Generator Embedded Thermocouples

[Rodmcm]

I am refurbishing a 50year old generator which has embedded type K thermocouples in the stator. We attached thermocouple/4-20mA transducers direct onto thermocouples at stator junction box. We find that the temperature reading on all 6 is fine when there is no generation, but raises by some 30 deg once the generator is on line and generating.Certainly not a heat effect as it rises as fast as we raise current (MW or MVAR). Tried earthing one side of thermocouple with no effect. Have tested and verified no pickup in the transducer or 4-20mA leads when generating. Only conclusion is pickup in thermocouple leads in stator. Any suggestions?

 

[dpc]

Sounds pretty normal. When generating, windings get hot due to resistive losses. This will increase with load. Even old Class A insulation can operate at up to 105 deg C.

Generators and motors can operated normally at temperatures high enough that you cannot keep your hand on the outside of the machine.

 

[itsmoked]

dpc; I believe Rodmcm is referring to an instantaneous rise in measured temperature rather than a reasonable rise over time.

T/Cs require such a small voltage for a given temperature I can easily see this problem. I'd rather RTD's or thermistors but..

Can you disconnect your 4-20mA converter and stick the leads directly to a DMM? This will tell you if the converter or the T/C is where your problem is occurring.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Rodmcm]

Yes we realise this, but the time scale is too quick. If we raise/lower MVAr instantly then the temp raises/lowers immediately, ie no thermal lag

 

[ScottyUK]

The stray fields around a generator are powerful. Most likely is that your t/c-mA converter is picking up the generator magnetic field and it is is being demodulated or coupled into the output of the transducer.

To give you some idea of the energy available in the stray field, we managed to get an old Philips Scopemeter running for a couple of seconds from its wall-wart plugged into an extension cable maybe 10' away from the outer casing of the machine. Only thing was, the extension cable wasn't plugged in. After the initial bemusement we figured the smoothing capacitor in the supply was being charged up by the stray field while the instrument was off and held enough charge to allow it to switch on for a couple of seconds before fading out. The battery wasn't even in the instrument.

We have no problems of the type you're experiencing with t/c's mounted in the stator of big 16kV generating sets and those t/c's are routed directly to DCS I/O cards a couple of hundred metres away. I honestly don't think the t/c is the guilty party. Can you install comp. cable and get the transducer away from the generator? Further the better - distance is usually the most cost-effective way to reduce magnetically coupled interference.

----------------------------------

Sometimes I only open my mouth to swap feet...

 

[Rodmcm]

Yes we thought of this. We shorted the inputs to two of the thermocouple units and ran the machine up. The temp reading oin the two units was funny but stayed the same for off line and on line conditions. From this we deduced that there was no interference to the transducer and 4-20mA wiring. The transducer box is actually mounted in the ais gap around the metal clad stator and the outer concrete support, ie where the water coolers exit air recirculates. A metal pipe in the stator surround allows access to these thermocouples.

The original scheme used a switched milivolt meter (no auxiliary supplies at all) with one side of the thermocouples commoned, ie a 7 wire system.

 

[itsmoked]

Shorting the TC inputs is NOT a difinitive test proving the problem to not be the complex instrument, verse the piece of wire.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[byrdj]

Is the junction box still open? I've had problems with test equipment trying to oerate beside an operating stator. I've used metal box as EMF shield. I would think J box would be EMF shield If cover is on.

I would also try previous recomendation to move one of the 4/20 transducers far away and connect it with extension lead and see if that helps

 

[Rodmcm]

To Keith Cress
Please explain why you do not think this is valid

 

[Rodmcm]

To Byrdj
Yes we used a metal box and it is a complete enclosure

 

[electricpete]

Just thinking out loud. T/C's seem inherently prone to noise. One would suspect that the unwanted noise is ac and the desired signal is dc. Putting a scope on the t/c output could provide a confirmation of the unwanted ac noise.

