Building a Thermocouple Vacuum Gauge


The operating principle of the thermocouple vacuum gauge and instructions for building a low-cost power supply & readout that is compatible with commercial gauge tubes.

The full version of this article appeared in Volume 1, Number 4 of the Bell Jar.


The thermocouple (or T/C) gauge is one of the more common and cost effective gauges for vacuum pressure measurement in the 1 Torr to 1 milliTorr range. The T/C is usually found in the forelines of high vacuum systems (i.e. between the roughing and diffusion pumps) as well as in single pump systems of the sort used to evacuate sign tubes.

Like most vacuum gauges, the T/C gauge does not measure pressure directly as do, for example, manometers of the McLeod or Bourdon type. Instead, these vacuum gauges depend on changes of a physical characteristic of the residual gas within the gauge tube. In the case of the T/C gauge, and all other thermal conduction gauges, that characteristic is the thermal conductivity of the gas.

A thermal conduction gauge may be thought of as a defective vacuum insulated thermos bottle (refer to Figure 1).

Each has a hot element (coffee for one, a filament in the case of the other) within a vacuum wall. There are two ways of removing heat: conduction (molecule to molecule) and radiation. For both coffee and warm filaments the primary path at atmospheric pressure is conduction. As it turns out, the thermal conductivity of air is nearly constant down to a fairly low pressure - about 1 Torr. Then it begins to change rather linearly with pressure down to a value of about 1 mTorr, whereupon conduction through the gas ceases to be a major factor. At that point, the dominant loss factors are conduction through wall and leads, and radiation.

What might be surprising to many people is that a fairly good vacuum is needed in a thermos. With a bit higher pressure, you might as well have no vacuum. In the case of the thermal conduction gauge, operation will only occur within the sloped portion of the curve. An interesting experiment would be to nick open a thermos bottle refill and measure the cool-off rates for hot water with the bottle evacuated to a number of pressures. The result would be a useful, but very slow, thermal conduction gauge.

The T/C gauge contains two elements: a heater (filament) and a thermocouple junction which contacts the filament. With the filament current held constant, as the pressure within the tube is decreased the filament will become hotter because of the improved thermal insulation provided by the increasingly rarefied gas. This temperature is sensed by the thermocouple junction. Measurement is accomplished by reading the thermocouple junction voltage on a sensitive meter which has previously been calibrated against a manometer. Simple T/C gauges may be obtained from a variety of sources such as Duniway Stockroom or Kurt J. Lesker Co. These gauges consist of the gauge tube itself, a power supply for the filament, and a moving coil (d'Arsonval) meter for displaying the pressure. Tubes usually have a 1/8" male pipe thread for coupling to the vacuum line and an octal (vacuum tube) base for mating with a socket. In newer gauges, the power supply is usually nothing more than a plug-in type ac adapter with a potentiometer for adjusting the current. Each type of T/C tube has its own calibration curve. Also, as there are some structural variations from tube to tube within a type, each has its own filament current rating. The current at which the gauge will conform to the calibration curve is imprinted on each tube. The gauges are calibrated for air. As different gases have varying thermal conductivities, the gauge will not be accurate when working with, for example, argon or carbon dioxide.

Making Your Own Gauge Controller

As was previously noted, complete basic T/C gauges are available from a variety of suppliers. Typical prices are in the $200 to $250 range, new. Given the basic simplicity of a T/C gauge, building one from available parts would not seem to be difficult. The gauge tube and the readout (thermocouple) meter are the only specialty components. For the tube, I don't think that there is much purpose to trying to build your own. Buy one of the cheaper ones (some suggestions will follow below). They can be had for about $40 new. The meters are specialized items in that they have to be compatible with the millivolt level, low impedance output of the gauge's thermocouple.The more commonly available milliamp/microamp meters have coil resistances many times the 55 ohms of the meter used in a typical gauge controller. Connect up a standard microamp meter to the gauge tube and it might budge, but probably not much. If you buy a meter as a subassembly (Dunaway sells the meters separately, but they are not cheap if your reference point is your junk drawer or a surplus catalog) you will get a very professional and calibrated readout as long as you use the tube for which it was intended. Even with this route, the complete gauge should come in at half the price of a new commercial unit.

A very satisfactory alternative involves the placement of an IC amplifier/buffer between the gauge and the meter. By selecting the right values of components, almost any meter can be coupled with any gauge tube as long as you know the tube’s maximum output and calibration curve. The next section will detail how to build an op-amp based T/C gauge using either of two inexpensive tubes.

