Suggestions for a Temperature Sensor

[amspire]

The variation between my five 100 ohm resistors is 0.008 ohms which is a lot smaller then your 1.000495 change.

I wonder if Australia and the US were a bit different back then, and we got resistors tailored for Australia? Australia almost certainly followed the UK standards, so we may have had no change, or perhaps a small one. Looking at my resistors, there seems to be two groups separated by about 0.006 ohms. The newer resistors are the higher ones.

And I know about calibration up the decades. I made a Hamon divider with some 25ppm 0.1% smd resistors I had, and they just change way to much to be useful for accurate decade calibration.  It does not take much power to make a SMD change temperature.

Have attached a photo of the 4 different types of L&N resistors I have.

Richard.

[IanB]

Also, if you haven't changed the oil, do so. I use plain mineral oil from the drug store, but a transformer oil would be best. What happens is the oil becomes acidic and slowly changes the value of the resistor.

I'm puzzled by this, as mineral oil is completely stable and there is no way for it to decompose. For it to become acidic it would have to be in contact with something that contaminates it, such as moisture from the air. Oil would be able to absorb a certain amount of water, and this water would in turn be able to absorb acid gases such as carbon dioxide. Or the oil might perhaps have contained impurities that have decomposed over time...

I don't know what is actually the case, but it would be a good idea to make sure the oil compartment is sealed and airtight in any event. I would guess also that food grade mineral oil is your best chance of a pure uncontaminated product at a reasonable price.

[Conrad Hoffman]

Those look really familiar- have you been messing around on my bench? I also have an old Guildline resistor, 10 ohms I think, but it lost its hermetic seal and thus its oil. I've never had the courage to try and fix it because I don't know exactly how it was put together. The bottom appears potted with epoxy or something. If you were sealing up an oil reservoir, naturally you'd put the cork in the bottom. Not! Last price I saw was something like $5k which seems like a lot for a resistor, but the specs are excellent. If you can find the square grey plastic General Radio Corp. resistors, they're not highly regarded, but my experience with them has been very good.

As for the oil, I've read about the acid problem in various places and I think it's well established. The L&N resistors are not air tight- that may be a requirement for stability, but I'm not sure. It almost seems the opposite. Modern oils are probably better than the old ones, so maybe the problem has gone away. There's also the issue of what they constructed the resistors with- I've seen glyptol, there's fabric or silk and I suspect they couldn't have gotten every trace of flux out of there. I do know if you smell the oil from an old L&N it's pretty awful.

BTW, there was a big Jim Williams article on temperature measurement that's good reading. It should be on-line somewhere, maybe EDN.

CH

[HLA-27b]

You guys are slowly getting into serious metrology. I enjoy reading it.

[IanB]

There's also the issue of what they constructed the resistors with- I've seen glyptol, there's fabric or silk and I suspect they couldn't have gotten every trace of flux out of there. I do know if you smell the oil from an old L&N it's pretty awful.

Yes, pure oil may be stable, but there are plenty of other things that are not. My worst bugbear is the kind of soft, tactile rubber or foam that is often used for cushion grips these days. After a few years it seems to decompose into a messy, molten gloop. Ugh!

[amspire]

I bought some mineral oil today from our local Chemist. In Australia, it seems to called Paraffin Oil rather then Mineral Oil.

I needed to fill the resistors all up anyway after years of seeping oil.  I will post some photos of the internals when I get back home in a few days.

Not sure if I can open the one on the left in the photo.

Here is where I am going with all of this. I want to test to see how some ordinary 50ppm/C metal film resistors behave with temperature variation, and any other interesting resistors that are on hand. I made an "oven" out of two metals strips with a foam sheath that allows me to fill with a lot of resistors between the strips and let them cool slowly. The leads come out each side so i can easily test them all as they cool through the temperatures. The setup allows me to cool over about 3 to 5 hours

I want to see how well different resistors in a batch match for temperature curve, how much error they have after a thermal cycle, and if any resistors are worth an attempt to stabilize with temperature compensation, or is there too much variation in the temp curves.

Richard

[Conrad Hoffman]

Cool! (hot?) The results should be interesting. I only did a casual investigation, but found that the inexpensive metal films I bought from Digikey ($8 for 200 back then) were quite good, but the tempco could be plus or minus and there wasn't much consistency in the bag. It's probably value dependent as well, since the materials are different. In making up my own resistors one of the better materials can be constantan. It's cheap and usually has a very low tempco. The problem is the huge thermal emf with copper, thus its use as a thermocouple. For AC or where the thermal emf cancels out, the stuff might be better than manganin. When I did my resistor tests I just immersed the resistor, clip leads and all, in a beaker of oil, warmed to the temperature desired. I've always been a big fan of instant gratification! The cheap metal films biggest weak point was the permanent change in resistance during soldering. After selecting them, and using a clip on the lead to sink the heat, I always needed to swap out one or two in a decade to bring things back to trim.

