S-TEC autopilot servo -- quality product?

[achimha]

Some months ago I got a brand new S-TEC autopilot, system 60-2 with all bells and whistles (GPSS, autotrim, altitude preselect, etc.). Don't ask me what I paid for it and how many alternatives I had...

The other day after flying to Paris, leaving the airplane there for 2 days (up to 33°C) I wanted to fly back and noticed that the autopilot would no longer operate the pitch axis. It would not notice the error (diagnostics pass) and when engaged happily block my trim wheel. Great!

So next trip is to the shop that installed it. The "computer" (what a euphemism for this 1970s analog circuit rubbish) was diagnosed to be fine so the servo was removed which is quite a job involving a slim employee climbing all the way inside the tailcone with hardly any air to breathe.

The servo was opened and what a surprise! It has a little motor which has a cog mounted onto its shaft. The cog is held in place with a little shear pin pressed through a hole in the cog and shaft. That pin has fallen out with the shaft rotating freely.

When the aircraft was parked in Paris at 33°C, the metal expanded and by intelligent choice of metals, the shaft expanded more than the pin which just fell out (the shaft was probably by coincidence positioned with the pin vertical).

S-TEC in their effort to even maximize their obscene margins no longer consider the servos to be field serviceable and no longer make spare parts available to its dealers. Every issue means sending the servo to Texas and (in case out of warranty) pay an obscene fixed price repair fee. Luckily my dealer still had some spare parts in stock including the shear pin and the repair was done in 10 minutes.

I find it totally amazing what crap quality companies like S-TEC deliver. The servos are very expensive and everything is certified, the company, the production, the product, its application in a given plane etc. They build the stuff exactly like they have been doing for 30 years and it's full of this "proudly made in USA" bull**** while it's just very very poor quality. Such a simple product and after 30 years of producing it, it's hard to believe.

Sorry, had to get this off my chest. I have no problem with products failing but I have a problem paying obscene amounts of money for what is clearly the result of poor build quality and sloppy quality assurance.

[peterh337]

I couldn't agree more.

The quality of GA avionics varies hugely, and autopilots tend to be at the crappy end.

You can read about the KFC225 here (http://www.peter2000.co.uk/aviation/kfc225/index.html) :) And that cost about 2x what you paid for yours.

Interestingly, when Avidyne (a company which BTW is all words and very little action nowadays, with desperate new product promises and cash-up-front deals for discounts when Box X is certified etc all suggesting they are very short of cash) started doing their DFC90 autopilot, they were going to use STEC servos, to ease the certification process. But Avidyne have recently announced (http://www.peter2000.co.uk/aviation/kfc225/avid-servos.pdf)that they are switching to the King (Honeywell) servos.

In the meantime Honeywell show no sign of wanting to fix their defective servo design whereby the amplifier self destructs if the current limit is activated for more than a few seconds. Their new SB11 just shorts the current sensing resistor with a piece of wire :ugh:

The Honeywell servos are mechanically very good and much better than STEC's but their electronics are rubbish.

The bottom line is that most GA avionics firms have very little in the way of brains in there. I suppose the pace of new product introductions has been so slow that all the smart people moved on years ago.

[Pace]

I flew two Seneca Five twins an earlier King autopilot equipt aircraft and a later S tech autopilot equipt aircraft.
We had no end of problems with the S tech.
Lock onto an ILS and while the King would fly the needles to almost the ground the S tech would start wallowing all over the place as you got closer in get worse and loose the plot!
An engineer described the S TECH as a load of cheap rubb+sh.

Pace

[achimha]

Unfortunately S-TEC are the only ones that offer retrofit autopilots. The best hope for the future is the Avidyne DFC90 which pilots seem to like very much but as Peter said, they're not making much progress. So far it's only Cirrus and I believe now Malibus. Cessna 182 are "announced". Not even the new Extra 500 is supported which ships with an Avidyne Entegra 9 glas cockpit but a horrible S-TEC 55x.

My dealer showed me the internals of the servo and how they can easily fail in a way you get trim run away which is a great way to get killed!

