Thermocouple into a rotating tank

Gustavo Raush wrote in news:g62gbc$5un$1 @defalla.upc.es:

Not to mention that the connection through the slip rings is another thermocouple!!

I don't know what kind of accuracy you need, but for best accuracy, the thermocouple and thermocouple amplifier must be on the same side of the slip rings.

--
Scott
Reverse name to reply

As long as it's the same thermocouple on both wires, it should be OK.

That means you're accuracy becomes limited by the cold junction sensor in the amplifier. If that were acceptable, a simple thermistor could be used instead of a thermocouple.

Heck, if that were OK, put a battery powered MCU with some DS2762's and a zigbee radio inside the tank and skip the slip rings completely. Or an MCU with 20 thermistors.

Scott Seidman escribió:

Hi Scott:

Thanks for your quickly response. The needed accuracy is about 0.3 to 0.5 ºC, worst values are not allowed because several heat transfer numerical models must to be validated. Do you know some published article which could be explaining experimental setup like this?. Not exactly the same but which could be given details for an infrastructure with similar problem to measure.

Thanks.

If you used nice high value 100K thermistors slip ring resistance would be no problem, just noise, and noise can be filtered. You can calibrate a temperature to a very close resoluton, but the problem is temperature range. Whats you range ? I lost track of the total number of channels needed, but I think your biggest problem is getting a calibrated system using high value thermistors of 1K or more. Most systems on the market are not that.

greg

GregS escribió:

think your biggest problem

Most systems

Hi Greg:

Your idea it not so bad. Could be is the best solution because the data acquisition units will be used is a well-known Agilent 34970+34901 module (Digital Acquisition Switching Unit+Customizing-Module). This DASU+Module tandem allows a wider range of thermistor sensors where the

10K value is permitted. It is not clear if the 100K is also included but I think 10 K is OK. Accuracy reported by the maker is 0.08ºC.

About the calibration sensors, in my case it is not a problem because I will be able to carry out a good one.

Thanks by the shown interest.

DJ Delorie wrote in news: snipped-for-privacy@delorie.com:

All good points. I have a number of slip ring systems, and have often threatened to replace them with telemetry based systems if any of the rings broke. I'd still need rings for power, as batteries wouldn't be an option, but that's sort of a different beast, and the ring count would still go way down.

As to whether the extra thermocouples would do anything bad, if the 'couples were wired directly into the rings, you wouldn't have the same thermocouples. Same if you extended the 'couples with copper wires. I've always found thermistors to be more worry-free from such design considerations.

--
Scott
Reverse name to reply

snipped-for-privacy@zekfrivolous.com (GregS) wrote in news:g62kgg$qgb$ snipped-for-privacy@usenet01.srv.cis.pitt.edu:

I think if the temp range permits, thermistors would be the way to go. Much easier.

--
Scott
Reverse name to reply

If both slip rings are made of the same metal and are the same temperature, and you use a suitable thermocouple extension wire on the outside, it works. The voltages induced by the slip rings cancel out, just like any other metal connector you use with thermocouples. I'm not sure what it would do for the accuracy, though - that would depend on any thermal gradient across the slip ring assembly.

You can do it with copper wire if you take into account the temperature of the slip rings when converting the thermocouple reading.

DJ Delorie wrote in news: snipped-for-privacy@delorie.com:

same

So, if you have a chromel-alumel thermocouple for example, and you attach it to a slipring of connector metal X, the chromel-X junction will exactly cancel the alumel-X junction??

I thought you need a second chromel-alumel at a ref temp wired in, then you could have two chromel-X (or alumel-X, your choice) isothermal junctions that cancel each other. Actually, I'm pretty sure of this, and there's an example at

I dislike doing precision thermocouple designs, but they're really not magic. In any case, thermistors are easier, but don't have the same temp range, and thermocouples can be faster.

--
Scott
Reverse name to reply

A common misconception in thermocouple design.

The junction is irrelevent. It's the temperature change ALONG THE WIRE that makes the voltage change.

A length of wire with a temperature gradient along it will act like a uV battery. The uV/C ratio depends on the metal. A thermocouple is two of these in series, meaning the voltage at the junction is different than the voltage at either connection. The chromel wire has a voltage change of X, the alumel wire has Y, you measure X-Y. If you had a third wire Z, you'd measure X-Y-Z (or X-Y+Z depending on how you interpret things). If you know the combined voltage and the temperature at all but one junction, you can calcuate the temperature at the remaining junction.

So, if the slip ring is isothermal, and you have two copper conductors to it (assume those have the same gradiant as each other) and a K thermocouple on the inside, what you have is FOUR seebeck effects happening, in series. The three junctions are the two slip rings and the tip of the thermocouple.

Since the two copper wires have the same composition and temperature gradient, they'll have the same voltage across them. Since we're measuring differential voltage, and the copper is providing a common-mode voltage, they cancel out.

The only catch is, you have to know the temperature of the slip ring.

In the pdf you reference, this is Figure 12.

Yes, more generally the error will be roughly (roughly because it's not quite linear) equal to the difference in temperatures between the pieces of metal X. This is probably your biggest issue.

Modern instrumentation allows a variable 'reference' temperature which is an isothermal connection point where the thermocouple wire is connected. That temperature is measured and used to compensate for the so-called "cold junction" temperature (often digitally these days, which allows for nonlinear comnpensation. You don't need to worry about it unless you're designing the signal conditioner.

The 60m ohm is no problem at all. Modern instruments have input impedances in the M ohms, but you might have issues with noise from the slip rings, depending on your desired bandwidth. Keep in mind that most instruments pass a small current (I typically use something like

500nA) through the sensor to detect failure. That will result in 0.1°C error for every 8 ohms of resistance (or resistance change).

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

What temperature range do you need?

A thermocouple mesaures voltage, so the resistance of the rings is not important. But they will form a thermocouple too, where your thermocouple wires are connected to the rings. If this is all at the same temperature, it should all cancel out, but it is likely to give problems.

Some other options are PT1000 elements or the modern precize NTC thermistors. Or maybe placing some electronics (that can handle the temperature range) in the tank and transmit the signals digitally? I would go for this last option, but notice that the electronics will have to take it's own temperature into account.

Pieter

How about using IR instead of contact? you could simply mount each receiver near the moving area of interest.. the other way would be to seal a complete control unit inside and have a set of slip rings to supply the power and the control unit can wireless transmit the values..

"

RTD's also may be a suitable replacement over using thermocouples.

The RTD's can be commutated easily over slip rings.

Just place a small cap across the connection.

"

Why are you using thermocouples? I'd look at RTD's, or National's line of sensors [LM35, I think], and some kind of scanner-mux. The slip rings could provide power and they or RF the data out.

--
A host is a host from coast to coast.................wb8foz@nrk.com
& no one will talk to a host that\'s close........[v].(301) 56-LINUX
Unless the host (that isn\'t close).........................pob 1433
is busy, hung or dead....................................20915-1433