Resistive Sensor Head (A2053) Manual

Contents

Description

The Resistive Sensor Head (A2053) is a Long-Wire Data Acquisition (LWDAQ) Device that measures the resistance of up to eleven resistive sensors. These sensors could be 1000-Ω RTDs (resistance-temperature devices), 100-Ω RTDs, 120-Ω strain guages, or any similar resistive sensor. You can use the A2053 with the Guage, Thermometer, or Flowmeter. For examples of the A2053 in action, see the Examples section.

Figure 2: RTD Head (A2053A) with Enclosure. Connectors 1 through 4 are marked in numberals the box itself. Pin 1 of each connector is at the left end. Connector 4, for RTD 11, is left open in the A2053A. The resistive sensor channel numbers are marked in black numerals around the outside of the box.

The RTD Head (A2053A), for example, connects to 1000-Ω RTDs and measures their resistance with 0.1-Ω precision over the range 860 Ω to 1160 Ω. The Strain Guage Head (A2053S) connects to up to eleven 120-Ω strain guages and measures their resistance with precision 0.01 Ω over the range 110 Ω to 130 Ω.

The A2053 connects to each of its sensors with a twisted pair of insulated, stranded copper wires. The resistance of these wires adds to the resistance of the sensor. If the sensor is a 1000-Ω RTD, a 10-m cable made out of twisted pairs pulled from a CAT-5 cable adds a constant +0.5 °C offset to the apparant temperature of the resistor. The resistance of the wires is roughly 1 Ω per 10 m. Because the resistance of such wires does not vary significantly with temperature, a one-point calibration of each sensor is sufficient to remove the effect of the wires.

Instead of running current continuously through the sensors, or using a Wheatstone Bridge to compensate for various electrical and thermal errors, the A2053 runs a small current through the sensor for a few tens of milliseconds. Because the A2053 is connected to a LWDAQ, we can digitize this voltage, display it on the screen, and take its average value. Once we have obtained this average value, we instruct the A2053 to run the same current through the bottom reference resistor on its circuit. With our 1000-Ω RTDs, we use a 1060-Ω resistor, which corresponds to a perfect 1000-Ω RTD at 15.38 °C. We record the voltage again, then select the top reference resistor. For the 1000-Ω RTDs, this resistor is 1100 Ω, corresponding to 25.69 °C. We now interpolate between the top and bottom references to obtain the resistance of the sensor.

In other words: the accuracy of the A2053 is contained in its precision resistors. Precision resistors are stable and inexpensive. They are small, rugged, and we can put them right next to our measurement circuits. The absolute accuracy of our RTD Head (A2053) is better than 50 mK, which is far less than the accuracy of the sensors we use.

In addition to resistance measurement, all versions of the A2053 provide a 15-V heating connection to any one of the sensors. Heating the sensors allows us to measure the cool-down time constant of the sensor in, for example, a pipe with flowing gas, and thus deduce the gas flow rate. The A2053F uses the heater to measure gas flow rate, as we describe here.

The A2053 mounts to a flat metal surface with four M6 bolts. The same bolts hold the A2053 circuit board to the top face of its enclosure. The drawing below shows the locations of the mounting holes. Each hole should be M6 threaded and at least 7 mm deep. Allow at least 40 mm height for the enclosure and its electrical connections.

Figure: Mounting Holes.

All versions of the A2053 connect to their eleven sensors through two eight-way plugs, one four-way plug, and one two-way 0.1" footprint that we usually leave unloaded to allow you to put a sensor right on the circuit board. Two wires connect to each sensors. The table below gives the LWDAQ element number corresponding to each pair of sensor connector pins. For an annotated photograph, see here.

See below for instructions on preparing sensor connections to fit the plugs on the A2035.

Versions

The following versions of the A2053 exist.

The A2053A uses EL2244CS op-amps. It performs well with the LWDAQ Thermometer.

The LWDAQ Flowmeter instrument, however, takes samples from a single RTD sensor for several seconds at a time without sampling the reference channels, and is unable to keep track of the offsets that grow in its EL2244CS op-amps during the measurement. We discovered this offset problem only after we had made several hundred A2053A boards with the EL2244CS. The A2053F is identical to the A2053A, but its three EL2244CS op-amps are replaced with three OPA2277UA op-amps in the same SOP-8 package. Therefore, the A2053F can be used as a Thermometer, but the A2053A does not perform well as a Flowmeter.