[peterh337]

Avidyne told me a few times they are really only interested in retrofitting the pre-G1000 Cirruses (a few thousand of those, perhaps) with the DFC90.

But also their story is the same every year, and to me (as a businessman) their general behaviour is that of a financially struggling company.

Like it or not, Honeywell handed the GA market to Garmin on a plate, and Garmin have now eaten everybody's lunch :)

But Garmin are not going to be certifying their -700 autopilot (which appears to be very good, though I would regard it as very unproven long-term) for too many types. They want to sell it with the G1000, mainly, and similar products.

[2high2fastagain]

I can sympathise with your frustration. My S-TEC went do-lally and scared the crap out of me. The yoke suddenly sent the ailerons into full deflection and the aircraft rolled inverted. I was reciting my VFR essay to ATC when it happened - I bet they periodically re-run their tapes to enjoy the plethora of non-CAP413 utterances I was screaming as I was showered with pencils, dried mud a checklist and an iPad.

After a thankfully relatively inexpensive repair I was told that the connection to the turn coordinator had apparently fallen off (or something like that I think). The autopilot would have cheerfully barrel rolled me all the way home at constant altitude (worth watching from the ground I'm sure).

I am now much more phlegmatic about the autopilot and my hand is never far away from the disengage switch. Interestingly, despite some initial training from my CFO and regular use, I can confess that I did wrestle unsuccessfully with the yoke for a full second before I had the wherewithal to push the disengage button.

[Pace]

2High

I also had the same problem with the S tech on the Seneca with uncommanded and abrupt rolls.
Do not know whether you ever punched in the APP mode on approach?
It was fine further out following the localiser and glide but at around 1/2 mile started hunting left and right on the localiser which got so bad you had to disconnect and hand fly.
As stated the King in the 1998 machine was a dream in comparison. The engineers never managed to sort the S tech.

Pace

[Katamarino]

I bet they periodically re-run their tapes to enjoy the plethora of non-CAP413 utterances I was screaming as I was showered with pencils, dried mud a checklist and an iPad

You really have to get a recording and share that here :D

[dublinpilot]

I wonder why?

Obviously they know about the problem, they know what is causing it, they know it's affecting their reputation and presumably know how to fix it.

Is the certification process a big barrier to fixing the problem? I mean if they want to fix it, they presumably have to go through certification again. Is it really that big of a problem to do?

[achimha]

The instructions of my 2012 made autopilot are written with a typewriter. That was probably part of the FAA certification of their production facility back in the 1970s.

[2high2fastagain]

Pace,

Nope, haven't dared to give it a go I'm sorry to say. I've flirted with the VS function (vertical speed) and did a fairly decent let down through cloud over the sea, but close to an airport I prefer to do my approaches with the old six-pack and the Mk1 eyeball (and sadly yes, the only six-pack I'll ever have will be mostly powered by a vacuum pump).

kat - there's no way I'm going to ask! The fine people who provide us with our radar service out my way only know me by call-sign and I'd like to keep it that way - at least as long as this one is concerned.

[peterh337]

Is the certification process a big barrier to fixing the problem? I mean if they want to fix it, they presumably have to go through certification again. Is it really that big of a problem to do?Normally a big firm like this has the authority to do mods without reference to the FAA.

I think there are perhaps several reasons they don't fix stuff:

1. All people who knew about it have left the company (true for most Honeywell products)

2. The product was bought in, or developed by a contractor who is no longer around to fix it (true for e.g. the KSN770)

3. The lawyers have banned any acknowledgement of the issue, so also banned fixing it - unless the fix can be done under the guise of some harmless sounding SB. This is a huge problem in GA as a whole. I know that e.g. Honeywell have gone to great lengths to tell individual customers that they are the only people with a problem, or that it happens only on the one aircraft model.

4. The company is no longer interested in anything beyond running some old cash cows.

STEC autopilots were never much good for performance because they use the TC as the roll reference, and use indirect means as the pitch reference.

Also the number of STCs which STEC got is so huge that it is totally obvious that they could not have possibly tested the system on every aircraft and done so at all the limits of the loading envelope and done each of these at both ends of the airspeed range. That's why we get so many reports of STECs oscillating. The KFC225 in comparison almost never had any problems and delivers really accurate control; its weakness is the servos.

