How EXACTLY do immesion heater tank thermostats work ?

[RogerJ]

My first post after extensive Googling. I'm qualified as an elctronics engineer (retired) Posting this question here in hopes that an enquiring electrcian will know the answer. The last time I changed a tank stat, I disasembled the old one and kept the 18" brass tube and central steel thin rod.. ( I do things like that a lot) I can see how it works but wonder if anyone here knows if that central thin rod is plain steel or Invar..
I have a use for a thin Invar rod in a DIY clock.. Thanks..

 

[Lister1987]

My first post after extensive Googling. I'm qualified as an elctronics engineer (retired) Posting this question here in hopes that an enquiring electrcian will know the answer. The last time I changed a tank stat, I disasembled the old one and kept the 18" brass tube and central steel thin rod.. ( I do things like that a lot) I can see how it works but wonder if anyone here knows if that central thin rod is plain steel or Invar..
I have a use for a thin Invar rod in a DIY clock.. Thanks..

Best bet is to ask a manufacturer?

 

[Julie.]

Very unlikely to be invar as invar has a very low expansion rate to keep the change in length of the pendulum at a minimum, in the case of a thermostat, you want the expansion to be quite high such that the increase in length is sufficient to operate the contacts.

It's the same whether the expansion comes from the expansion of the whole rod, or from a sensor at the remote end transmitted by the rod, you wouldn't need to maintain a low rod length expansion rate.

 

[Avo Mk8]

Very unlikely to be invar as invar has a very low expansion rate to keep the change in length of the pendulum at a minimum, in the case of a thermostat, you want the expansion to be quite high such that the increase in length is sufficient to operate the contacts.

It's the same whether the expansion comes from the expansion of the whole rod, or from a sensor at the remote end transmitted by the rod, you wouldn't need to maintain a low rod length expansion rate.

Hi Julie
I did a little research on this yesterday for my own interest. and found a patent application that specified Invar for the central rod.
The principle of operation, at least for that stat, was that the outer brass tube lengthens with temperature increase, and since you want to get that movement back to operate the switch, you don't want to cancel it out with a material that also expands.
So with the brass outer and Invar centre rod, you get maximum 'gain'. if you use steel, you would get a lot less movement, and hence less 'sensitivity'.

 

[Julie.]

Hi Julie
I did a little research on this yesterday for my own interest. and found a patent application that specified Invar for the central rod.
The principle of operation, at least for that stat, was that the outer brass tube lengthens with temperature increase, and since you want to get that movement back to operate the switch, you don't want to cancel it out with a material that also expands.
So with the brass outer and Invar centre rod, you get maximum 'gain'. if you use steel, you would get a lot less movement, and hence less 'sensitivity'.

Ah OK.

I assumed it would be the central rod that would expand, but yes it makes perfect sense.

 

[brianmoooore]

It's the outer tube that provides the expansion. The inner rod is just to couple that expansion back to the switch mechanism.
Any expansion of the inner rod is undesirable, but as long as it's less than the outer tube, the 'stat will still work, but will be less sensitive the closer the expansion rates of the two materials are.
You can make a 'better' 'stat with an invar inner rod, but an 'adequate ' one with steel, and since steel is cheaper than invar, I expect that it's steel that is used.

 

[RogerJ]

It's the outer tube that provides the expansion. The inner rod is just to couple that expansion back to the switch mechanism.
Any expansion of the inner rod is undesirable, but as long as it's less than the outer tube, the 'stat will still work, but will be less sensitive the closer the expansion rates of the two materials are.
You can make a 'better' 'stat with an invar inner rod, but an 'adequate ' one with steel, and since steel is cheaper than invar, I expect that it's steel that is used.

Thanks Brian and Julie...I think you may now have guessed I'd got that far and was hoping to find if the central rod was the best answer (Invar) - or the cheaper, less sensitive option, steel.. I've no way here to tell the difference myself so thought there was a chance someone would know. I suppose it's possible that it may even differ between manufacturers.
So unless there is someone out there who knows EXACTLY we'll leave it there.. Roger

 

[RogerJ]

As it's New Years Eve I hope you won't mind me going off topic and say why I hoped it might be Invar ;-)
I've built an accurate electricaly powered pendulum clock, with a 1 metre (1 second) Invar pendulum rod. The bottom supported 15lb steel bob however expands upwards with heat causing the clock to go slightly faster with increase in temp. If I could transmit that slight rise of the bob top to a suitable point up the main rod... with a thin Invar rod...and add a small weight to the top of it, I could cancel the effect out.. happy New Year all..Roger

 

[brianmoooore]

Couldn't you get the same effect with a small additional weight attached to a brass rod, under the main bob? That weight moves down to slow the pendulum to counteract the expansion of the main bob speeding it up.
Alternatively, couldn't you attach the pendulum arm to the bob near its centre of mass, so that part f it expands upwards, and part down?

