temp sensor

[John Ferrell]

I missed this thread the first time around. I will share my experience....
In 1959 I was a Lab Tech in the Reactor Physics Division at Battelle Memorial Institute. We were studying Heat Transfer conditions in Reactor Cores. We used Iron Constantan thermocouples that were fabricated on site. The sensor ends were spot welded to the site to be measured with whatever means we could make stick. The TC wire was continuous back to the operating table. Tests frequently required more than a dozen sensors. Iron Constantan was preferred because it provided higher output than Chromel-Alumel and it was determined that we could measure it more accurately. Each TC connection had its own melting ice bath for reference and there was a couple of multiple double pole switches that put the TC's in the measuring bridge one at a time. The bridge was powered by a standard cell and the Potentiometer was a Leeds & Northrup K2. The balance indicator was a very sensitive Galvanometer. The object under test was heated by either water, steam or electricity. We did the best we could in each case to establish calibration figures for each TC.

The most interesting subject was the study of the nature of water turning to steam in a vertical section under moderate pressure. That is, pressures less than 2000 psi. In addition to temperatures we measured fluid density in the test section using a Cesium Source (radiation) and radiation detector. All of the rig was built onsite for the task. The radiation detector was powered by a case of 45 volt batteries in series. There were still tube type portable radios in general use so every appliance store stocked B-Batteries. The sensor tube had a mica(?) window to pass Beta rays. A deflection magnet (permanent) permitted us to lose the Alpha's.
Did I mention that the whole rig was behind a 3/8" steel enclosure about 10 feet tall?

Sometimes the measurements made obvious sense and sometimes they were so scattered that it looked like human error was the only explanation. I have always thought that those cases were the ones that were of most importance. My personal conclusion was that the interface where water turns to steam at any pressure is unstable. Clearly, higher temperatures and pressures of this event yield higher energy transfer. Although this is desirable for the harnessing of the energy in a reactor core it is of utmost importance that the condition NOT occur in the core. The varying density of the water-steam is faster than mechanical control of the reactor control rods can respond. The resulting power surge SHOULD result in a Scram condition (Emergency Shutdown). If the Scram fails to shut down in time and the reactor goes Super Critical you have a Chernobyl disaster kind of incident.

The point I am trying to make is that Thermocouple technology is more reliable in the long run than any other and should be utilized any time safety is a parameter. Redundancy and fail safe-fail soft techniques are always required if the measurement is important. I received a few Thermocouples from EBAY (China) in the last 2 days for less than $20. I will know more about the subject this time next week. Don't be afraid of thermocouples, they work fine!

Sorry if I bored you!
de W8CCW

[LGR]

John, your conclusion makes no logical sense. The fact that a TC worked reliably doesn't mean the alternatives are inferior. TC is still the best choice for very high temperatures, but Pt RTD is at least as reliable, and more accurate, provided that the temperature doesn't get too high.

I have an "army story" about a TC in a chemical reactor too, and it worked perfectly in a temperature run-away situation. But an RTD would have worked just as well, and would have been cheaper to install. But if you expect the point that you're measuring to melt down, TC probably is a better choice.

Also, keep in mind that a lot of J-type TCs get installed outdoors where rust is a problem. You had an indoor installation that avoided that. Interestingly, moderate temperature can be more of a problem than high temperature because the high temperature keeps the iron dry. If you have to use a TC, type K is just as good, and doesn't have that potential problem.

[janni]

Thanks for the story, John . With some experience there's indeed no need to be afraid of thermocouples. Nowadays one may almost directly interface them to some delta-sigma converters, use digital sensors for cold junction compensation and filter and linearise in code. No big deal .

[joseLB]

wow, with so many experts in temp sensor I´m really affraid to give my suggestion. But here it goes, and please, don´t throw me a bunch of stones, I´m a bit sensitive, ok . Also, that´s the advantage of who is not an expert, like me, we can try strange solutions, he, he.

Using 1 or 2 LM35 (plain cheap linear temp sensor), the metal case ones.

1- put side A of a iron bar or whatever on the hot air flow.
2- at side B, at a distance that will not destroy it, put a lm35 (bar size must be experimentaly defined)
3- as you do not have that need for precision... this bar will act as a temp "divisor". Also, I´m supposing temp speed is not a must.
4- connected lm 35 to a pic adc with ref. voltage = a dione = ~0,67v
5- in order to reduce effect of free air temp changes over bar, use a 2nd lm35 somepoint in middle of the bar
6- so, for a such free air temp and desired temp on bar side A, you measure the temp on both2 lm35.
7- heats or cold the free air , measure both lm 35 again. More measures you do, the better.
8- interpolate results, and I hope you have your temp sensor... cheap, not so fast response, but who knows, it can work...
Jose

[John Ferrell]

Jose, your approach is sound. No need to apologize whatever the results are. If the strategy provides a solution within your bounds it is especially good. If not, what you learn from the project is also valuable. The science of heat transfer and control is critical for steam engines, micro electronics and everything between! Perhaps you can extend your stratagy to a cheap, simple plan to measure soldering iron temperature. I am pretty well equiped, but I still cannot justify the cost of high end soldering stations...

LGR: Sorry I missed your post earlier. Somehow I overlooked the email that let me know there was an addition to the thread. Perhaps I did not state the time frame of my experience. We simply did not have Platinum RTD's in 1959. Also, I was not aware of them until I read your post. All I know at this point is what you have said, MicroChips AN687 Ap Note, and http://www.patents.com/us-5666593.html. Other than expensive and somewhat fragile they do appear to be a better solution than thermocouples. The warning about self-heating when driven beyond 1 ma seems to me to be ideal to implement as a flow sensor for both liquids and gases. I will shop and see if I can afford a couple.