[achimha]

STEC autopilots were never much good for performance because they use the TC as the roll reference, and use indirect means as the pitch reference.

You're a bit unfair here Peter. In fact, rate of turn based autopilots do work quite well for smaller aircraft. In most situations, there will be no real difference. In slow flight, they tend to oscillate because they have no means of determining bank, only the resulting roll condition. Above 60 KIAS, they usually do a very good job. The pitch axis is controlled using ambient air pressure (and its delta over time for climb/descent) as well as accelerometers to detect turbulence. I agree that attitude based autopilots are superior but they are also a lot more complex and -- the biggest drawback -- they have a strong dependency on a certain AI/HSI. HSIs are very expensive to repair and you can't easily replace them with something else when you have a KFC225.

The big difference between S-TEC and KFC225 is that the former is a retrofit autopilot and the latter only comes (or rather came) with factory new aircraft. S-TECs are field installed (with varying quality) and you find them in a lot of aircraft. If installed well, they do their job but often they're not installed well. In my airplane, I cannot complain about its accuracy and comfort level -- when it works! The Cirrus aircraft were equipped with S-TEC until some years ago and they work well. In your case, the aircraft manufacturer installed the KFC225 and flight tested it. That is of course a much better package than a retrofit autopilot that was mostly paper tested.

[S-Works]

I have an S-Tec in my Cessna with Alt, GPSS etc and it has been faultless. I like the fact that it is separate from from the AI (which has failed more than once).

I also don't like the AI linked units after having a nasty experience mid channel in a Senecca in IMC when the AI failed and the auto pilot followed it to turn us upside down.

Its like everything in aviation, over priced and out of date technology!

[A and C]

Teach your wife/girlfriend to keep the aircraft S&L in the cruise and fly the interesting bits yourself, it's got to be cheaper than fixing this 60's technology and there is probably less chance of her rolling the aircraft on its back !

[peterh337]

Done that already :ok:

BTW, any autopilot is vulnerable to a failure of the pitch or roll source - whatever it is. A TC can fail, just like an AI can fail. The KI256 just happens to be an expensive bit of kit because it contains a flight director (and because Honeywell are milking it).

There is an exact electrical replacement for a KI256, but electrically driven so not vulnerable to a vac pump failure, from Castleberry in the USA, which is TSOd, but you would need to apply for a field approval (N-reg). And you would need a separate power source for it, obviously (a second alternator, or a sizeable backup battery).

[achimha]

Next S-TEC disaster!

After the low quality servo was repaired, 40 minutes into the flight in calm air, the electric trim stopped working. The AP detected the condition and provided manual trim commands so better than the last fault.

Turned out the fuse on the trim board inside the pitch "computer" melted. I replaced the fuse and tried again. Looks like the board runs the trim servo in one direction without indication and as soon as it gets an opposite direction trim command, it shortcuts the circuit and the fuse blows. Looking at the circuit, there is no protection against driving the servo with both polarities at the same time, i.e. shortcut the circuit. Amazing design.

[wigglyamp]

Of course you can also get rid of your KI256 by fitting a Garmin G500 or Aspen EFD1000 EFIS with the appropriate autopilot interface unit and then for the mandatory back-up you can have a standard vacuum horizon from RC Allen, SigmaTek etc.

Incidentally, I have an S-TEC 30 in a PA28 and it's been pretty flawless for 7 years. We do install STec systems at work and know of the shortcomings but as with all things at the cheaper end of the Market, you get what you pay for. Once an installation is completed properly they tend to be pretty good. STec rate-based systems tend to have poor performance in higher speed aircraft but for the top end of their Market, STec make the 2100 Intellipilot which is attitude based and this does seems to work extremely well. Unfortunately the 2100 is only STCd on a limited number of types - typically twin turboprops such as Cessna 425, Twin Commander and KingAir.

[UL730]

Teach your wife/girlfriend to keep the aircraft S&L in the cruise and fly the interesting bits yourself, it's got to be cheaper than fixing this 60's technology and there is probably less chance of her rolling the aircraft on its back !

