Are Type K Thermocouples Being Used in Error? (Steven Heale)

I have some experience with thermocouples. First, the article would lead you to believe that K-types are linear at 41 microVolts/C. They are not linear. Here is a primer on thermocouple use
from Omega: http://www.omega.com/prodinfo/thermocouples.html Note the link to the voltage/temp conversion chart and from the chart, note the non-linearity. Also, if you are measuring voltage with a voltmeter, you need to subtract the cold junction voltage. In a purpose built reader, this is done automatically inside the box where the cold junction lives, usually with a calibrated thermistor. I use mostly type K, but sometimes use type E at liquid hydrogen temperatures where K types have a much smaller voltage constant. Finally, there are grades of wire, mostly based on chemical purity. The better grades have closer tolerance on the alloys for better accuracy.

The article mentioned a brick manufacturer that was considering changing away from K-type because of alloy migration. I want to point out that the article did not mention the time scale. Are they holding at high temperature for years? How many hours at temperature caused a drift of how many degrees? If you are using beaded thermocouples and are concerned about this, it's easy to chop them off and reweld. One way to reweld is to use an oxy-Acetylene torch, and another is to use a variable voltage high current source (a variac does well) and scratch the twisted wires across a steel pad. I've made thermocouples using wire down to 40 gauge (.005 mm^2) using this scratch method, and I've used the oxy-acetylene torch with wires as big as 12 gauge (3.3 mm^2).

@AlainCo
You are referring to Green Rot, which is common in type K in reducing conditions.
The center of an open hot tube, with air in it, actually becomes very reducing, since the high heat creates a local pressure that forces most of the air out.
(Think hot air balloon). At 1000°C, a small gauge type K can go green and fail in a couple of hours in such an open hot tube.

@Dan21
The old timers used to quickly dip the twisted junction into mercury that one side of 120V was fed in order to weld it. Nasty, but effective.

Quote from Paradigmnoia

AlainCo
You are referring to Green Rot, which is common in type K in reducing conditions.

Here is Alain's earlier post on the subject, which was enlightening. My takeaway: k-type thermocouples exposed to 1000+ C and hydrogen can do unexpected things. One wonders whether subtly failing thermocouples might explain some reports of excess heat.

@Eric Walker
Thermocouples fail with decreasing output, as far as any reports I could find, universally, since they generate voltage in order to operate. Properly sized (diameter) thermocouples usually fail after months of service if you follow manufacturer recommendations for temperature and reasonable precautions.
Null TC failure might look like a lower T, but typically this drop as the TC begins to fail is only a small percentage of change from the real temperature, followed by total failure or intermittent connection of broken ends in some cases. A short across the two TC leads will probably move the temperature reading to the location of the short, rather than the hot end. A short could generate an additional Seebeck voltage depending on the shorting metal type and how the leads are making contact.

Rarely, but not super uncommon, is the MgO powder in metal clad protective tube-type TCs have a bit of moisture (absorbed from the air in storage) and a weak voltage develops as steam is released and reacts with the MgO, the TC wires and possibly the enclosing metal sheath. A slow bake out of the protected TC before experimenting will prevent this.

High reading TC's are almost always caused by galvanic action, or other electric leakage. AC leakage is easy to spot and can be filtered. Most TC problems with electrical leakage are very obvious, since most electrical sources will leak a voltage orders of magnitude higher than the normal TC Seebeck potential. Galvanic action can be very weak, and in the order of magnitude of a thermocouple in cases of dissimilar metals and gasses or fluids. Accidental Seebeck voltage can be generated by certain metal combinations. These types of DC electric leakage into a TC are as likely to reduce the TC voltage as increase it, depending the polarity.

In a furnace, a low reading TC can be catastrophic, since the burner will be fired higher in response to a lower temperature signal. This may ruin a product or process or even destroy the burner if it runs over the operating range.

There is a TC problem related to the type of Parkhomov-like experiments being performed that is not being discussed: at ~900C and above, the normally insulating alumina/mullite ceramic tubes become electrically conductive. If the heater wire is in contact with the ceramic and the thermocouple is in contact with the ceramic, a current can be conducted from the heater to the thermocouple. Resistance values of ~20k ohms have been measured. If the heater is driven by high voltage AC, the conduction can show up as high common mode AC signal on the thermocouple. Even if the TC is being measured in a differential mode, the high common mode AC signal can cause measurement noise and DC signal offsets as the AC gets rectified in the contacts and electrical input of the ADC. An intervening ground is needed. This can be achieved using an inconel shielded thermocouple with the TC junction insulated from the sheath, and then ground the sheath.

