Failed EGT Probe ?

One can design is to work with either type equally, so this may be down to a limited competence of the designer.

The main thing is to make the system immune to interference. For example I sometimes see a “walkabout” in the EGTs and this turns out to be linked to the wingtip strobes. (Slow walkabout, with a period of tens of seconds or more, is likely to be valve issues).

PetitCessnaVoyageur wrote:

So switching to grounded type is not straightforward, as suggested above.

You should use the onces that are required for your installation.

Peter wrote:

The main thing is to make the system immune to interference. For example I sometimes see a “walkabout” in the EGTs and this turns out to be linked to the wingtip strobes. (Slow walkabout, with a period of tens of seconds or more, is likely to be valve issues).

Using unground one’s can be one way of making the system more immune to interference. It will prevent ground loops, and make it a true differential signal. This methode is also commonly used on better intercom installations, where at least microphone signals are also considerd to be true diffirential signals.

It will prevent ground loops, and make it a true differential signal.

That I don’t understand.

Having been in electronics since the 1960s and working a lot in comms (RS485 etc) I know ground loops are an issue there. But only if there is a significant current flowing through the ground lead. That is easy to avoid – by having an isolated (totally floating) interface or (if the common mode voltage can be assured by system design to be within reasonable limits) by using a differential amplifier.

In the case of an EGT probe, the voltage produced is very small (millivolts) and one has to do this right. But this is easy. Precision A-D converters (well beyond 16 bits) have been around for years and these EGT+CHT instruments have only one ADC and multiplex it. So measuring the signal (both sides of it) plus measuring the cold junction temperature is straightforward.

If one of the probe wires is connected to its tip (a grounded probe), then obviously the corresponding instrument input must not be grounded to the airframe – otherwise you will get a ground loop due to some of the various avionics, lighting etc currents flowing through the ground lead. But there is no need to design the instrument so its input ground is connected to the airframe; that’s just stupid, and invites damage because e.g. the starter motor return just connects to the engine so there is a real risk of getting a few hundred amps returning via the instrument!

If the probe is ungrounded then that risk doesn’t exist, and then the instrument needs to have its input designed so that the common mode voltage is defined.

Anyway, this debate is academic in the case of a real instrument which presumably you are stuck with…

I would be amazed if Garmin released the schematic of the thermocouple amplifier – even to dealers. Garmin release almost nothing down to that level. They obviously don’t want to create the “wooden hut repair shop” type of useful repair support which King’s freely available MMs helped to create.

Peter wrote:

That is easy to avoid – by having an isolated (totally floating) interface

That is exactly what they do? Any induced interference would be about equal at both wires. This, combined with being differential makes the interference have no influenance. You would keep only the wanted small signal.

They must have a differential input, measuring between the two wires, regardless of the type of probe being used. The signal is much too small otherwise. They will also have lots of filtering, due to the massive interference, with most probe wires being unshielded, etc.

But that isn’t quite the same point as having a type of input circuit which requires a grounded probe, or an ungrounded probe, to work properly.

I don’t think the input circuit is fully floating (optoisolated etc). Very few thermocouple instruments do that. Even industrial temperature controllers isolate the mains (obviously ) and most isolate the RS485 comms interface. Many years ago I designed an 8-input TC interface on which all 8 inputs were fully floating but it was a lot of circuitry, with a transformer with 8 secondary windings to power the floating stuff. Nobody does that in avionics.

Peter wrote:

having a type of input circuit which requires a grounded probe

Why does it need a totally (fully floating) input circuit? It surely is a question of common mode range. You need fully floating if the common mode range can be in the hundreds of volts, but I doubt that a grounded probe would need more than a few (low single digit) volts…

I agree, of course.

What I was getting at is that if somebody designs an instrument for a grounded probe, they may well be relying on the ground connection to establish the bias point (the signal ground) for the input amp, and if you float the probe then the Vcm could be out of limits (stupid, yes…). So in such a case you could use an ungrounded probe if you grounded the bottom end of it.

Or, if somebody designs an instrument for an ungrounded probe, they may be setting up the signal ground with a resistor divider (etc) and then the probe being grounded will bugger that up.

What I am saying is that an ungrounded probe should be usable in any situation (but you may hve to ground the bottom wire yourself) whereas a grounded probe may be usable only in an instrument designed for one.

Thermocouple instruments tend to use a pullup resistor for open probe detection (which is why a probe which is about to be eaten by the hot gas initially shows a high temp reading, before failing totally) which is another reason for a limited Vcm range.