My wife (PPL/IMC) and I fly together and we have rewritten the checklist - to cover several items such as familiarisation with the AP CB and all the ways to disconnect in the event of a trim runaway. I fly with my left thumb permanently covering the AP disconnect (BD Altimatic V) although in 2000+ hours it has worked without incident.

This made chilling reading - H&W team flying a 182 fitted with a KAP 140 and an inadvertent autopilot engagement in altitude hold mode before takeoff.

http://www.aaib.gov.uk/cms_resources.cfm?file=/dft_avsafety_pdf_501522.pdf

I have no personal experience of an S TEC AP.

[achimha]

http://www.aaib.gov.uk/cms_resources...pdf_501522.pdf

Interesting read. So this accident made Bendix introduce the audio warning TRIM IN MOTION. I rented a 2005 C172 some weeks ago and I was annoyed by how long you have to push the buttons for the AP to take your command and the TRIM IN MOTION annunciation over the audio panel. Now I know why and it's actually quite a good solution.

Trim runaway errors are evil. My avionics guy told me about a situation they had in a Fairchild Metroliner (http://en.wikipedia.org/wiki/Fairchild_Swearingen_Metroliner) with trim runaway. This aircraft does not have a manual trim wheel and two pilots together were not strong in enough to pull back the elevator. Almost got killed.

[172driver]

Interesting - and chilling - read indeed.

Slight thread drift: how many of us actually LOOK BACK at he trim tab during the checks or confirm during walk-around that the 'take-off' position on the wheel actually corresponds to the correct trim tab setting?

Perhaps not an issue on you own a/c which you know well, but a good idea on a rental!

[achimha]

I would like to try out a takeoff run with the trim tab fully deflected to either position and see how that feels. I wonder if that is a dangerous experiment.

[172driver]

I would try a simulated t/o at altitude. However, I believe what gets you in this kind of situation is not the actual flying, but the WTF :eek: factor. Pushing full down with all your force is totally counter-intuitive at t/o and I guess for most pilots the time to figure out what has gone wrong would be too long.

After once discovering the trim tab position way out of agreement with the wheel, I've made it a habit of looking back at the tab immediately before lining up. Not sure that would have saved the day in the accident described, as it is conceivable that the trim ran away during the t/o run. In this case, all bets are off.

[peterh337]

That 2x fatal is well known.

as it is conceivable that the trim ran away during the t/o runIt has been speculated that since their AP was mounted at the bottom of the centre stack, they may have engaged it (pressed the AP button) with a knuckle, when advancing the throttle for takeoff.

However, with a correctly functioning autopilot, the system doesn't ever drive the pitch trim servo except as a response to a signal from the torque sensor in the pitch servo. The KS270C pitch servo has a -3V to +3V output, corresponding to max torque one way / the other way, and the AP computer uses this to drive the pitch trim servo to reduce that signal back to within certain limits.

So, to get the pitch trim servo to run at all, during the takeoff phase, you would need to engage the autopilot (by pressing AP) and then ignore the fact that it is driving the yoke back towards you (or away from you; the direction depends on the actual pitch angle at the time AP was pressed, and on the subsequent pitch angle, because merely pressing AP captures and tries to maintain the pitch angle, and it does it solely by driving the pitch servo) and you would either have to resist that motion (which would be bl00dy obvious, since during the takeoff phase the elevator hardly moves, apart from its initial "self-alignment" when you go to full power) for several seconds, or ignore the motion until it hits one of the stops, and only then will the pitch trim servo get actuated.

The above assumes the system was working as it should be.

I have no reason to say this relative to that accident but loads of people fly with defective or partly defective autopilots because (a) the costs of fixing it and (b) the extreme scarcity of avionics shops in Europe who know anything about them. Most planes I flew in during training, that had autopilots fitted, had them placarded INOP or something similar, but that doesn't mean the CBs were pulled ;)

how many of us actually LOOK BACK at he trim tab during the checks or confirm during walk-around that the 'take-off' position on the wheel actually corresponds to the correct trim tab setting? It is a pre-takeoff checklist item on the TB20. It's obviously critical to have the elevator trim in the right place (more or less) prior to takeoff, on any aircraft type.