Also note that SS is a proton conductor at temps >800C. You cannot use SS sheathed k-type TCs in a closed hydrogen chamber at T>800C. You can use them to higher temps if the TC junction is open to the air so that the hydrogen conducted through the SS can easily diffuse to atmosphere.

Quote from BobHiggins

There is a TC problem related to the type of Parkhomov-like experiments being performed that is not being discussed: at ~900C and above, the normally insulating alumina/mullite ceramic tubes become electrically conductive. If the heater wire is in contact with the ceramic and the thermocouple is in contact with the ceramic, a current can be conducted from the heater to the thermocouple. Resistance values of ~20k ohms have been measured. If the heater is driven by high voltage AC, the conduction can show up as high common mode AC signal on the thermocouple. Even if the TC is being measured in a differential mode, the high common mode AC signal can cause measurement noise and DC signal offsets as the AC gets rectified in the contacts and electrical input of the ADC. An intervening ground is needed. This can be achieved using an inconel shielded thermocouple with the TC junction insulated from the sheath, and then ground the sheath.

For sanity reasons: Why do these experiments not use a battery? This would provide a bottom of the energy used. The efficiency of invertid rectifiers is well known.
If You do this inteligently (more than one battery) then you can charge and unload the same time.
An even more clever (higher energies) approach could be a mechanically decoupled motor/ generator module!

Bob Higgins brings up a special case which is interesting.
A coil leaking voltage into a weak conductor like mullite will probably leak a low voltage into the thermocouple, because it will probably not leak the full potential supplied to the coil, but instead a potential that is between two adjacent coils wraps, which can be fairly low depending on the number of wraps. This depends on the thermocouple circuit very much, as much as the coil characteristics and the coil cement (if used). The potential between one end of the coil and the thermocouple could also be delivered, but should be very obvious, since it will give a high numeric change to the TC reading (probably overlimit). A voltage between two wraps could easily be similar in voltage to a real Seebeck voltage. The placement of the thermocouple can significantly affect the leakage. Imagine a case where the thermocouple is installed so that the leads from the junction are parallel with the coils, and between two coils.

Additionally, not only is mullite conductive at high temperatures, but sodium based cements are even more conductive.
In the first video below is an exaggerated case, but remember thermocouples operate in the mV range.
The second video shows thermal runaway in conducting glass, which may explain some mullite-sodium cement tube failure modes.

@Wyttenbach

Certainly a battery could be used, but it would have to be a big one! For example, take GS5.3 by Alan Goldwater and MFMP ... This experiment ran for days with an average input power of approximately 500W. Let's say this was 3 days = 72 hours. Then the battery required at 100% conversion efficiency would be a 36 kWH battery. For a 12V system, this would be 3000 AH extracted, or full discharge of about 30 car batteries. To avoid damage the lead acid batteries, you don't want to ever discharge them by more than 50%, so you would need 60 car batteries. This would cost more than the rest of the apparatus!

@Paradigmnoia

Note that most thermocouples should be best measured in differential mode by the ADC in the data acquisition system. In such a configuration, the leaked AC would amount to a common mode signal while the junction voltage would be differential. You also get some common mode signal from picked up stray AC on the twisted thermocouple leads - the leaked AC voltage could be greater than this, and also if the heater drive is triac switched, you will get all of those switching voltages as common mode on the TC. You don't want the common mode to be too high amplitude or too high frequency, because either case can cause common mode to differential conversion. Conversion can take place even at the clamp contacts of the TC input to the ADC because you have different metals and oxides present that can produce a partly rectifying contact.

@BobHiggins
Thanks for your elaboration.
The coil leakage scenario is something I will be carefully trying out by experiment at some point. I am not sure how the wrap-to-wrap voltage and the TC to coil end voltage interact in real life. I did have AC come through a TC that had the coil intentionally embedded in glass (liquid at operating temperature, to improve heat conduction to the tube). That was a bit scary and luckily did not damage the TC logger which was battery powered. But it did cause the logger to log data in channels that were not even plugged in, and a bunch of low T records in the active TC channels. The glass had a considerable amount of ZnO in it, which may also act like a rectifier.

Edit: the 9999 are normal open TC readings.. The TC signal temperature actually dropped.
There is a drop in the second TC channel before the first channel gets it (just above highlighted area).
Below is a slice of the data logger file from one of the conduction shorts.
Edit: Graphic summary of experiment added.
Note that the bulb continued to operate for about 4 minutes with the glass broken probably due to the rarefied hot air inside the tube. The liquid glass sealed the ceramic crack, but ran though the crack onto the bulb inside.