CH

[amspire]

The oil bath works great, but I want to be able to leave the resistors on the bandoliers.

I will see how it goes.

I did notice that Constantin seems to have something like 10x the EMF of mangagin, and I gather it really should be welded, not soldered.

I have also been surprised how much difference you can get between different batches of the same wire. Definitely, there is a big difference in stability between making all resistors in a divider out of one batch of wire, and making it from different batches.

The super stable resistors seem to be made of Evanohm and they use multiple resistors with different temp coefficients that all cancel out.

Richard

[IanB]

I'm sure you have this one nailed already, but when winding your own resistors it is important to use a winding pattern that avoids the wire stretching when it cools below the temperature you wound it at. Also to prevent problems if the wire warms up, when it could snag and then stretch when it cools down again.

[amspire]

I have replaced the oil in some of my Leeds and Northrup Rosa/NBS Standard resistors, following Conrad's advice. The construction of the resistor element in different vintages of resistors is very consistent. They stuck to the formula very closely.

I have added some photos.

This design has been around since the early 1900's, and metrology labs were still using them in the 2000's. I am sure they are still in use today. They are not the best standard resistors, but their behavior is extremely well known. It all comes down to confidence, and the labs are slow to trust new resistor designs.



Here is the insides of the newest resistor - the 10K one. The central brass tube is the temperature well so the resistor temperature can be measured. The resistor is wound with a cloth covered Manganin wire on a laquered brass former.  Brass and Manganin have a very similar temperature coefficient of expansion.

For full accuracy, the maximum dissipation is 0.1 watt but if you only need 0.02% accuracy, you can go all the way up to 1 watt.

The covering is important . Manganin is badly affected by any stress on the wire caused by winding, and silk-cotton blend was the most successful material they could find to minimize stress in the wire as they were winding it..

To calibrate the 10K resistor, they added a small 2.5 ohm resistor wound on one of the terminals.



The 100 ohm resistor is pretty similar, but no extra calibration resistor. They just stripped the insulation off each end of the manganan wire calibrated it by adjusting the brazed joint to the copper terminating wires.



If you look at the closeup, you can see the wire was wound as a pair of wires, so it will be a Bifilar winding to minimize inductance.



Richard

[Conrad Hoffman]

Good photos! I've never seen the extra calibration resistor before, but Guildline added a parallel resistor to fine tune my 10 ohm unit.

Just to give an idea of how badly stress can affect precision resistors, I once wound a 100 ohm resistor on a brass form. I kept the winding slightly loose and made many measurements. The thing was dead stable. I then sprayed it with a coating of Krylon Electronic Clear. The performance was terrible! It immediately developed an offset from the original value, plus a large temperature coefficient. Baking and ageing it didn't help. I washed off the Krylon in solvent and the value and temperature stability came right back to normal. IMO, Krylon is not a very rigid coating, probably more flexible than Glyptol. When Julie Research was winding precision resistors they always left the wire completely loose on the bobbin inside the plastic shell. You can saw one apart today and easily reuse the wire- if you can spot weld it.

Did you know that manganin is used as a sensor for the pressure waves from an explosion?

CH

[HLA-27b]

Thanks for the photos Richard, very interesting indeed.
The cloth cover means that these were not annealed after winding. Apperently this has no significant impact even at these levels of accuracy.
If one wanted to redesign this thing to be able to withstand annealing that would increase cost at least tenfold.

Making the thermal well out of brass neatly prevents thermall stress but adds quite a lot thermal mass.  Another curious point is the capacitor formed between this core and the winding and its effects on the resistance.

[robrenz]

Thanks for the photos Richard, very interesting indeed.
The cloth cover means that these were not annealed after winding. Apperently this has no significant impact even at these levels of accuracy.
If one wanted to redesign this thing to be able to withstand annealing that would increase cost at least tenfold.

Making the thermal well out of brass neatly prevents thermall stress but adds quite a lot thermal mass.  Another curious point is the capacitor formed between this core and the winding and its effects on the resistance.

In looking at evanohm material the exact tempco is achieved by exacting heat treatment after cold working. see page 2 of this   hpmetals.com/download/Evanohm-R.pdf  so annealing above 475degF woud be a very bad thing.  I dont know if this applies to Manganin but I bet Conrad does.

[HLA-27b]

In looking at evanohm material the exact tempco is achieved by exacting heat treatment after cold working. see page 2 of this   hpmetals.com/download/Evanohm-R.pdf  so annealing above 475degF woud be a very bad thing.  I dont know if this applies to Manganin but I bet Conrad does.