 

[RogerJ]

Hi Brian.. The second option is the solution used in high class work.. It involves drilling the bob with a shelf half way up.. and using an Invar tube back down to the rating nut.. Beyond my workshop capabilities.. But a small weight on a brass weight below is possible..food for thought there ! Thanks..

 

[Avo Mk8]

Hi Brian.. The second option is the solution used in high class work.. It involves drilling the bob with a shelf half way up.. and using an Invar tube back down to the rating nut.. Beyond my workshop capabilities.. But a small weight on a brass weight below is possible..food for thought there ! Thanks..

And then the next improvement is to deal with changes in barometric pressure! Happy New Year

 

[RogerJ]

And then the next improvement is to deal with changes in barometric pressure! Happy New Year

Yep..the final frontier ! I have a net friend in Canada building a similar clock but with micrprocessor (Arduino) control of the pendulum swing with plans to deal with baro. pressure. Sadly I can't program such things ! My clock is here: A DIY, NON Hipp Toggle master clock. - http://www.rogerj.co.uk/clock2.htm
@ AVO Mk8. . My first job intoduced me to the AVO 8 (and the Tektronics 535A) but I couldn't afford one in the '60's. Twenty years ago they were cheap as chips and I now have 12.. one of every Mk I think.. but mostly use a ten quid digital now..

 

[Avo Mk8]

@ AVO Mk8. . My first job intoduced me to the AVO 8 (and the Tektronics 535A) but I couldn't afford one in the '60's. Twenty years ago they were cheap as chips and I now have 12.. one of every Mk I think.. but mostly use a ten quid digital now..

If you've got one of the very first wood cased Avo's I'm really jealous!

My father 'bribed' me to work at my A levels by saying if I got the grades I needed he would buy me an Avo 8 (my choice of reward - what a geek). It's been in use for 55 years until my wife bought me an Avo 410 (isn't a 'real' Avo!) as a birthday present last year. Wow, I can measure capacitance!

Well done with the clock

 

[RogerJ]

If you've got one of the very first wood cased Avo's I'm really jealous!

Nope... Nothing before the black plastic Mk1.. I got as far forward as getting a DA116.. Also not a "real" AVO !

And finally.. @brianmoore.. Quote "Couldn't you get the same effect with a small additional weight attached to a brass rod, under the main bob? That weight moves down to slow the pendulum to counteract the expansion of the main bob speeding it up."

Thinking about it a bit more.. that option would be cumbersome - clock geeks will know that the most sensitive place to move a weight on a pendulum rod it half way up.. Here a small weight would suffice.. BUT in this instance a THIN plain steel rod of the about 18" would move a small weight in sympathy with the bob.. However expansion of that rod would increase the "gain" - the opposite of its use in the stat. But it would have a completely different thermal hysterisis from the bob and may not be such a purist solution..

 

[DPG]

Interesting stuff. Will have a proper look at your web page later.

 

[PEG]

I trust all of you are in good shape,after celebrating the new year

I include the OP.....who raised a glass,4 seconds after the rest of us

 

[RogerJ]

LOL...Thank you PEG and Happy New Year to all on the forum.. In the event last night the clock ws only about 0.1 seconds slow last eve (was reset last Thursday) and I got the the time from Jools Holland anyway..

 

[brianmoooore]

I got the the time from Jools Holland anyway..

Can you anymore, in these days of digital everything?
Back when TV and radio were all analogue, if you listened to Big Ben on any device, the chimes were in perfect sync, and, I assume, accurate.
Nowadays, there's a noticeable difference between devices of the same type, such as TVs in adjoining rooms, radio different again, and if you start adding streamed sources, can be a minute or so behind the others.

 

[RogerJ]

Good point..In the good old analogue days transmission route delay was neglible. Big Ben is maintained by its keepers and no doubt on new years eve is tweaked to be as close as possible to GMT at midnight.. The pips are of course electronically generated by a source fed from a Caesium atomic clock..
I believe the delay via freeview is about 2 seconds.. Cable, Satellite and internet delivery may be a bit longer (and all different) Delay on DAB radio is significant and depends a lot on each radio's decoder (speed ,quality).. On NYE Jools is good enough for me and I reckon I was in sync with all indoor celebrators

 

[brianmoooore]

I regularly watch F1, using feeds from various places for commentary, pictures and data screens. There can be up to 90 seconds difference between identical picture feeds, and anything up to two and a half minutes difference for the other stuff.
Radio Five Live commentator usually says exactly when pit exit light goes green, specifically for the purpose of synchronising their commentary with whatever picture feed you're using.

