A strange slaved compass problem

P

peter-h

Line Up and Wait
Joined
Sep 24, 2007
Messages
613
Location
UK
Display Name
Display name:
peter-h
I have put a little movie here
http://www.zen74158.zen.co.uk/temp-files-ph/ehsi-valid-error.wmv

The compass system comprises of a KMT112, KG102, and the Sandel SN3500 EHSI.

It has been working fine since February 2011.

I have recently noticed the weird behaviour whereby the EHSI compass scale changes from white to red (indicating a "heading invalid" signal from the KG102) when one turns right through 090 or turns left through 270.

It happens for turn rates between about rate 0.5 and rate 1.5. Higher turn rates do not show it, and lower turn rates do not show it.

The KG102 has just been replaced with a freshly overhauled unit from the USA, so it can't be that.

As I see it, it could be the KMT112 fluxgate magnetometer, the EHSI, or the wiring between these. But I doubt the last two since why would it occur only on those headings?

It cannot be reproduced on the ground, regardless of the rate of turn. It also cannot be triggered through high-G maneuvers in any direction. And steady flight on the headings in question does not cause it either.

Can anybody offer any ideas?

If it is likely the KMT112, I can buy one of them, but presumably the compass accuracy will be lost and a new compass swing will be needed. How bad is this likely to be? 5 / 10 / 50 degrees?

It doesn't appear a safety critical issue but obviously needs to be fixed sometime, and in many places in Europe there are no avionics facilities to set up a freshly installed KMT112 (if there is a setup; there is a compass adjustment in the EHSI).
 
Last edited:
Peter: I tend to suspect the EHSI - the behavior you are experiencing has a "programming" feel to it, and I suspect (admittedly, without a shred of actual knowledge) that the Sandel unit is the only part of the equation which has sufficient "smarts" to contain such an anomaly.

That said, however, I would be interested in seeing how the heading information is conveyed from the remote compass to the Sandel - is it a serial data stream or some sort of parallel scheme? If the latter, then this behavior could well be symptomatic of a bad connection on one line only.
 
peter-h said:
I have put a little movie here
http://www.zen74158.zen.co.uk/temp-files-ph/ehsi-valid-error.wmv

The compass system comprises of a KMT112, KG102, and the Sandel SN3500 EHSI.

It has been working fine since February 2011.

I have recently noticed the weird behaviour whereby the EHSI compass scale changes from white to red (indicating a "heading invalid" signal from the KG102) when one turns right through 090 or turns left through 270.

It happens for turn rates between about rate 0.5 and rate 1.5. Higher turn rates do not show it, and lower turn rates do not show it.

The KG102 has just been replaced with a freshly overhauled unit from the USA, so it can't be that.

As I see it, it could be the KMT112 fluxgate magnetometer, the EHSI, or the wiring between these. But I doubt the last two since why would it occur only on those headings?

It cannot be reproduced on the ground, regardless of the rate of turn. It also cannot be triggered through high-G maneuvers in any direction. And steady flight on the headings in question does not cause it either.

Can anybody offer any ideas?

If it is likely the KMT112, I can buy one of them, but presumably the compass accuracy will be lost and a new compass swing will be needed. How bad is this likely to be? 5 / 10 / 50 degrees?

It doesn't appear a safety critical issue but obviously needs to be fixed sometime, and in many places in Europe there are no avionics facilities to set up a freshly installed KMT112 (if there is a setup; there is a compass adjustment in the EHSI).
Click to expand...

The only thing that comes to mind is that this must be related to the direction of the Earth's magnetic field "lines of force" which form an angle to the horizon approximately equal to your latitude. At your location those lines are getting pretty steep so when your airplane is banked to the right heading east or to the left heading west the field is nearly parallel with the plane's vertical axis and as a result the output of the flux gate is very weak and/or ambiguous at that time in the turn.

It's possible that this is "normal" behavior at high latitude or perhaps there's some sensitivity threshold that's mis-adjusted.
 
Last edited:
It is notable that the issue occurs on reciprocal headings according to the direction of the turn, and merely flying *on* the suspect heading never shows it.

I did email Sandel twice but with no reply so far. Maybe the man I emailed is away.

The schematic is here and the general project notes are here. The SN3500 gets the heading data from the old KI-525 HSI connections; that part was not changed and indeed I could reinstall the old HSI within an hour or two (but it is a pig of a job).

I don't know enough about the KMT112 but it looks like something inside it is "catching" on those headings. But it has no moving parts apart from a "dangling" sensor, AFAIK.

I am sure this behaviour has not been there before. I also don't think it can be a loose wire because it is just too specific and repetable, yet occurs only in very narrow circumstances.

