Thermo-Electric Generators.

Let us smooth your persuit of ccda, a mcts 70-536 exam or a hard-hitting oracle certification.

Updated: 13 June 2008
Pouillet, Ruhmkorff added
The Pouillet thermopile: c1840 New
The Ruhmkorff thermopile: c1860 New
The Markus thermopile: 1864
The Becquerel thermopile: 1864
The Clamond thermopile: 1874
The Noe thermopile: 18??
The Hauck thermopile: 18??
Dr Stone on thermopiles: 1875
The Gülcher thermopile: 1898
The English Mechanic thermopile: 1898
The Thermattaix: 1925
The Cardiff Gas Light & Coke Co: 1930s
The Russian Lamp: 1959 Updated
Back to Home PageBack to The Museum

Thermo-electric generators convert heat directly into electricity, using the voltage generated at the junction of two different metals. The history begins in 1821 when Thomas Johann Seebeck found that an electrical current would flow in a circuit made from two dissimilar metals, with the junctions at different temperatures. This is called the Seebeck effect. Apart from power generation, it is the basis for the thermocouple, a widely used method of temperature measurement.
The voltage produced is proportional to the temperature difference between the two junctions. The proportionality constant a is called the Seebeck coefficient.

A series-connected array of thermocouples was known as a "thermopile", by analogy with the Voltaic pile, a chemical battery with the elements stacked on top of each other. The Danish physicist Oersted and the French physicist Fourier invented the first thermo-electric pile in about 1823, using pairs of small antimony and bismuth bars welded in series. The thermopile was developed by Leopoldo Nobili (1784-1835)and Macedonio Melloni (1798-1854). It was initially used for measurements of temperature and infra-red radiation, but was also rapidly put to use as a stable supply of electricity for other physics experimentation.

George Simon Ohm was probably the most famous user. In 1825 he was working on the relationship between current and voltage by connecting wires of differing resistance across a voltaic pile- pretty near short-circuiting it. After an initial surge of current rapid polarization of the pile caused the voltage to decrease steadily, greatly complicating the measurements. Ohm took a colleague's advice and replaced the voltaic pile with a thermopile, and the results were much better. Note that this is only four years after the discovery of the Seebeck effect, so the idea of the thermopile must have been quickly developed. Unfortunately I have not so far been able to identify the first use of the effect for power generation rather than measurement.

As an aside, Ohm's law met with a very cool reception in his own country; one account soberly states: "Unfortunately, Ohm's law was met with resistance." The Prussian minister of education pronounced that "a professor who preached such heresies was unworthy to teach science." This is the sort of thing that happens when politicians try to involve themselves in science, and in that respect we have progressed little since then.

Here are displayed some early thermo-electric generators or "thermopiles". I have tried to put them in chronological order but not all have a definite date, so this is rather iffy.
The maximum power is obtained from a thermopile when its load resistance is equal to its internal resistance, as with all electrical sources. Since the internal resistance of a chain of thermocouples is very low, this means that it can supply big currents but only low voltages, unless a large number are wired in series.

Left: Thermopile by Pouillet: circa 1840.

This, I think, is the earliest thermopile I have found so far. Unfortunately I have no details on it, and its operation is obscure. It is not clear where the heat is applied; perhaps one brass tank held hot water and the other cold? If so, that would be a much less effective source of heat than a gas flame.

In each tank, one of the L-shaped pipes appears to go into a glass vessel, for reasons unknown.

Claude Pouillet (1790-1868) was a pioneer in the detection of infra-red radiation. He used a "pyroheliometer"- essentially a water calorimeter- to measure the intensity of solar radiation. The apparatus shown above is NOT the pyroheliometer; however it may be some sort of measuring instrument rather than a power source as such.

Example in CNAM, the Conservatoire National des Arts et Metiers, in Paris. Author's photograph.

Left: Thermopile by Ruhmkorff: circa 1860.

