow to the same angle. If one synchro is held in de same position, the other synchro is very hard to move and will be forced to the position where both synchros have the same angle position.)
The synchro [CT] angle is mechanically connected to the actual physical position of the load. The three coil wires are connected to the synchro which indicates the desired position. Therefore the reference winding of synchro [CT] is used to determine the error. If the angles of the two synchros [CX] and [CT] are the same, the output of the reference winding [CT] is 0 V. Therefore the error voltage is 0 V and the motor does not have to move. If the angels doesn't match, there is an error voltage and the AC controlled motor is fed with an AC voltage to 'send' the motor to the desired position. The result is that the load angle corresponds exactly with the angle of the input synchro. Or, that the instrument indicator is in de same position as the gyroscope position...
Michel Waleczek from France reverse engineered the error amplifier boards. There are four identical amplifier boards which control the AC controlled motors. The schematic is shown below of one board. This is a screenshot of video #103 of Michel. The input signal  at the left is the input from the actual position synchro. The output  is the -20...+20 VAC signal to the AC controlled motor to move the mechanical parts. Based on the 3k9 and 270k resistors it's clear that the gain of this amplifier is approximately 69. 270.000 / 3.900 = 69,23. The two diodes at the input are likely limiting diodes to prevent over voltage of the amplifier.
I found out that the roll and pitch control loops have three position synchros instead of one! I haven't reverse engineered the wiring (yet), but I made an educated guess of the purpose of these 'extra' synchros. I expect the synchros are connected to the output connector of the instrument. I expect that the synchro outputs are connected to the gyroscope unit for angle comparing with the actual position of the gyroscope. In a steady state, the 'extra' synchro should have the same angle as the gyroscope synchro angle. If these synchro angles are the same, everything is as it should. If a component fails like a broken amplifier board, broken motor or something, the visual representation of the instrument is different than the gyroscope position. Therefore the control loop is open en tries (without success) permanent to obtain the desired position. The effect is that the position of the'extra' synchro doesn't match the position of the gyroscope. I assume that the 'unknown error flag' at the left bottom appears to indicate that the shown information on the instrument doesn't match with the gyroscope position. This makes a close loop system to check if the visual information matches the actual gyroscope position. If the rotation of the gyroscope is faster than the instrument motors are, a temporarily error could be shown. But if the gyroscope spins faster than the instrument can show, the pilot has other problems to worry about than a wrong pitch and roll indication. ;-)
The front panel is equipped with some lightbulbs. There are at least two electrically parallel lightbulbs placed. After some simple 'reverse engineering', the connections are retraced to the rear panel connector. Pins [k] and [X] are used for the ligh bulbs power supply. Based on some tests is fairy fate to say that the bulbs are designed for 5 Volt operation. By applying 5 Volts, there is a current draw of 840 mA in total. That results in 2 Watts per bulb (if two bulbs are placed). In an air plane there's plenty of electrical energie available, so the power consumption is less relevant. It's likely that led lighting would be the 'modern' choice. Well, nowadays these mechanical instruments are replaced by electronic devices of course... There's a picture below showing the panel when the lighting is turned on.
There are two orange flags on the frontpanel. The original purpose is unknown but it's likely that both 'flags' are error flags for indicating a not desired state. My guess is that the [+] flag at the left is shown if there's no electrical power supplied to the instrument. My guess is that the 'horizon' flag at the left corner is used to indicate the gyroscope state. I think that the 'horizon flag' is used to indicate that the gyroscope and instrument are up and running. The gyroscope takes some time to spin up and therefore it's likely that the instrument indicates that the gyroscope is at full speed and therefore ready to use. Both flags are controlled by some external equipment. After some simple 'reverse engineering' it turned out that if 24 V (28 mA) is applied to wires [g] and [h], the [+] flag disappears. The electrical energy powers a solenoid which retracts the error flag from the front panel sight. The wiring of the 'horizon flag' is not (yet) known.
My goal is to make a demo setup. The instrument is nice, but it's even better to have the instrument functional. The original gyroscope is rare en probably rather expensive, so that's not for me. I planned to make a panel with the instrument with actuators to simulate the function.
One switch for power on/off. One switch for the dial lighting. One switch for the [+] error flag at the right bottom. One switch for the 'horizon' error flag at the left bottom.
One knob for the desires pitch setting. One knob for the desires roll setting. One knob for the desires horizontal yellow marker setting. One knob for the desires vertical yellow marker setting.
After studying the possibilities and sharing information with Michel, the best way to simulate the positions is by using actual synchros. Synchros are not that cheap and I need four of them... Then I came up with the idea to use the 'extra' synchros from the device. Since I don't have the gyroscope unit not want to fly with the instrument, I don't need the safety feedback loop. So my plan is to use the four surplus synchros from the device to use as an external gyroscope simulation. Well, the idea's are here, now it's time for further studying and building the demo setup. To be continued (someday)...