One thing I didn't try is flying with the KFC225 AP in NAV mode. In HDG mode it works fine, but it only uses the HDG ERR voltage then. In NAV mode, I believe, it uses the actual heading value, no?
 
I doubt if it's wiring or a connection, and intermitent problems don't only apprear at certain headings.

You say the DG was recently replaced. Was it replaced for exactly this issue or for another issue and now you are ruling it out because it is a freshly changed part? I bring this up because DG's are a high failure item, would be a likely cause, and just because a part has been recently installed does not mean that it could not be "bad from stock". Addtionally as DG's go bad you can usually see it on the ground when turning because your rate of turn on the ground is much, much faster than in flight.

If you go into the config menu on the HSI do you get any logged faults or error codes after the flight? These newer units are pretty good at self test. If it's not faulting itself (meaning the EHSI) than I'm thinking the flux valve may be going bad and the excitation signal from it to the DG is weak at certain headings.

Does your aircraft have a switch where you can operate the HSI in the unslaved mode? If it does, fly them same way in this mode and see if the probelm disappears.
 
The DG was replaced after this issue was seen, but the replacement has not changed it - as far as I can tell.

I no longer have the KA-51 (KA-52?) slaving accessory because that item is removed when you install the Sandel EHSI.

I am interested in finding out whether this *could be* the KMT112 fluxgate, because that item is relatively cheap to just change, and this aircraft is nearly 10 years old now.

I am happy to have changed the KG102 even though it has not made this particular issue go away because the new one (purchased from the excellent Castleberry Instruments) starts up about 1/3 faster than the old one.
 
Okay, pull up the maint menu as follows:
Prior to applying power to the SN3500, depress and hold the [BRG] and the
[MEM] pushbuttons, then apply power to the unit. Continue to hold until the
first maintenance menu appears. This protocol insures that maintenance
menus cannot be called up accidentally during flight.
B. Once the Maintenance Menu is entered, press the [NEXT] or [PREV] softkeys
to cycle the MAINTENANCE MENU pages. Use the UP/DOWN arrow keys
for selections, and rotate the right knob to adjust and select. On some menus
additional soft key legends will appear as prompts.
C. Escape the maintenance menus by pressing and holding the course pointer
knob (left knob). This will allow normal operation of the unit to test the effects​
of settings. Re-enter the maintenance pages pressing and holding the course

When you are in, the compass menu is #3 I think. Go to the config and assure everything is selected properly.

Assure you have the KG102 for the HDG listed and the 112 for the fluxgate. The EHSI does the internal slaving.

So it would either be the flux valve or the EHSI... The flux is the cheaper item to replace, and I would go there first because as mentioned before, these newer electronics are pretty good about faulting themselves if they have a problem. You didn't mention any fault codes, so I would try overhauling or replacing the flux first.

Last, you mentioned you could swap out the EHSI for your old HSI and see if the problem stll exists. I don't think that will work because your old HSI requires an amp to drive the compass card and that part is now removed.

Hope this helps...
 
Mark,

Many thanks.

The config I originally set up is here. I suppose it may have changed (got corrupted); will check.

AFAIK everything in the EHSI is "just software". There is no slaving hardware in there. They emulate a KI-525 in software, decoding the KG102 gyro (synchro) outputs directly and doing the same with the KMT112. So (as a hardware/software developer myself) I find it hard to see how it could be the EHSI, and be doing this only on those two reciprocal headings, and each time doing it for just 1-2 seconds. If there was a more consistent problem with the Gyro Valid signal, that could be the EHSI because there could be a fault on that particular analog input (which I think goes into a MUX which then drives an ADC).

Unfortunately, my understanding of the KMT112 operation is limited, but if a faulty KMT112 could explain this problem then it is worth changing. One can get them for about $700. A replacement SG3500 is about $9000.

The wiring was done so the KI-525 can be reinstalled directly, just by putting it in and removing the SN3500 conversion harness, and reconnecting the two connectors to the KA51 slaving accessory (which remains in the panel but is unused). But this is obviously an undesirable end outcome...
 
Last edited:
At the risk of being obvious (I am a long time separated from chasing bits and signals for a living, but... I used to!), have you monitored Sandel connector P1-19 while flying, to determine whether the SN3500 is momentarily losing its "Gyro Valid" signal when the momentary glitch occurs? Is it logistically possible to monitor this signal level while (excuse, me, whilst) flying?

Apologies in advance if I have overlooked something obvious here, but it seems to me that it would be very useful to know whether the Sandel is simply announcing a reported condition (loss of "Gyro Valid" signal), or is somehow internally generating a momentary heading invalid indication through a combination of conditions not properly handled in programming.
 
Indeed - that would have been my next option, to attach a wire to the KG102's Valid output and monitor it in flight with a scope. I have a scope and a 24V to 240V inverter which I was using for vibration measurements.