Actually, the T drop before the big jump in the other channels is slightly similar to the Jiang experiments, but his were all in the same TC channel.
Maybe that is coincidence.

Quote from BobHiggins

Then the battery required at 100% conversion efficiency would be a 36 kWH battery. For a 12V system, this would be 3000 AH extracted, or full discharge of about 30 car batteries.

The following company builds large, 'grid usable' batteries.: http://www.eosenergystorage.com/ 200$/kWh

It sounds affordable.

You don't need a LiH battery. This is far too expensive.

Quote from Paradigmnoia

Thermocouples fail with decreasing output, as far as any reports I could find, universally, since they generate voltage in order to operate.

I interpret the discussion of your post to indicate that green rot and gradual TC failure will not be much of a concern in LENR experiments, as they will result in a lower voltage (temperature) reading than would otherwise have been the case, unless this kind of failure occurs on the null TC (which, presumably, it could).

Quote from BobHiggins

If the heater wire is in contact with the ceramic and the thermocouple is in contact with the ceramic, a current can be conducted from the heater to the thermocouple.

This sounds like a common setup. I do not recall reading about grounded TC sheaths, either in the cold fusion conference reports/JCMNS articles or in the hobby experiments. Perhaps I was just not paying attention to this detail. What amount of uncertainty would this particular error introduce, I wonder.

Quote from Paradigmnoia

Note that the bulb continued to operate for about 4 minutes with the glass broken probably due to the rarefied hot air inside the tube. The liquid glass sealed the ceramic crack, but ran though the crack onto the bulb inside.

Can you tell more about your setup/reactor?

@Eric Walker
Yes. That is a good summary. I looked very hard for examples of TCs failing with higher output, but they were always due to external voltages when the temperature reported by a TC was higher than the real temperature.

@eros
This was one of several attempts at simulating excess heat, to characterize what it might look like. There was no PID control. In this test, a very early version, I rushed the job, and did not put a dimmer control on the bulb. I foolishly just turned it on. The temperature rose about 300C in 12 seconds, and continued to climb over a minute and a half or so. The tube was 3/4 inch OD ceramic, with 1/2 inch ID. The coil wire was embedded in a thin layer of custom glass, and the tube had been run before several times empty. The bulb was 500 W, a thin halogen bulb for a work lamp. In this experiment, it would have been equivalent to a COP of two, if I had managed to get the coil to full power. The tube failed early, but the coil seemed fine, so after confirming that no more ceramic breakage was likely, I turned on the coil. Almost immediately the bulb broke, with a pop, and the light brightened. At this point the TC signal went crazy, so I turned off the coil. The bulb was still lit. The tube was open on both ends. The thermocouple signals returned to normal, but the outer thermocouple was clearly stuck in the glass. It was carefully pulled out of the glass with a ceramic rod, and it was noted that the insulation only was sticking to the glass.The inner thermocouple was wrapped around the inside bulb. After the experiment cooled, it was found that glass has poured through the cracks in the ceramic, and spilled onto the inside thermocouple and bulb. The tube had to be cut open on a diamond saw to examine the inside. There was glass from the drips inside the broken bulb, and bulb glass stuck to the tube around the bulb, in the outside glass. The outside glass was beer bottle brown, so telling the glasses apart was easy.
Of particular note was that a sudden heat burst could be easily fatal to the tube, and that the glass conduction improver was as much of a problem as it was solving another.
The glass was able to make the outside temperature almost the same as the inside temperature, however, when heated with the coil. I presume that the glass improved heat transfer by conductivity such that radiative heat transfer was supressed to a much higher temperature than typical. The glass would liquefy at around 670C, and actually wick up to the coil.

@Alan Smith
Quartz is not fused alumina. I saw that you said that once before, but assumed it was accidental.

Quartz is fused silica (SiO4), while alumina is made of aluminum oxide (Al2O3).
Sapphire glass is made from alumina, but that would be rather expensive.
Mullite is combination of silica and alumina.

The Parkhomov cement is almost certainly a high temperature conductor, being a sodium silicate, and additionally possibly a rectifier since it has (in Parkhomov's version) a lot of zinc oxide.

Rossi used Durapot 810 cement, and this may be close to ideal for electrical insulating ability. It is designed as a thermal conductor.