The device uses 115 VAC 400 Hz. The +115 VAC 400 Hz sinewave signal has to be connected to pin 'A' of connector DH1. The power return is connected to pin 'W'. The instrument used AC voltage since the frequency is used for synchro synchronisation. Therefore DC voltages can not be used. The frequency is 400 Hz instead of the 'usual' 50 or 60 Hz since a higher frequency results in much smaller transformers. There are two small transformers built in for generating -20 VDC and +20 VDC for example.
115 VAC 400 Hz is an 'odd' power supply signal and a standard avionics power supply is rare and expensive. There's a 'trick' to create the desired power supply signal. For the lab setup I use a (Rigol DG1022Z) signal generator. The generator is set to generate a 400 Hz sine wave of 1 Vpp. This signal is fed into an audio amplifier. I've got an mono '100 V' amplifier that that has a 4 and 8 Ohm output and also a '100 V' output. The 100 V output is used to drive a lot of loudspeakers for broadcasting announcements in an office building for example. When the amplifier is fed with the 400 Hz sinewave, the output of the audio amplifier at the 100 V output is approximately 130 Vrms. By reducing the 'audio output', the voltage can be reduced to 115 Vrms.
If no 100 V line amplifier is not available, a 'normal' audio amplifier can be used by connecting a power transformer 'backwards' to the 4 or 8 Ohms output. A 240:24 V transformer reduces the voltage by ten, by 'reversing' the transformer, the voltage is increased by ten. When the 24 V side of the transformer is wired to the 4 Ohms output of the amplifier, the output signal of the amplifier is multiplied by ten. The desired higher voltage is available at the '240 V' connection of the transformer. There will be losses since a 50 Hz transformer is not designed for 400 Hz operation, but I guess that's no problem for engineering/testing...
There are two connectors at the rear end of the device. DH1 is the primary connector and DH2 is a secondary connector. DH2 does not have to be connected for 'normal use'. After some reverse engineering I found that the connector is for two independent synchro transmitters (pitch and roll) and the 'horizon error flag'. More information is shown below. DH2 connector
Image of connector DH2.
DH2 connector wiring
Reverse engineering notes.
Explanation The connector has two synchro transmitters, the 'horizon error flag' and three ground pins connected. Pins 'A' and 'L' are not connected. Pins 'M' and 'N' are connected to the 'horizon error flag'. When applying 28 VDC, the error flag with a horizon marking hides. Pins 'V', 'S' and 'F' are wired together and are wired to the ground pin of connector DH1. The pitch resolver transmitter [I] has six wires. Pins 'K' and 'T' are the 26 VAC 400 Hz input for the reference winding. (Pin 'U' turned out to be a intermediate reference winding tap.) Pins 'J', 'G' and 'H' are the output wires of the position windings. The roll resolver transmitter [H] has six wires. Pins 'P' and 'E' are the 26 VAC 400 Hz input for the reference winding. (Pin 'R' turned out to be a intermediate reference winding tap.) Pins 'D', 'C' and 'B' are the output wires of the position windings. Suspected application Connector DH2 has the wiring to two resolvers and one error flag. I thought about this and I have a suspect what this is used for originally. Since original documentation is nowhere to find, this educated guess isn't validated (yet). The gyroscope pitch and roll position are send to the indicator. The indicator positions are compared to the gyroscop positions and the 'error' is fed to the motor so the indicator synchronises with the gyroscope position. This is a controlled loop that works great. But if one link fails, it's not detected. If there's a broken wire, a broken amplifier board, a broken cog, a broken synchro ore something else, the indicator doesn't match the gyroscop position! The result is that the indicator doesn't show the real (gyroscope) position. This could be very dangerous since the pilot doen't know the plane's position in zero visibility... My educated guess is that there's an extra control loop. The two synchro's send the real position of the indicator (not the position that it should be). I think that this position infomration is send back to the gyroscope for comparison. If the gyroscope synchro position matches the real position of the indicator, the indicator shows the desired information. If something fails and the shown position doesn't match with the gyroscope position, this could be detected since the synchro signals don't match. I expect that the 'horizon error flag' is powered by the gyroscope unit as long as the gyroscope and visual indication match. If the error flag is hidden, the indication is right. If the gyroscope position doesn't match the visual indication of the position (due to some fault), the error flag is shown to inform the pilot that the indication is wrong. I assume that there's also a little delay or accepted angle error built in. If the plane pitch/roll changes faster than the indication chanes and therefore the visual representation is lagging, the pilot could be starteled by the appearing flag. To prevent primature warning, the lagging time could be compensated with some (small) delay. With a small delay or accepted angle error, the error flag will only appear if the position indication is seriously wrong due to failure instead of a 'slow' system.