EGT - How hot is too hot?

[A64Pilot]

Only failures I’ve seen were in turbines, from excessive over temps literally melting the probe tip, once on a T63-A700, hot start, motor cooked.  Of course I’ve seen several where dirty connections, loose grounds etc. caused erroneous readings, but as those were fixed without changing the TIT harness I don’t consider that a harness failure.

Turbines usually have several, like 7 probes all in a harness that forms a ring that is mounted so that it measures the temp at the entry of the first stage turbine, that being the hottest place.

A Garrett oddly measures the temp of the Exhaust gasses after it’s passed through all the turbines, which I was always taught wasn’t very accurate for a turbine engine, a turbine usually gets the worst heat stress at the first stage turbine so you want to measure temp there, depending on how efficiently the turbine is operating the temp drop to exhaust can be large so you don’t really get a good idea how hot the hottest section is.

But Garrets are weird, good motors, just different.

‘The EI guy could tell us, they have sold thousands and I bet failures are pretty quick and are from manufacturing defects more than anything.

[MikeOH]

My $0.02 regarding TCs as an EE who has a little (not a lot!) of experience designing with them:

1) You cannot calibrate a thermocouple (there is nothing to adjust!) You can only characterize their response.

2) There is a difference in stability (drift over time) dependent upon their quality.  The issue is the purity (homogeneity) of the thermocouple wire.  That is the reason TC wire is sold for use in fabricating a TC (high purity) AND sold as extension wire (lower cost); the extension wire is of lower quality/not as pure.  Over time, at high temps, the impurities 'move' within the TC wire and cause loss of accuracy.  For extension wire at a relatively uniform and low temp this is not a factor.

3) Unwanted/unknown TC junctions can introduce significant errors due to temperature differences between the unknown junction pairs.  Asymmetries in circuits can also be significant: e.g. use of an integrated circuit manufactured with a Kovar lead frame vs. copper.

4) Noise/poor circuit design is often a bigger problem; it doesn't take much to introduce a 41 micro-volt error!  Even 100 degrees is only 0.004 Volt.        

5) US color code: The red wire is actually the low side.  The TC type is indicated by the other wire's color: yellow is K, white is J, blue is T, orange is N, purple is E.

                                        

[A64Pilot]

You “calibrate” a thermocouple at least for the FAA by obtaining a calibration card, that shows multiple test points and how far off it is from actual at each point, really very similar to the compass calibration card that I’m sure we all have.

You actually get the same type of card for the altimeter and A/S indicator too.

[MikeOH]

1 hour ago, A64Pilot said:

You “calibrate” a thermocouple at least for the FAA by obtaining a calibration card, that shows multiple test points and how far off it is from actual at each point, really very similar to the compass calibration card that I’m sure we all have.

You actually get the same type of card for the altimeter and A/S indicator too.

THAT is NOT calibration in a technical sense, it is a correction card; just like a compass correction card.  A calibration would REMOVE the errors and obviate the need for the correction card.

[EricJ]

18 minutes ago, MikeOH said:

THAT is NOT calibration in a technical sense, it is a correction card; just like a compass correction card.  A calibration would REMOVE the errors and obviate the need for the correction card.

It's also correcting for variances in the entire sensing system which includes the probe, cables, connectors, and voltage measuring circuit.   Even if the probes are consistent, if the rest of the sensing system is sensitive to installation variances more than the required accuracy, then a "calibration" is required like you mention.   Thermocouple circuits are sensitive to the metals used in the cables and connectors, and correcting for them in individual circuits can matter sometimes.

[MikeOH]

57 minutes ago, EricJ said:

It's also correcting for variances in the entire sensing system which includes the probe, cables, connectors, and voltage measuring circuit.   Even if the probes are consistent, if the rest of the sensing system is sensitive to installation variances more than the required accuracy, then a "calibration" is required like you mention.   Thermocouple circuits are sensitive to the metals used in the cables and connectors, and correcting for them in individual circuits can matter sometimes.

100% agree; the system as a whole gets calibrated.

 I’m being a bit pedantic in regards to the terminology for the TC itself.
 

[Echo]

Ostentatious.  So much better, you know because four syllables is so much better than three.

[A64Pilot]

14 hours ago, MikeOH said:

THAT is NOT calibration in a technical sense, it is a correction card; just like a compass correction card.  A calibration would REMOVE the errors and obviate the need for the correction card.

I put calibrate in parenthesis for a reason.

What I described is what the FAA calls instrument “calibration” and if your doing any Certification work needs to be done within 90 days of the flights, only a very few instruments are actually capable of actual calibration. The rest you get Calibration cards for, their terminology, not mine.

You even “calibrate” the OAT by putting it in ice water, and comparing its ambient reading to a lab thermometer. That was the one we could sign off ourselves. We calibrated all our own torque wrenches and many other tools. Most “calibrations” were really validations, because your not calibrating a prop protractor for example, but it still gets a calibration sticker.

However a magnetic compass is capable of calibration, I’ve done literally hundreds, but you still need the card because even after calibration, it’s never perfect, but even if you find one that is, you still have to have the card.`

I had two sets of instruments to keep from buying down time and would rotate them out.

 

[N201MKTurbo]

When I used to do work for NIST on DC voltage standards, they taught me a lot about metrology. They use calibration to describe measuring the current and future value of a standard along with determining the uncertainty of those numbers. The standard would require recalibration after the uncertainty exceeded some agreed upon value. (The uncertainties increase with time).

NIST never adjusts anything. Adjusting is the act of changing the value of the standard to get it closer to the correct value. If a standard is adjusted, it invalidates the calibration and must be recalibrated.