The gas burners are inside the black body of the device; the spigot for the gas supply pipe is at lower left. The brass tanks hold the cooling water for the cold junctions.

An interesting feature is the sliding contact at the top, which allows the output voltage to be altered by connecting a variable number of junctions into the circuit. The output terminals are at top right.

Heinrich Daniel Ruhmkorff, electrical researcher and instrument maker, is best known for the remarkable improvements he made in the induction coil. However, it appears he was also in the thermopile business. Ruhmkorff was born on 15 January 1803, in Hanover, Germany, and died 20 December 1877, in Paris.

Example in CNAM, the Conservatoire National des Arts et Metiers, in Paris. Author's photograph.

Left: Markus's thermopile: 1864.

The EMF of a single couple was quoted as "One-twentieth of a Daniel cell" which makes it about 55 milliVolts. The negative metal was a 10:6:6 alloy of copper, zinc and nickel, similar to German silver, and the positive metal was a 12:5:1 alloy of antimony, zinc and bismuth. The iron bar a-b was heated and the lower ends cooled by immersion in water. A defect of this design was a rapid increase in internal resistance as the two alloys oxidised at their point of contact.

Markus' thermopile won a prize in 1864/5 from the Vienna Society for the Promotion of Science.

From "Electricity in The Service of Man" published in its 3rd edition in 1896; the thermopile section appears to have been written much earlier, and certainly before 1888. It was originally published in Germany and was written by Dr A R Von Urbanitsky.

Left: Becquerel's thermopile.

This was invented by M. Edmond Becquerel (1820-1891), at a date unknown. The junctions were composed of copper sulphide for one metal, and German silver for the other.
It appears that D is a trough of cooling water for the cold junctions, supplied at B and exiting at C. There appears to be another trough on the other side of the central burner E. Gas for the burner is supplied via pipe A.

Edmond Becquerel was the father of physicist Henri Becquerel, who discovered radioactivity

From "Electricity and Magnetism", 1891.

Left: The Clamond Thermopile: .

This pile, developed in association with Mure, used a zinc-antimony alloy for one metal and iron as the other. It was gas-fired, and could liberate 0.7 oz of copper per hour by electrolysis while consuming 6 cubic feet of gas in the same period. The output current was quoted in this outlandish fashion because electroplating was the main application of these devices; possibly practical ammeters did not yet exist.
The diagonal connections that join each ring of couples in series can be seen between the two vertical strips.

The gas pipe can be seen coming in from bottom right. The little coffee-pot thing in the line is actually a gas pressure regulator.

From "Electricity in The Service of Man"

Left: The Clamond Thermopile: plan view.

The solid sectors A were made of the alloy, while the cooling fins F were made of sheet iron to act as cooling fins for the cold junctions.

From "Electricity in The Service of Man", a much longer book than "Electricity in The Service of Chameleons"

Left: The Clamond Thermopile: reality.

Note gas feed with tap running into the centre of the pile.

This example is in the History Museum of the University of Pavia in Lombardy, Italy.

Left: The Clamond Thermopile: section.

Showing the multiple annular burners in the centre of the pile. Gas enters through tube T.

According to the French journal La Nature for 1874, one of these piles was in use at the printing works of the Banque de France, presumably for electroplating.

Picture from La Nature 1874.

Left: The Improved Clamond Thermopile: 1879.

The EMF of this pile was no less than 109 Volts, with an internal resistance of 15.5 Ohms. The maximum power output was therefore 192 Watts, at 54 Volts and 3.5 Amps.

This pile was fired by coke. The hot junctions were at C, while the cold junctions D were cooled by sheet iron as in the original design above. What purpose was served by the tortuous path T-O-P taken by the hot gases is unclear, because there seem to have been no hot junctions in the inner sections.
This beast was 98 inches high and 39 inches in diameter.

It was a serious piece of machinery, quite capable of delivering a lethal voltage.

From "Electricity in The Service of Man"

Left: The Noe Thermopile.