The A2053S is designed for use with 120-Ω strain guages, such as the 032UW by Vishay. It is incompatible with 1000-Ω RTD, and also with 100-Ω RTDs.

Connectors

The connectors used on the A2053 (as well as other assemblies like the A2044) come from the C-Grid family made by Molex-Waldom. Do not confuse the C-Grid family with the C-Grid III family. Connectors from these two families do not mate together.

The C-Grid plugs are single-row headers with pins on a 0.1-inch pitch. Each plug is enclosed in a plastic shroud. The shroud provides a keeper for a latch on the mating socket. The table below gives the Molex-Waldom and Digi-Key part numbers for C-Grid plugs, receptacles, and crimp terminals we use in our assemblies.

We strip a couple of millimeters of insulation off each sensor wire and crimp to each wire a female terminal. There are two crimps on each terminal. One should crimp the wire and the other should crimp the wire insulation. We like to use gold-plated crimps, but if you are going to connect your sensors only once, tin-plated crimps are just as good, and less expensive.

When we have the terminals on the wires, we push the terminals into a receptacle. They lock in place. We can remove them again with the help of a pin. The receptacle itself can provide a locking latch and polarization, or it can be without polarization and without the locking latch. When you have plenty of space to grab the connectors when you want to remove them, we recommend the locking receptacles. But when there is little space for your fingers, you can use the receptacle without the latch. The disadvantate of the connector without a latch is that it does not provide polarization either, so you can plug it in the wrong way around. You must look at the triangle in the receptacle face and match this up with pin one on the plug.

When we crimp our terminals, we like to use the manufacturer's crimp tool (part number 11-01-0209). But we can also get the job done with a pair of needle-nosed pliers and a little more time and effort.

Sensors

With adjustments to its reference resistors, the A2053 can be used with any resistive sensor. We obtain best performance with sensors of resistance between 100 Ω and 1M Ω where the resistance of the sensor varies by only ±10% over the measurement range.

A 1000-Ω RTD has nominal resistance 1 kΩ at 0 C. Its resistance increases by approximately 4 Ω/C. The combined resistance of the wires leading to a sensor adds to the resistance of the sensor itself, causing an increase in measured temperature of 0.25 C/Ω.

Example: We observed a 50 mK/m increase in measured temperature with our standard stranded-core twisted pair cables. The longest connection to an RTD in ATLAS is nearly five meters, which introduces an increase in absolute temperature measurement of 0.25 C. But the ATLAS temperature sensors will be calibrated a 20 C, thus removing any absolute error in the error in the absolute temperature measurement, either due to cable resistance or the sensor itself. Our ATLAS temperature sensors are accurate to ±0.3 C at 0 C, so the error due to connecting wire resistance is comparable to the error due to the sensor itself. Furthermore, if we know the length of the cable, we can calculate and remove the effect of its resistance.

The effect of the wires will be ten times greater for 100-Ω RTDs. That's why we prefer 1000-Ω RTDs for use with the A2053.

The dominant application of the A2053 so far has been the A2053A with 1000-Ω RTDs in ceramic packages with radial steel wires. In 2004, we bought five hundred RTDs with 300-mm teflon-insulated leads from Enercorp for $5 each. We installed these in the ATLAS detector. In 2008 we bought one hundred RTDs with short, tin-plated steel leads for $2 each. Most RTD leads are steel, and so must be tinned with the help of acid flux before solderin. (See solder joints for details of soldering steel wires.) Another source of radial-wire RTDs is Omega Engineering, which sells radial-lead 100-Ω RTDs for $1 each in packages of 100, or 1000-Ω RTDs for $1.50 each. At Digi-Key you will find surface-mount 1000-Ω RTDs and radial-wire 100-Ω RTDs.

Specification

The A2053 complies with the Long-Wire DAQ specification. It supports no device-dependent LWDAQ jobs. It has no Device Type and no Elment Numbers.

To determine the command word that will implement a particular operation on the A2053, write out sixteen bits in a row, starting with bit sixteen (DC16) on the left, and ending with bit one (DC1) on the right. Set each bit to zero or one as you require. The left-most four bits form the most significant nibble of the sixteen-bit command word. The right-most four bits are the least significant nibble. Translate each nibble into a hex digit, and you have the hex version of the command word.

The T1 through T11 bits select Sensor 1 through 11. The HEAT bit applies 15 V to the selected sensor or sensors, heating one or all of them. Each heated sensor consumes 15 mA from the +15 V supply, and receives 200 mW of heating power. The WAKE bit wakes up the board by turning on the ±15 V supplies. The TB and TT bits select the bottom temperature reference resistor and the top temperature reference resistor respectively. The LB bit enables the logic loop-back driver, and is used by a LWDAQ Driver's loop job.