Therocouples are not as simple in reality as the theory would seem. They must be calibrated in place for precision measurments. The are sensitive to stray currents their mass affects their time response. But it remains they are rugged and cheap!
de John, W8CCW

[joseLB]

Jose, your approach is sound. No need to apologize whatever the results are. If the strategy provides a solution within your bounds it is especially good. If not, what you learn from the project is also valuable. The science of heat transfer and control is critical for steam engines, micro electronics and everything between! Perhaps you can extend your stratagy to a cheap, simple plan to measure soldering iron temperature. I am pretty well equiped, but I still cannot justify the cost of high end soldering stations...

Hi John
I agree, it´s a very simple idea, that could suit the initial thread posts, as seems that the original poster would like something not so precise and mainly cheap. So, if my idea (I use a lot lm35, but not in so high temperature). My main goal was to see with all you if it has any chance to work.
The idea behind:
1- iron bar will have a temp gradient. I don´t remember if it is linear with distance or logarithimic, square root, etc.
2- to have this gradient, must exist a temp loose to ambient air thru the bar lenght.
3- at bar start, temp gradient is smaller, as ambient air is close to furnace. as is goes far, probably gradient is higher, as ambient air is colder.
4- to decrease bar lenght, he could use dissipators, like those used in transistors, but probably will be more dependant or ambient air variations
5- ambient variation is a key issue. That´s the reason for the 2nd lm35. Let´s imagine it at about 1/3 from the bar end and the 1st at the bar end. with that the gradient at the 1/2 bar end will be known at any time, besides temperature at each one.
6- I supose the mathematics to do all this gradient is above of my knowledge.
7- so meauring these 2 lm35 temp for distinct ambient temperatures, and knowing the furnace temperature, he could create a table with 3 entries for each measure(FurnaceT, lm1T, lm2T). Time interval must be constant while creating table.
8- now he can interpolate the 2 lmsT above to infers furnaceT.
9- Take n measures with same time interval above
8- I have no idea what´s the time for a furnace temp change to reach bar end. But, he could even work with tendency, I mean, as bar end temp start to change, it takes actions to revert it.
9- lots of experimentation would have to be done byt our friend
11- do you think it has any chance to work?
Regards
Jose

[John Ferrell]

Remember! I am not an expert here, just a student...
When you add iron mass you will be slowing the measurement process down. Increasing surface area will speed up heat loss. I have been thinking in terms of the sensor being on the head of a 3 inch nail with the point at the heat source. Also, controlling the heat with something like PWM so that corrections do not exhibit latency. Any drafts will add to the challenge. I think a microprocessor can handle it after you develop the algorithim! I may have an idle control on my soldering iron after all!

de John W8CCW

[LGR]

I agree with John. Keep the mass of the sensor to a minimum. That keeps the time constant to a minimum. Regardless of control mode (Dany has a nice PID control code in Libstock), less time constant is better.

[joseLB]

I agree with both you, a needle instead of a bar.... John, let´s you think that you are a student, but by sure the first of the class...

John, I understood that you will try it?

[John Ferrell]

I don't think so! The idea is yours, I am just following your lead.
LGR showed me the way to a really good sensor that I did not know existed.

I am looking in my supplies to see if I have one of the sensors to experiment with.

[joseLB]

well, at this time I don´t need any hi temp measure. I was just trying to help as all you was, I suppose. So, it will stay untested, but joining all ideas, needle, etc., it could be a cheap way and resonable solution, uh?
Jani, any point here??
Regads
Jose

[piort]

hi gang,
i decide to jump in this thread
Another way to measure very high temperature is to use metal expansion constant. We know that white cast iron expand of 0.0000065 inch by Fahrenheit degree. So you take a piece of metal in expansion chamber, in contact with temperate spacer in contact with pressure sensor. you set 0 pressure at 70 degree F , raise the temp to 200 degree F and read the result. Simple math do the rest Another way is if you know the movement of the pressure sensor like pressure measurement is 10mv by 0.00001 inch , simple math do the thing again BTW when you are at 1600 degree F or more (like in metal forging), you can leave with +/- 5 degree precision...

hoping that will alimenting the discussion

[janni]

well, at this time I don´t need any hi temp measure. I was just trying to help as all you was, I suppose. So, it will stay untested, but joining all ideas, needle, etc., it could be a cheap way and resonable solution, uh?
Jani, any point here??

Hi Jose . Sorry to disappoint you, but you're rewarming 6-years old thread . As the OP (zkt) is inactive for 4 years, I don't think he needs a new solution or is able to test your idea. As for the idea itself, anything that may work is worth considering , but your solution would need a practical frame in which it's pros and cons could be considered. It's not exactly a direct method and depends on many factors so it's applicability would strongly depend on practical aspects.

Piort's indirect methods would also be worth considering in specific applications .

Happily enough, in most practical cases direct temperature measurements are possible, and biggest problem is to choose a method best fitted for the task (or least expensive one ).

[joseLB]

Hi Jani, more than disapointed, I´m worried.... 6 years old.... I think that I´m becoming old, I didn´t saw that!!!!

Anyway, at least the ideia is there....

btw, I saw this announce (IR digital termometer, with laser, for about US$30,00). In portuguese, but they tell it can measure/shot up to 10 meters (let´s say, 2 meters), and goes from -50 degrees centigrade up to 380 degrees. This kind of thing work? By sure it´s chinese...
http://produto.mercadolivre.com.br/MLB- ... a-380c-_JM.

[LGR]

IR is good for telling the difference between an object and its surroundings, and it's also good for very hot items. There might be calibration issues with something that's just warm.

OTOH, they're making cheap IR thermometers for measuring body temperature now, so maybe it's a good choice.