The hot junctions are the pointed things directed inwards to the central burner. The cold junctions are cooled by radiation and convection from the vertical strips on the outside.

The inventor, Fr. Noe, came from Vienna. The output EMF of this pile was about 2 Volts, with an internal resistance of 0.2 Ohm. This was for a pile with 128 couples.

From "Electricity in The Service of Man".

Left: One thermocouple from the Noe Thermopile.

The hot junction is a copper pin in a brass case, surrounded by "an alloy" which is presumably the other half of the junction.

The connecting wires visible here on each side were of "German silver". German silver (better known nowadays as nickel silver) is the generic name for a range of bright silver-grey metal alloys, composed of copper, nickel and zinc; it contains no real silver.
These wires were essential to join the thermocouples together, but reduced its efficiency as they conducted heat away from the hot junctions to the cold ones. The problem is elegantly solved in modern semiconductor versions by using alternate P and N type materials that do not require these connections.

From "Electricity in The Service of Man".
Left: The Noe Thermopile in reality.

This high-performance version is surrounded by little cylindrical fins for cooling the cold junctions, permitting a greater output.

This example is in the History Museum of the University of Pavia in Lombardy, Italy.

Left: Hauck's thermopile.

The EMF of a single couple was quoted as "0.1 of a Daniel cell" which makes it about 110 milliVolts; this seems rather high to me. The current capacity using 30 couples was "capable of making a platinum wire 1.2 inches long red-hot" which is not a very useful sort of spec, since we have no idea how thick the wire was.
The Hauck pile was fired by gas, using something looking very much like a Bunsen burner. The cold junctions were water-cooled by a series of little cylindrical tanks, and there was an obscure little glass device in the middle; possibly to show the rate of gas flow?

These devices appear to have been produced commercially in different sizes, with two or three placed on a common frame. They were used for science education and electroplating. To put a time marker on this, it was 1843 when Moses Poole took out a patent for the use of thermoelectricity instead of batteries for electro-deposition purposes. This was long before practical dynamos and alternators.

In the days when chemical cells needed a lot of attention, something that provided power at the strike of a match must have had its attractions.

From "Electricity in The Service of Man".

Left: Article in Nature: Nov 18,1875.

Doctor Stone reads an article on thermopiles.

This gives some interesting practical details on the problems of brittle thermocouple materials and the difficulty of avoiding oxidation when iron was used as one half of the couple, as it was in the Clamond pile. There is also the interesting suggestion that petroleum should be vaporised at the cool junctions, reducing their temperature, and the resulting vapour burnt at the hot junctions.

Attempts to find out more about Dr Stone have so far failed.

This article comes from the English journal Nature, not to be confused with the French journal of the same name.

Left: Gülcher's thermopile: c 1898.

It looks as if it was gas-fired, with the gas going in through the spigot on the right, but unfortunately that is all I know about it at present.

This example is in the History Museum of the University of Pavia in Lombardy, Italy.

Left: Commercial thermopile: 1898.

A handy thermopile with wall-mounting bracket. It is gas-fired, with the gas going in through the central spigot. The output terminals are bottom left. Manufacturer unknown, but if it really could be supplied by "any respectable electrician" it must have seen some commercial success.

Cooling looks like it might be any issue; presumably it relied on convection and radiation from the cylindrical outer surface, as there are no signs of water cooling arrangements. I would have thought that would have reduced its effciency markedly. There are no visible fins to improve cooling.

If the biggest model gave 2.5A at 8.5V, that's a healthy output of 21 Watts.

Bottone was a regular contributor to discussions in the English Mechanic at the time.

From English Mechanic 9 Sept 1898, p98

Coke-firing is clearly not an attractive option unless you had a big thermopile like the Improved Clamond above. Coal gas was far superior for table-top models. But when did gas supply to buildings start? Here are a few historical nuggets that show that gas could be laid on rather earlier than you might think. But for a year or two, the Duke of Wellington could have written his despatch reporting his victory over Napoleon at Waterloo by gaslight.