Aside from checking the wiring routing, twisting and shielding to remove the induced voltage, there might be some filtering options. Maybe your transducer has some kind of an input filter to remove ac noise? If not, maybe try a cap across the transducer input (acts like an open circuit to dc but a shorting impedance to ac).



=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[itsmoked]

Hi Rodmcm; Because the problem can be the converter's inability to reject common mode noise coupled onto the T/C not 'the T/C'. Shorting the T/C may just remove the large common mode signal, leaving you, thinking all the problem is with the T/C.

It's rather like having a pressure transducer that is not measuring small changes in pressure because the head pressure is way too high and the transducer is pegged. It's not the 'pressure's fault'.

You will get a very strong 50/60Hz magnetic coupling in that environment with wire twisting being able to help only so much. If your instrument can't handle that correctly you will get erroneous readings.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[ScottyUK]

Expanding on Keith's reply:

The t/c is a low impedance source which is quite difficult to couple noise into, as is the surrent output of the transducer. The rest of the circuit within the signal conditioner will be typical op-amp electronics, lots of small semiconductors with high impedance signals, all prone to pickup of interference.

Screening LF magnetically coupled noise is virtually impossible at a practical level. Earthed metal enclosures are great for reducing capacitively coupled noise, but have negligible effect reducing magnetically coupled noise, unless the earthed metal just happens to be a mu-metal box. At 50Hz or 60 Hz it would be a very thick walled mu-metal box to be effective. Keep away from it remains the easiest solution.


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[electricpete]

"The t/c is a low impedance source which is quite difficult to couple noise into.."

I would have said the t/c is a low-voltage signal source connected to a high-impedance circuit which would be very easy to couple noise into. Would I be wrong?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[ScottyUK]

Hi ePete,

A t/c is definitely source of very low voltage, but it's a source with a very low internal impedance. It's not easy for an external interfering source to impress anything onto the t/c signal, unless the t/c cable forms a loop antenna by virtue of not being a twisted pair. That would be a very poor installation and asking for problems. The signal conditioning circuit itself is likely to have a lot of fairly high impedance nodes internally within the circuit and thus be susceptible to external interference, hence my suggestion to move it away from the source of the interference.


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[itsmoked]

Yes but because the signal of interest is in the differences between 1-2uV any coupling is going to be problematic if the measuring instrument does not 'filter' out the coupled voltages. I'd say impedance doesn't matter in the case of a loop antenna\transformer. In those field strengths twisting is not going to be enough anyway as you have to untwist somewhere to 'hook it up'. The the insulation still leaves a minuet loop and the loops don't effectively cancel as any turns of the wire path subject partial twists to differing fields....

Back to the converter, it better cope correctly with the 50/60Hz and it should be non-susceptible to the fields,(somehow), directly. Speaking of which, the best way to deal with the 50/60Hz in a converter like this one is to to synchronously read the T/C voltage with the noise source, that is, the 50 or the 60Hz. This means any 'noise' is always the same magnitude and can be calibrated out. I say this because if the converter is expecting 60Hz but is in 50Hz service (or vis-verse) you will lose that feature.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Rodmcm]

As a matter of interest for those who participated in this we found that there was an induced wave (almost as if the fundamental had a second harmonic imposed) in the thermocouples that could not be filtered out by a thermocouple unit. We provided an LCR filter on the front end and it all worked. Sorry I misinformed you the Thermocouples are type T m( Cu/Constantine) not K.

 

[itsmoked]

A common error! Glad you got it working.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[ScottyUK]

Hi Rod,

Are these 1500 rpm machines (1800 rpm in the 60Hz world)? I'm curious as to the origin of the 2nd harmonic frequency.



----------------------------------

Sometimes I only open my mouth to swap feet...

 

[Rodmcm]

They are BTH 150RPM 40 pole 50HZ units. We guessed at the 2nd harmonic from a look at the squarish wave present to the transducer. The filter rounded it out to more like 50HZ and the transducers inbuilt common mode rejection did the rest.