Well the document says 475 degrees Celsius which is 887 degrees Fahrenheit. Also it does not say it is a bad thing, on the contrary it says it is more or less mandatory.

GENERAL INFORMATION:
 
The alloy is supplied with 90% cold reduction 
which has a positive TCR of about 70 PPM/ºC.
A stabilizing heat treatment (approximately 
475ºC) during manufacture of finished parts
adjusts the TCR to a desired value.  The heat 
treatment virtually eliminates any drift 
tendency of the resistivity.  A heat-treat curve
for each melt is developed at Hamilton and is
made available as a guide in regulating TCR.

[robrenz]

Well the document says 475 degrees Celsius which is 887 degrees Fahrenheit. Also it does not say it is a bad thing, on the contrary it says it is more or less mandatory.

My bad,  The point I was making is the anneal sets the tempco and is unique to each melt of material. so the heat treat curve they mentioned will have to be very precise to achieve a very low tempco. Interesting side note Evanohm is made by Carpenter technology and they are about 25 miles from my home and the Hamilton company the pdf is from is only about 10 miles from my home.

[Conrad Hoffman]

I too can be approximated by a sphere, but I digress... In a past life I did a lot of work with Invar and Super Invar, so I'm a big fan of Carpenter Technology. They also make some of the nicest machining stainless steel around, Project 70 I think. Though I've read all the same stuff about the heat treatment, I've zero experience trying it. Most of my efforts have been to wind very stable standards without heat treating, though I've been known to do a mild bake just by applying a suitable voltage for an hour or so.

BTW, here "paraffin oil" is the stuff used in odor free kerosene lamps when you don't want to use kerosene. It's definitely not what we call mineral oil, and I doubt you'd want it in a resistor. I'd do some searching and confirm what's what. There was a big discussion of this topic on one of the machinists forums a while back, and it's amazing what different substances you can get under a given name, depending on what country you're in.

CH

[robrenz]

link to Carpenter Evanohm data. 
Evanohm R http://cartech.ides.com/datasheet.aspx?I=101&TAB=DV_DS&E=27
Evanohm S http://cartech.ides.com/datasheet.aspx?I=101&TAB=DV_DS&E=26  looks like the  better tcr

[IanB]

BTW, here "paraffin oil" is the stuff used in odor free kerosene lamps when you don't want to use kerosene. It's definitely not what we call mineral oil, and I doubt you'd want it in a resistor. I'd do some searching and confirm what's what. There was a big discussion of this topic on one of the machinists forums a while back, and it's amazing what different substances you can get under a given name, depending on what country you're in.

All of these substances are related, and practically speaking it comes down to the purity and the viscosity of the stuff you are examining.

As a chemical term paraffin refers to the family of straight chain unsaturated hydrocarbons (the alkanes). The small molecules are gases and liquids like propane, butane and (components of) gasoline, the heavier molecules are found in kerosene, and heavier still in oils and waxes.

Paraffin oil, also called mineral oil, is the kind used medicinally as a laxative and in food production machinery as a lubricant. If the stuff you have comes from a chemist or pharmacy, is safe for internal consumption, and is thick and viscous, then it is going to be pure good stuff. When the molecules get even bigger you have paraffin wax, which is solid at room temperature. Paraffin wax is inert and is a very good electrical insulator.

If what you have is thin and pours easily, and is sold as a fuel, then it is not the right stuff, although chemically it will be rather similar. If it is odorless it probably is pure, but it will have the wrong physical and mechanical properties.

[HLA-27b]

What I know aboout heat treatment of metals comes from a book called

Metallurgy of Steel for Bladesmiths & Others
who Heat Treat and Forge Steel

by Em. Prof. Dr. John D. Verhoeven

a highly recommended read and it is available for download.

As far as I understand it, annealing stress relieving etc. are always fairly exacting processes. Interestingly it is not about the temperature of the metal per se, but rather the rate of change of the temperature between two specific temperature levels.

In our case the annealign temperature is around 475 degC. This means that the metal has to be heated to a temperature somewhat above this and soaked at that temperature for certain amount of time. Then the real annealing happens by allowing the metal to cool at a rate not exceeding X degrees per minute until reaching the lower limit say around 350 degC, this is the annealing process proper. The cooling rate below the lower limit is of no consequence.

A possible fly in the ointment is that high temperatures cause metals to oxidize and this is something we definitely want to avoid in a resistor which is meant to be stable. So the annealing of manganin or evanohm has to be carried ideally under vacuum.

P.S: The spherical cow has to do with our tendency to simplify things which we can not comprehend. It is not about surface curvature

[alm]

Using an inert gas like nitrogen or argon would be much easier than vacuum, and has less issues with leakage since it can be positive pressure. These gases may not be readily available in a home lab, but neither is the ability to evacuate almost all of the air.