By 1819, 288 miles of gas pipes had been laid in London to supply 51,000 burners.

The first commercial town gas supply in the USA began at Baltimore, Maryland in 1816, lighting residences, streets, and businesses.

By 1850, all public lighting in France was by gas.

I have so far no been able to discover when gas was introduced in the German states; can anyone help? Anyway, I think I have shown that a gas supply was in fact ready and waiting for the thermopile.


Left: Yamamoto patent: 1905.

This thermopile was patented in Japan in 1905 by one Kinzo Yamamoto. Few other details are known; much information was destroyed in the Tokyo Earthquake of 1923.

The P-type material is made of bismuth, antimony and zinc in the proportions: Bi:Sb:Zn=12.0:48.0:36.8. In the figure, D is a P-type "Bullet" and E is a Nickel electrical connection. (Probably that should be nickel-silver: see above)F is the pin to collect heat flow from the flame. A is an electrical and thermal metal connection. B is a cooling fin.

This design has an unmistakable resemblance to the Noe thermopile above; in fact it appears to be a very faithful copy. It was presumably intended for powering radios, but this is pure guesswork on my part.

It appears that Great Britain was rather slow in electrification compared with other European countries. Light could be provided by gas, and heating by coal, but electricity was needed to run radios and a gas-fired thermopile was one way to get it. Alternatively, you took your lead-acid filament accumulator into town to get it charged for you, which was somewhat less than convenient.


Left: The Thermattaix: circa 1925.

Not a name that exactly trips off the tongue. The voltmeter on the front registers from 0 to 10 Volts; a suitable voltage range for charging accumulators running 6.3 Volt valve heaters. The black knob below the meter obviously controlled something- presumably the gas supply.
It appears this device was designed to charge lead-acid accumulators rather than power the radio directly. This may have been because output voltage fluctuations would have had little effect on accumulators, but would have been very bad for the filaments of valve heaters.

This example is in the Science Museum in London.

The magazine Amateur Wireless, in April 1929 carried an advert for the Thermattaix, apparently claiming that it could work your wireless set by gas, petrol, electricity or steam. Electricity? It goes on to claim that amongst their customers were gas companies, the Italian airforce, architects of note and big game expeditions in Africa and India.


The gas-fired machine below, which seems to have no name, but was sold by the The Cardiff Gas Light & Coke Company, was brought to my attention by John Howell, who says that his father sold a number of these when working in South Wales during the 1930's; that's what triggered this page. I must admit that I had never heard of such a thing in Britain before- they must have been fairly rare. I would have thought that by 1930 the provision of mains electricity would have been well advanced. However, apparently not.

Whether The Cardiff Gas Light & Coke Co made this machine themselves, or bought it in, is currently unknown.

Left: The gas-fired thermo-electric generator: 1930s.

Well, it was certainly the invention of a generation, but not of the generation that advertised this machine, as you will have seen from the thermopiles above.

It is believed it contained thermopiles (ie series arrays of thermocouples) that produced 2 Volts @ 0.5 amps for valve filaments/heaters and 120 Volts @ 10mA for the HT.

Thermocouples do not generate much voltage, but since they are simply junctions between two kinds of wire, connecting many in series is feasible. One of the most useful combinations is Ni/NiCr, ie nickel/nickel-chromium. This has a thermovoltage of about 4 mV/100K and a usable temperature range up to some 1000 K. This is very likely the type of couple used in this generator; it implies that 40 mV is about the most you can get from each thermocouple, so 50 in series would have been needed for the 2 V filament supply and 3000 in series for the 120 V HT. This sounds possible, though probably rather protracted to assemble and maybe heavy on labour costs. It would be interesting to know what the retail price was.

Picture kindly provided by John Howell.

Left: Advertising blurb for the thermo-electric generator. Probably printed on the other side of the page above.