 

[pc1966]

Some pendulum clocks were extraordinarily accurate, such as these:

Shortt–Synchronome clock - Wikipedia

en.wikipedia.org

 

[pc1966]

If you have the rod already, you could try measuring its electrical resistance and see if it falls in to the range for steel, or is significantly higher as alloys with chrome/nickel tend to be (such as stainless).

If you have a 5A power supply, or a car battery and a headlamp bulb (along with safe means of holding it as it will get HOT!) you can use that to put a decent current along the rod. Then measure the volt drop from two points on the rod (not on the power supply leads or clips) and from a measurement of the cross-sectional area and length you ought to find the material's resistivity. You will need something that can measure a few mV probably, but a cheap-ish multimeter ought to do.

Carbon steel is about 1.43E-7 Ohm.meter, while stainless about 6.9E-7 Ohm.meter, Other cases can be found here:

Electrical resistivity and conductivity - Wikipedia

en.wikipedia.org

EDIT: Just to add some background info about the Kelvin (4 wire) resistance measurement method:

Kelvin (4-wire) Resistance Measurement | DC Metering Circuits | Electronics Textbook

Read about Kelvin (4-wire) Resistance Measurement (DC Metering Circuits) in our free Electronics Textbook

www.allaboutcircuits.com

 

[RogerJ]

Some pendulum clocks were extraordinarily accurate, such as these:

Shortt–Synchronome clock - Wikipedia

en.wikipedia.org

Hi pc1966. I'll address your following post shortly.
The clock stuff: All practiising electrician will know that their craft basically begins with Michael Faraday (Michael Faraday - Engineering and Technology History Wiki - https://ethw.org/Michael_Faraday) - nearly 200 years ago.
Soon after (1841) Alexander Bain patented the first electric clock (Distant Writing - Bain - https://distantwriting.co.uk/bain.html)

A free swinginging pendulum devoid of ANY interference WHATSOEVER will keep good time. In my clock, with the advantage of 21st century electrics, were to be encased in a chamber with a fixed low vacuum it would probably equal the performance of the Shortt/Synchronome. "Interference" in mechanical pendulum clocks, may come from poking the pendulum on every swing and extracting energy to move the hands.

In 1842 Mathtias Hipp came along with a better version (A DIY Free Pendulum, Hipp Toggle master clock. - http://www.rogerj.co.uk/clock.htm)
In 1895 Frank Hope-Jones formed the Synchronome company (Synchronome Clocks history - https://www.synchronomeclocks.com/pages/about-us) and his master clocks, with one second pendulums were in use well into the 1960's - when they finally succumbed to the Quartz crystal and other esoteric resonating sources. Along the way..in 1921, collaboration between Hope-Jones and William Hamilton Shortt produced the time reference you linked too.. Now we all know !!

 

[RogerJ]

If you have the rod already, you could try measuring its electrical resistance and see if it falls in to the range for steel, or is significantly higher as alloys with chrome/nickel tend to be (such as stainless).

If you have a 5A power supply, or a car battery and a headlamp bulb (along with safe means of holding it as it will get HOT!) you can use that to put a decent current along the rod. Then measure the volt drop from two points on the rod (not on the power supply leads or clips) and from a measurement of the cross-sectional area and length you ought to find the material's resistivity. You will need something that can measure a few mV probably, but a cheap-ish multimeter ought to do.

Carbon steel is about 1.43E-7 Ohm.meter, while stainless about 6.9E-7 Ohm.meter, Other cases can be found here:

Electrical resistivity and conductivity - Wikipedia

en.wikipedia.org

EDIT: Just to add some background info about the Kelvin (4 wire) resistance measurement method:

Kelvin (4-wire) Resistance Measurement | DC Metering Circuits | Electronics Textbook

Read about Kelvin (4-wire) Resistance Measurement (DC Metering Circuits) in our free Electronics Textbook

www.allaboutcircuits.com

I have the rod and may attempt to try it. It has also occured to me that if I supported it off the side of a stout board, fixed one end and with a dial gauge at the other..... I could pour boiling water over the rod and see if it expands...
Thank you..I'll follow up the links.. Roger