Example: To select the top temperature reference, we set the following bits: TT (DC6) and WAKE (DC8). All other bits should be cleared. We compose the following binary number: 0000 0000 1010 0000, which translates to $00A0 (our symbol for hexadecimal numbers is "$"). To heat the same sensor we transmit $02A0.

The A2053 does not respond correctly to the loop job. The loop time we measure with an A2053 is always 0 ns because the A2053 drives R LO when we send it the command that usually sets up a device for loop-back of T.

Measurement

You will find the schematic of the Sensor readout circuit here. The schematic gives the component values for the A2053A, which reads 1000-Ω RTDs. Resistor R25 sets the current flowing down out of U16-1 (the collector of the current source transistor), along K+ (the current does not go into Q18, because Q18 is the heater, and we assume it's turned off), and so to all the sensors via the common K+ connection. But the current flows through only the one sensor whose selection switch (these are the mosfets Q5..Q17) is turned on. The voltage on K+ is the voltage developed across the selected sensor when the measurement current flows through it. When R25 is 2 kΩ, the current is 350 μA. Across a 1000-Ω resistor, this current develops a voltage of 350 mV.

Figure 1: RTD Head (A2053A) Circuit Board, Top and Bottom. The RTD connectors are 0.1" shrouded headers. R23 is the center reference resistor. R26 is the top reference resistor. R27 is the bottom reference resistor. R24 sets the current through the center reference resistance. R25 sets the current through the measurement resistance, which can be R26, R27, or a resistor connected to one of the eleven sensor inputs. Normal operation of the A2053 requires that R24 and R25 be matched to 0.1%.

Resistor R24 sets the current flowing into the center reference resistor, which is R23. We make R24 close to R25, so that the sensor current and the center reference current are the same. That's why the schematic specifies R25 and R25 to 1%. (If you use 5% resistors from the same reel, they almost always agree to 1%. But we used 1% 2 kΩ resistors from the same reel in our mass-produced A2053As.) The center reference resistor has the resistance we expect of our sensors in the center of our desired dynamic range. In the case of the A2035A, we want to measure temperatures close to 20 °C in an experiment hall with 1000-Ω RTDs. These RTSs have resistance 1000 Ω at 0 °C, and their resistance increases by roughly 4 Ω/°C above that. So the A2053A center reference is 1080 Ω.

The two voltages K+ and K− proceed to the input of a differential amplifier shown here. The amplifier consists of two op-amps in package U15. It amplifies the difference between K+ and K−, which we call K, and sends the amplified difference back to the LWDAQ driver for low-pass filtering at 10 kHz followed by sixteen-bit digitization.

The useful voltage range of K is around ±40 mV. When the sensor resistance equals center reference resistor, K will be zero. With a reference current of 350 μA, K will be +40 mV when the sensor resistance is ≈110 Ω higher than the center resistance. Likewise, K will be −40 mV when the sensor is ≈110 Ω lower. The 350-μA reference current gives us a dynamic range of ±110 Ω. With 1000-Ω RTDs, this ±110 Ω gives us a dynamic range of ±28 °C, which is why the A2053A operates over a 55 °C range.

The accuracy of the A2053 comes from its use of two reference resistors. These are R26 and R27 in the schematic. The A2053 treats these just like sensors. But they are not sensors, they are known resistance values to which we compare the sensors. Resistor R27 is the bottom reference resistor and R26 is the top reference resistor. In the A2053A these are 1060 Ω and 1100 Ω respectively, and both are accurate to 0.01%. They are model SM5 wire-wound resistors from Reidon. They represent 1000-Ω RTD temperatures 15.38±0.03 °C and 25.69±0.03 °C respectively.

To measure the resistance of a sensor, we select the bottom reference resistor, then the sensor, then the top reference resistor, and we measure K for each. We interpolate between these values of K, and our known reference resistances, to obtain the sensor resistance. We then convert the sensor resistance into temperature or strain using the sensor manufacturer's data sheet.

A2053A

These component values in the schematic give the A2053A the following properties.