The TRIM IN MOTION is IIRC what you get after 9 seconds of continuous trim servo activity, and after 15 seconds (IIRC) the AP will disconnect and hand the plane back to you, sir :E

The system I fly behind that had annunciation years before that accident, but there may be autopilots which didn't have it.

[172driver]

However, with a correctly functioning autopilot, the system doesn't ever drive the pitch trim servo except as a response to a signal from the torque sensor in the pitch servo.

Peter, suggest you read or re-read the AAIB report. According to it exactly that can happen.

[peterh337]

I can see that, yes, but mine (a similar system except that the altitude preselect is integrated) doesn't do that, and I cannot see why this type of autopilot (with a pitch servo) should drive the pitch trim when on the ground.

On mine, the only way to make the pitch trim servo run is to either subject the pitch servo to a "torque" (in flight, or by pushing/pulling on the yoke either on the ground or in flight for several seconds), or of course by operating the electric trim switch.

I will make some enquiries in the USA regarding this, to see if Honeywell changed the firmware since 2001, or whether the KAP140 really does this.

[AN2 Driver]

Interesting read.

I've been toying with the idea of upgrading my Mooney with either a S-TEC 30 as a minimum solution or a 60-2 combined with an Aspen, the latter as I don't see the point of having a FD compatible PFD without an AP which delivers the FD signal.

Any experiencs with the -30 and 60-2 will be appreciated.

Friend of mine used to fly a Twin Com with a -30 over most parts of the world for thousands of hours and told me he never had as much as a twitch, much less a real problem. Of course, the -30 is a rather simple system with only an alt hold function.

I did my initial IR on a KFC150 equipped Seneca II and later also flew on a Seneca I with the old Piper Alticontrol system. Never had any problems with either. Never flew with an S-TEC system at all.

[peterh337]

I wonder if wigglyamp might comment on that AAIB report?

He knows about autopilots.

I have been asking around US forums and they seem to be 50/50 on the AAIB report being confused.

Personally, and not knowing the KAP140 or the particular C182 installation, I cannot be sure that the accident installation was not faulty.

There is a curious way in which a fully functioning autopilot might drive the trim during the takeoff run: when full power is applied for takeoff, the airflow "straightens" the elevator into a neutral-ish position. If you engaged the autopilot, by pressing say the AP button just prior to that moment, it would "grab" the yoke instantly, and the centralising-airflow effect on the elevator would apply a torque on the pitch servo which would cause the pitch trim to be driven. However, this sequence of events would commence at the moment full throttle is applied, and I wonder if there would be anywhere near enough time for the Cessna electric trim to wind the trim all the way UP in the duration of the takeoff run.

[172driver]

Peter, according to the AAIB it's the other way round, i.e. the LACK of airflow initiates this sequence of events. Quoted from the AAIB report, my bold:

If any one of the autopilot AP, HDG or ALT buttons had been pressed before the take-off roll commenced, the autopilot would have engaged and attempted to maintain the vertical speed existing at the time of engagement. In effect, since the vertical speed would have been zero, the autopilot signals would have been consistent with attempts to maintain level flight. In the absence of aerodynamic loads, the elevator would have drooped under its own weight, causing a difference in tension in the elevator servo capstan cables. This difference would have caused the pitch trim servo to be signalled in the normal manner, thus causing movement of the trim tab. With insufficient airflow to produce any aerodynamic force on the tab that would lift the elevator and eliminate the tension differential, the auto-trim function would have continued to apply nose-up trim until the trim actuator cables reached their stops.

[Ariel_Arielly]

Dear freinds,
A freind of mine has the S-Tec auto pilot in his E-AB Lancair 360 A/C.
My freind asked me to help him identify the problem in the auto pilot:
There is an external toggle switch 3 positions: ON - OFF - TEST.
In test position it operate the auto pilot in test mode but in ON position nothing works. no lights no eny sign of operation.
Has any one of you met this problem? of have any idea what can be wrong?
Thanks,
Ariel