 

[Avo Mk8]

Roger
Sorry to be a spoilsport - coming back to your original question (Invar or steel), it occurs to me, and no doubt has to you too, that a simple experiment might be worth trying (but only if the length of your thermostat rod will fit in your oven!)
Cut a wire cotahanger to be as close as practicable to the length of the rod, and bung them in the oven at as high a temp as you like! Take them out, stand the pair next to each other and see if they're the same length!!
Or maybe there's a flaw in this scheme

 

[RogerJ]

Roger
Sorry to be a spoilsport - coming back to your original question (Invar or steel), it occurs to me, and no doubt has to you too, that a simple experiment might be worth trying (but only if the length of your thermostat rod will fit in your oven!)
Cut a wire cotahanger to be as close as practicable to the length of the rod, and bung them in the oven at as high a temp as you like! Take them out, stand the pair next to each other and see if they're the same length!!
Or maybe there's a flaw in this scheme

Done ! I left the bitter end of the brass tube on (as a socket) and cut a length of steel fence wire to fit under the conical under face of the "nut" on the other end of the stat rod. the other end was trapped under the nut and screwed down so they lay alogside each other. With other half safely watching Paddington Bear I set the oven to 180 and after 10 minutes put the combination in. The fence wire bowed away from the specimen by 1/4"..soon cooled down to as was. A strong indication therefore that the stat rod is Invar
Simplest ideas !!!!

 

[pc1966]

This is also an interesting run through the history of "time keeping" but phrased as oscillator stability:

http://www.leapsecond.com/ten/clock-powers-of-ten-tvb.pdf

 

[RogerJ]

This is also an interesting run through the history of "time keeping" but phrased as oscillator stability:

http://www.leapsecond.com/ten/clock-powers-of-ten-tvb.pdf

Interesting indeed.. I follow the gist of the sequence OK but I would need to have attended the lecture to understand a lot of the info the slides carried..

 

[johntheo]

Is it possibe that the thermostat is actually a pressure switch with some sort of thermal fluid in the sensing rod, cut a bit off the end??.

 

[johntheo]

No, not a pressure switch.

 

[RogerJ]

John Ward, the presenter) didn't get it EXACTLY right either He was corrected more than once in the comments....

 

[johntheo]

Where I find this stat interesting is where a single stat is used in a top mounted immersion where normally sink & bath lives are switched manually depending on the volume of HW required, one terminal on each heating element are permanently linked to each other and the thermostat switches the common neutral.
the thermostat rod length is matched to the longest (bath) heating element, on a standard HW cylinder with 11" and 23" elements the stat length will be 22/23".

I wonder what effect just fitting a 11" stat would have. Sink selected no problem. Bath selected and assuming a cold cylinder then the temperature will rise uniformly until 60C is reached and the stat switches Off, if you then start using HW the cold water will displace the HW until the contents of the cylinder are almost exhausted, temp falls to (by say 5C, hysteresis) to 55C, stat switches ON, cylinder reheats. The only? advantage I see with using a 22" stat is that it may switch on once a smaller vol of water has been used, some of the stat will be at 60C and some at say 15C (mains temp) so perhaps this is enough to switch on the stat.

 

[johntheo]

Actually, think I got a bit mixed up, just think the stat pocket is really only 11" or so, no mystery.

 

[johntheo]

Just to clear up the above point, all top mounted twin immersions with a common stat means that the stat pocket & stat will never be longer than the short element, both elements will work fine, and the stat will cut out at its setpoint temperature. If the stat was the same length as the longer element, and if this (bath) element is selected it will cut out at its setpoint temperature and cut in and start reheating the cylinder once the incoming cold water is in contact with a fairly small length of the stat.
If the shorter (sink) element is selected then the stat just might cut out (if at all) but at a very high temperature as "1/2" the stat will be ~ at mains water temperature and the other "1/2" might be at 100C if a vented cylinder.

I did a few simple tests on a new 11" stat that I found lying around, the stat, when fully immersed in water switched out at 62C and switched in at 50C, with 4.5" immersed, it cut out at 73C and switched in at 67C. Stat nominally set to 60C for both tests.

 

[RogerJ]

All very interesting but..... The point of the thread was to discover if anyone knew for certain that the thin steel rod inside the stat's brass tube is Invar.. I think we're 90% certain it is.. That's all..

 

[EricMark]

Must admit I don't know exactly how it all works, the willis immersion heater is clearly different from direct, and going in from side of tank is also different from going in from top, but it seems unlikely that all thermostats work in the same way.