The automatic control feature is intriguing. Given that any radio of the time would have had a Class-A output stage, whose current drain does not depend on volume, there seems no need to compensate for load changes. What might have been more useful (and possibly what the copywriter meant) would have been control to stabilise the 2V filament supply against changes in gas pressure. Excess filament voltage would have seriously reduced the life of the valves.

You may have heard of "steam radio" but this advertisment offers "gas radio".

Picture kindly provided by John Howell.

Left: The gas-fired thermo-electric generator: 1930s.

With no sign of a connection for an outside flue, I can't help wondering how much carbon monoxide these things produced.

Apologies for poor picture quality.


Left: A Russian thermo-electric generator based on a kerosene lamp.

This lamp was introduced in 1959, once again to power radios. Presumably there were parts of Russia that Stalin's electrification program had not reached. The output voltage(s) are unknown, but since a picture is known to exist of it powering a valve radio, HT must have been generated somehow, possibly by a vibrator power supply.
(In this context a vibrator is an electromechanical device, similiar to an electric bell, that chops low-voltage DC into crude AC that can be applied to a step-up transformer. They were widely used in car radios before semiconductors arrived)

I have just been informed by Pine Pienaar that he has seen one of these things, and it yielded both 1.5 and 90 Volts, so it could replace a composite dry battery with the same output voltages. Such batteries were once widely used to operate small radios.
Such radios typically used four 7-pin valves and needed a 90V HT supply at around 12mA and a 1.5V filament supply at 125mA or 250mA depending on the valves used.

This example seems to be missing its metal chimney. (see pictures below)

Left: Cutting about the Russian thermo-electric generator.

This confirms that the 90V HT was generated directly.

Presumably "invested" should read "invented".

Cutting kindly provided by Ed Maurus, original source unknown.

Left: Russian thermo-electric lamp partly dismantled.

Pablo Reyes tells me that there are thirty cooling fins. The terminal plate has 5 terminals, duly numbered 1 to 5. That's presumably two isolated thermo-electric generator banks and an earth terminal.

Photo kindly provided by Pablo Reyes.

These machines are alive and well, being used in remote places where small amounts of electricity are required and the complications of an internal-combustion engine and alternator are not welcome. Modern versions use a thermopile made up of a series array of lead-tin-telluride semiconductor elements, rather than simple thermocouples. These thermojunctions are much more efficient than simple thermocouples, and have been available since the mid-1960s. They are commonly used (working in reverse, of course) to cool the little sofa-side beer refrigerators which are now quite common.

This gives a very good account of semiconductor thermojunctions and how they work: Thermoelectrics by Tellurex (external link)

For one example of modern gas-fired thermoelectric generation, see: Global Thermoelectric. (external link)

Thermoelectric generators can also be heated by radioactive decay, and such devices are used to power interplanetary space probes and the like, where distance from the sun means that solar power is not an option. See: Free Dictionary: RTGs

Even so, I was thinking that thermoelectric generators must be very rare- and then I found one working away in my garden shed. They are everywhere around us!
They are used in central heating boilers to control the pilot-light valve. When the pilot is burning, the thermopile generates about 750 mV- enough to actuate a small solenoid that keeps the pilot valve open. This sadly doesn't mean you can run a central-heating system with no electric power, as the main gas valve is operated by mains power switched by the room thermostat; in any case, the pump wouldn't run.

Left: A modern thermo-electric generator or thermopile made by Honeywell for boiler control.

The voltage output is 750 mV with the "Cold" Junction at 416 degC (780 F) and the Hot Junction at 760 degC. (1400 F) I know that 416 degC is not exactly cold, but this thing is mounted inside the boiler combustion chamber.
Assuming Ni/NiCr thermocouples are used, we can deduce from this that the device contains about 55 thermocouples in series.

So why is a thermopile used for this job? Presumably because it is very simple and reliable; it is hard to see how a thermopile could fail to the danger state- it can hardly generate electricity when it isn't hot.

Back to Home PageBack to The Museum EntranceTop of this page