[amspire]

Just to give an idea of how badly stress can affect precision resistors, I once wound a 100 ohm resistor on a brass form. I kept the winding slightly loose and made many measurements. The thing was dead stable. I then sprayed it with a coating of Krylon Electronic Clear. The performance was terrible! It immediately developed an offset from the original value, plus a large temperature coefficient. Baking and ageing it didn't help.

The main windings on the L&N resistors are definitely not varnished. They are actually wound fairly firmly,  but they can probably do that with the silk/cotten cover as it probably has enough give to prevent any pressure spots in the wire.

The cloth cover means that these were not annealed after winding. Apperently this has no significant impact even at these levels of accuracy.

The 1 ohm Thomas resistors have manganin wire that is annealed at about 500 degC, but they can do that as the windings are self supporting - they do not need a former. It is hard to do with any multilayer wound resistor as insulation cannot survive the temperature. Also at that temperature, the wire oxidizes, and it then has to be pickled to remove the oxide. Much easier to do with a free standing Thomas resistor then a resistor wound on a former.

Edit: Or as Hal-42b suggests, heat in a vacuum or an inert gas.
 
I have read that the standard way to stabilize a new Manganin wound resistor is to age it at 120 degC for 24 hours, so I would guess that L&N did that with these resistors. Not as good as annealing after winding though.

Making the thermal well out of brass neatly prevents thermal stress but adds quite a lot thermal mass.  Another curious point is the capacitor formed between this core and the winding and its effects on the resistance.

These resistors are usually need about 2 hours in a temp controlled oil bath to stabilize before measuring.  The newer ultra stable standard resistors have built in temperature sensing often by a second slightly less stable standard resistor thermally connected to the main resistor. By measuring the difference in resistance between the two resistors, you can use a chart to compensate for temperature errors. It means the newer resistors can often be used immediately without waiting for anything to stabilize.

All the information about Evanohm is great. I was wondering how the TCR was controlled. The IET Labs standard resistors are made using multiple Evanohm resistors that have TCR's that cancel out. So it looks like you can give different batches of wire different heat treatments to give different TCR curves, and then you can carefully mix them together to get the most stable blend.

Richard.

[amspire]

Reading the Evanohm data sheet, it looks like it does not have one of the chronic problems of Manganin. With Manganin, if you heat up the wire to 100 DegC +, the resistance will almost certainly change significantly and permanently.

Evanohm does not seem to change significantly at temperatures up to 200 DegC.

Very impressive!

[amspire]

It sounds like "Mineral oil" can have an extremely variable composition, so it is hard to know if it is corrosive or not.

The oil recommended by Tinsley Precision Instruments in the UK for their standard resistors is:

Castrol Whitemor WOM14
http://www.tds.castrol.com.au/pdf/10021_whitemor_wom_14_2010_10.pdf

This is a highly refined food-grade mineral oil.

I have used "Paraffin Oil B.P." from the chemist in some of my resistors, I will try it and see how it goes. The "B.P." means it passes the British Pharmacopoeia 2009 specification which includes testing for presence of any acidity or alkalinity, Sulphur compounds and Polycyclic Aromatic Hydrocarbons. I suspect it will be OK. The neutrality test is to boil a 50/50 mixture of oil and water. Then add a drop of phenolphthalein TS and it must not turn a shade of pink. Then add a drop of very diluted Sodium Hydroxide solution and it should turn a slight pink. I have the exact procedure if anyone needs to test their oil.

The question I have is at what temperature does it solidify? I suspect it is something like -9 degrees, but I have put some of the oil in the freezer to see if it stays liquid. So far, it has thickned, but no solids.

Richard

[IanB]

If mineral oil is food grade you can assume it will not be corrosive 

The basic constituent of mineral oil, long chain paraffinic hydrocarbons, is pretty much inert. Any problems will come from additives or impurities. For instance, lubricating or industrial oils may contain various additives to improve their properties for their intended use.

Acidity, alkalinity, sulphur compounds or aromatic hydrocarbons are all things that should not be in pure paraffin oil.

It probably won't solidify in a domestic freezer. The freezing point will be somewhere below -22 C most likely. But does it matter to you when it solidifies? Will this make a difference to any use you have for it? (I suspect not.)

[amspire]

It probably won't solidify in a domestic freezer. The freezing point will be somewhere below -22 C most likely. But does it matter to you when it solidifies? Will this make a difference to any use you have for it? (I suspect not.)

I just wanted to make sure it was still fluid enough to circulate heat at all practical temperatures for the resistor, and it looks like it does. I am sure it wouldn't cause any damage if it did solidify.