However, with the KG102 replaced, any Gyro Valid signal issues could be only the wiring, and the Socata wiring is done to a very high standard.

I just got a reply from Sandel saying that this kind of issue is seen when a KMT112 fluxgate starts to go weak.

So I will get a new one of those... my one is 10 years old.

Re your 2nd paragraph, yeah.... a possibility but not easy to establish because only the programmer at Sandel is likely to know the full details. The SN3500 does it all in software (synchro decoding, slaving, the lot) AFAIK. Their reply does indicate that the change from white to red is triggered not only by the Gyro Valid signal being invalid (which is obviously seen at system startup) but also by various other things like the KMT112 outputs being below some Sandel-set amplitude. And yes the KMT112 output will vary with the heading, won't it? A fluxgate output is very variable; it is more the phase which is relevant AIUI.
 
Last edited:
Gotcha.

Over 20 years in the law now, and I find myself more interested in this troubleshooting, bit-chasing issue.

I'll look forward to hearing the result of the flux gate (make sure they don't send you a flux capacitor by mistake, or you'll have issues over 88 MPH!) swaperoo.

I'd have enjoyed seeing pictures of a big, honkin' Tektronix scope on the seat, in flight...
 
SCCutler said:
I'd have enjoyed seeing pictures of a big, honkin' Tektronix scope on the seat, in flight...
Click to expand...

Heh... they make some darn good "handheld" scopes these days, though. LCD screens, a little hard to read in sunlight, but a LOT lighter than the old stuff.

Of course, who am I to talk... I still drag an IFR 1500 around to work on radios... I should take the lead-acid battery out of it.
 
That TEK scope is actually quite light, and is solid-state. The others I have I would not want to carry on the airplane: TEK2465B 400MHz, and a couple of old but really good Iwatsu scopes from the 1980s :)

I keep a lookout for a used IFR4000 but they cost way too much.

Anyway, I have ordered a KMT112 and will report the findings.
 
Hate to say this, but the problem is from the KG102A, that is a "rate of turn" error, in the KI525A should show a red HDG FLAG, intermittent. You should find another KG102A as a loaner, and try again.
 
Last edited:
If it was strictly a rate of turn issue, shouldn't it be reproducible on the ground? And why would it happen only on turns through 090 or 270?

I think Lance's idea makes the most sense: it must have something to do with the orientation of the airplane (bank angle) with respect to Earth's magnetic field. The fact that the direction of turn that provokes the malfunction is opposite for eastbound vs westbound headings points so strongly to that that I can't imagine any other mechanism that could account for that particular set of observations.

But whether it's the KMT-112, or the DG or the interface between them, I don't know enough to say... and I should know more, as I have one of these systems myself.
 
The KG102 was changed for a new (overhauled) unit, and this made no difference.

The KMT112 was changed for a brand new unit, and this made no difference.

There is not a lot else left...

It cost me $2000 net (after exchange costs, but excluding shipping avionics to/from the USA - another $400 or so) to swap the two above units. I don't really regret it as both were 10+ years old anyway, but there is not a whole lot else it can be other than the SN3500 EHSI itself, or maybe a very subtle wiring problem which perhaps attenuates the KMT112 excitation voltage (the output of the KMT112 does apparently vary somewhat with the heading). I have emailed Sandel 2x with no reply so far.
 
peter-h said:
The KG102 was changed for a new (overhauled) unit, and this made no difference.

The KMT112 was changed for a brand new unit, and this made no difference.

There is not a lot else left...

It cost me $2000 net (after exchange costs, but excluding shipping avionics to/from the USA - another $400 or so) to swap the two above units. I don't really regret it as both were 10+ years old anyway, but there is not a whole lot else it can be other than the SN3500 EHSI itself, or maybe a very subtle wiring problem which perhaps attenuates the KMT112 excitation voltage (the output of the KMT112 does apparently vary somewhat with the heading). I have emailed Sandel 2x with no reply so far.
Click to expand...
Peter, I believe you could simulate this problem on the ground simply by extending the cable to the magnetometer and removing it from the plane so you can simulate it's movement in a turn. I think you'll find that you'll get the same failure if you orient the magnetometer to a 90 degree heading then tilt it as if it were in a 30° bank to the right. If that doesn't precipitate the problem you might also need to unmount the gyro and have someone rotate and tilt it as well although I doubt that would be necessary for a mechanical gyro like the KG102 (Sandel's solid state replacement gyro senses bank angle so it almost certainly would need to be moved along with the magnetometer.