An Op-Amp Based T/C Controller

The meter side of this controller is based on a single stage op-amp amplifier configured in the inverting mode. To establish the component values in the circuit (the values of the input and feedback resistors) one needs to know the load resistance for which the T/C tube was calibrated and the maximum output voltage of the thermocouple at “full” vacuum. The latter corresponds to a full scale deflection of the meter and is taken at a pressure of 10-4 Torr. The tubes we shall consider are the 531 and 6343 (and their Kurt J. Lesker equivalents). Relevant data on these tubes is shown in Figure 2.

The circuit is shown in Figure 3.

Since the input impedance of an inverting amplifier is set by the input resistor, the value for this should be 55 ohms. I elected to measure the output with a 30 k-ohm/volt multimeter set on the 1 volt scale. Thus the gain of the amplifier would have to be set to up the 14 mV T/C output to 1 volt, a gain of 71.4. As the amp’s gain is set by the value of the feedback resistor, Rf, divided by the value of the input resistor, Ri, Rf should be about 3.9k. As it turned out, the closest values I had on hand were 47k and 3.3k which would give a gain of 70.2. Close enough, I figured.

The op-amp used was a 741 and the circuit was assembled on a Radio Shack proto pc board, catalog number 276-159. I used a regulated +/- 15 volt supply but a couple of 9 volt batteries would work as well. Likewise a 50 mA meter (surplus of course) with a series resistor could be used in place of the multimeter. Do include the offset pot for zeroing.

On the filament supply side it does not matter which pin you select as the positive pin. However, it is essential that the filament supply be independent of the amplifier circuit (i.e. no common ground). Otherwise you will end up just amplifying the filament voltage (the filament and the thermocouple are electrically connected). The filament pot (as well as the offset pot) are 10 turn wirewounds. Fair Radio Sales and other surplus electronics houses have them for about three dollars per.

To get the gauge going, connect the tube to your system with the threaded connection (use Teflon tape or other sealant) or just slip it into a piece of tight fitting rubber vacuum tubing and tighten with a hose clamp. Octal sockets are available from Fair Radio and 4 conductor telephone type cable is good for the tube to controller connection (this should be no longer than 10 feet or so). Be sure to have the filament current control at the lowest setting so you don't burn out the tube. Begin to pump down the system and set the offset pot for a “0” reading on the T/C meter. Then begin to bring the filament current up to the value marked on the tube. The T/C meter should begin to creep up indicating that (1) the circuit is working and (2) that you are pulling a vacuum.

Many T/C tubes don’t do well when operated at atmospheric pressure. To preserve your tube, don’t apply filament power until you are sure that you are drawing a vacuum in the system. Also, avoid getting contaminants in the tube and position it at a location in the system plumbing where oil cannot back up into it.


Now, all you need to know is the correspondence between the meter reading and pressure. The table of Figure 2, with data points scaled directly from production gauges, gives a reasonably accurate set of points with which to develop a calibration curve. Even with the sloppy resistor selection, my prototype controller tracked a commercial gauge pretty well. Also, bear in mind that T/C gauges are not particularly accurate instruments. Most often they are used only as rough indicators of pressure where 10 to 20 percent accuracy is acceptable.

Other Thermal Conductivity Gauges

There are two other common types of thermal conductivity gauge.

The Pirani gauge has a fine wire filament that has a high temperature coefficient of resistance. The wire acts as both the heater and the sensor. Usually a Pirani gauge is part of a Wheatstone bridge circuit that also includes a temperature compensating element. Well designed Pirani gauges offer better accuracy and response time than do thermocouple gauges (often tens of milliseconds vs. several seconds).

The thermistor gauge is the least common of this class of gauge. Using a small thermistor, the principle is very similar to that of the Pirani. However, the element is more massive and the response time is slower. Roy Schmaus of the University of Alberta described thermistor gauge in Volume 4, Number 2 of the Bell Jar. Details are also provided on Roy’s Web Site.

Home-made Pirani gauges can be made from small light bulbs, carefully opened, or even from model aircraft engine glow-plugs. (See Volume 4, Number 1 of the Bell Jar.

Home     tBJ Articles & Publications    Chambers & Supplies    Contact tBJ
©2008 Stephen P Hansen, the Bell Jar