I mentioned in an earlier post that I'm pretty certain this is related to the fact that the axis of the magnetometer is nearly parallel with the vertical axis of the airplane (and the magnetometer) in a right turn passing through 90° or a left turn passing through 270°. As a result the signal from a two axis magnetometer will be very weak, non-existent, or erratic in those scenarios. I have wondered if Sande's solid state gyro wouldn't solve your problem and whether or not they have a three axis magnetometer that would work.
 
Last edited:
Subject to monitoring the gyro-valid signal, I am throwing down on Sandel's logic, or (perhaps) interface circuitry.

Assuming no transient invalid signals on the inputs, a periodic and entirely predictable spurious indication like you have described, on a system which is fully-functional with its original parts & pieces (should I say, "kit"?), almost has to be the SN3500's programmed logic. It is either not "hearing" properly, or it is misinterpreting what it "hears."

BWTHDIK?

---
Edit:

As an aside, one of my airport mafia buds is putting in an Aspen, and he keeps trying to tempt me with the 3500, but it seems to me that by the time I spend the $$$ to install it, I'd be more than halfway to an Aspen for myself. What sayeth you?
 
Last edited:
Problem #1 is that I cannot be 100% sure the problem was not always present. To spot it, you have to be watching the EHSI during that 1-2 second moment, during a turn through the right 090 or 270 point. Given the SN3500 was installed ~ March 2010, I may have simply missed it. I have just flown to Greece and back, 20hrs in the air, and never noticed the problem...

I am suspecting the SN3500 (a reduction in sensitivity on the KMT112 signals would probably do it) but I will get all the wiring checked at the next Annual. (I don't have an avionics shop to hand, and am not allowed to work in any hangar unless having paid maintenance done :) ).

Monitoring the KG102 Valid wire in flight is fairly doable and I will do that if the wiring check shows nothing. But the KG102 has already been changed, for a fairly new unit ("fairly new" in the avionics scene means a serial number sometime after WW2 ;) but actually the S/N is similar to my old one which was year 2001) so I would be amazed if this showed anything. And it came from Castleberry Instruments which I find 100% excellent.

Based on my (not complete) understanding of how the KG102+KMT112 system works, I don't think rotating the KMT112 alone will do anything useful, because the "display device" (the SN3500 in this case) works as follows:

a) at aircraft startup (or any other time the gyro is not Valid) it grabs a heading directly from the KMT112; this rotates the compasses to the current heading almost immediately after power-up, and if the KG102 dies you still have a flyable aircraft, with the compass being just a little erratic

b) once the KG102 Valid signal goes valid, it then tracks the KG102's gyro outputs, and

c) as a slow background process, it uses the KMT112 heading to continually correct the heading

The KI525 does it more obviously (I have the schematics) whereas the SN3500 does it all in software.

So just rotating the KMT112 is not likely to do anything much, because the new heading will not be reflected in the KG102's outputs.

One old avionics man told me that the HSI will show the Heading Invalid flag if the KMT112 heading is changing more quickly than the KG102's slaving outputs can track it. I don't know if this is true, but I have already changed both of those! And the fault does not reveal itself in fast turns e.g. 45 or 60 degrees.
As an aside, one of my airport mafia buds is putting in an Aspen, and he keeps trying to tempt me with the 3500, but it seems to me that by the time I spend the $$$ to install it, I'd be more than halfway to an Aspen for myself. What sayeth you?
Click to expand...
A very good friend of mine is replacing his EFD-1000 for the 3rd time, following another total failure. He has had two total failures in IMC and one in VMC. His unit lasted just marginally over 1 year so while Aspen are supplying a new one, he is having to pay for the installation (4 figures). Apparently Aspen suggested that the failure was caused by a nearby thunderstorm... I cannot believe somebody would say that with a straight face to a pilot :mad2: Anyway, you can guess my answer :) You can't blame these failures on a mis-located remote sensor. I think the stuff is basically junk. I would not have it if you installed it for nothing.

I really like the SN3500. It is the best instrument in the panel. Totally sunlight readable, very clear, works just beautifully. I did the installation design myself. Such a pity that Sandel have got overtaken by the "glass" boat, at the light end of GA.
 
Last edited:
Peter, I have worked on the entire KCS55A system, I know how to check every component with the KTS-152 and KTS-153 test panels, believe me, the problem is in the gyro. Few shops around the country can do a perfect KG102A, most of them ignore how to "synchro" the signals from the gyro can and the gyro electronics. When the gyro can is not sending the right signal, during the "rate of turn" the gyro electronics will "reject" this information. The optic disk in the gyro can is not synchronized, plain and simple.
Could you find a loaner? maybe from another airplane? swap the KG102A and try again. or ship the KG102A to another shop for evaluation only.
:rolleyes:
 
Why would it be happening only on the 090 and 270 cardinals?
 
You must log in or register to reply here.
Back
Top