• Compatible RTD: 1000-Ω at 0°C, temperature coefficient 3850 ppm/°C.
• Measurement Current: 300 μA.
• Dynamic Range: 20±35°C.
• Sensor Self-Heating: See RTD manufacturer specification.
• Sensor Accuracy: See RTD manufacturer specification.
• Precision: 20 mK.
• Accuracy: 50 mK at 20°C.
• Reference Temperatures: top 25.69±0.03°C, bottom 15.38±0.03°C.
• Reference Resistors: top 1100±0.1Ω, bottom 1060±0.1Ω.

In other words: the A2053A measures resistance to 0.1 Ω precision over a dynamic range of 860 Ω to 1160 Ω, which is 300 ppm of the dynamic range.

A2053F

The A2053F is identical to the A2053A, except that it uses precision op-amps, the OPA2277 instead of the EL2244CS. The precision op-amps avoid offset drifts during the prolonged measurement period of the Flowmeter.

A2053S

The A2053S operates with 120-Ω strain guages like the 032UW. We want the center of our dynamic range in strain to be 0% strain, so our center resistance, R23, must be 120 Ω. The sensor's guage factor is around 2.0, so its resitance increases by 2% for each 1% strain (length increases by 1%), or 2.4 Ω for each 1% strain. To give the A2053S a ±4% dynamic range, we need to measure resistance between 110 Ω and 130 Ω. With a measurement current of 4 mA, the dynamic range of the A2053S would be 110−130 Ω. But we find that a measurement current of 4 mA gives poor performance. The measurement current source transistors heat up, and perform less well as constant-current sources. We find this is the case for measurement currents greater than 1 mA. So we use 1 kΩ resistors for R24 and R25, and set the measurement current to around 700 μA.

Now we pick the top and bottom reference resistors. Let them be 125 Ω and 115 Ω, wire-wound 0.01%, < 10 ppm/°C temperature coefficient. These values correspond to at +2.08±0.005% and −2.08±0.005% strain respectively.

These component values give the A2035S the following properties. For some raw data, see here.

• Compatible Strain Guage: 120-Ω at 0% strain, 2.0 guage factor.
• Compatible RTD: 100-Ω at 0°C, temperature coefficient 3850 ppm/°C.
• Measurement Current: 700 μA.
• Dynamic Range: 60 to 180 Ω, ±100% strain, −100 to +300 °C.
• Sensor Self-Heating: See manufacturer specification.
• Sensor Accuracy: See manufacturer specification.
• Precision: 10 mΩ, 40 ppm strain, 25 mK.
• Accuracy: 100 mΩ at 120 Ω, 400 ppm at 0% strain, 0.5°C at 20°C.
• Reference Temperatures: top 64.94±0.03°C, bottom 38.96±0.03°C.
• Reference Strains: top +2.08% strain, bottom −2.08% strain.
• Reference Resistors: top 125±0.01Ω, bottom 115±0.01Ω.

A2053L

The A2053L is for use at cryogenic temperatures (L is for Low). We calibrate the A2053L with boiling nitrogen, which is at −195.8°C, and subliming carbon dioxide, which is at -78.5°C. We use 1000-Ω platinum RTDs, just as we would with the A2053A. We bought ours from Enercorp, and they came with teflon-insulated leads soldered to the steel wires of the RTD elements. Because we do not require more than a few degrees accuracy from the A2053L, we don't bother with 0.01% reference resistors. We obtained our reference values by immersing a selection of 1000-Ω RTDs in liquid nitrogen (−196°C) and squeezing them between blocks of dry ice (−78°C, see here for squeezing).

• Compatible RTD: 1000-Ω at 0°C, temperature coefficient 3850 ppm/°C.
• Measurement Current: 38.9 μA.
• Dynamic Range: −50±150°C.
• Sensor Self-Heating: See RTD manufacturer specification.
• Sensor Accuracy: ±5°C −196°C.
• Precision: 0.5°C at −196°C
• Accuracy: ±5°C at −196°C.
• Reference Temperatures: top −78°C, bottom −196°C.
• Reference Resistors: top 700 Ω, bottom 200Ω.

To obtain the above characteristics, we set R24 and R25 to 18 kΩ, R26 (top reference) to 700Ω, R27 (bottom reference) to 200Ω, and R23 (center reference) to 400 Ω. We used the A2053L in the tests we describe here. Because of our calibration process, in which we measure the resistance of a reference RTD at the temperature of boiling liquid nitrogen and subliming dry ice, we expect the A2053L to be exact at −196°C and −78°C with our reference RTD. When we tried out other RTDs, we found that they agreed to within ±1°C at both temperatures.

Operation

We read out the RTD Head (A2053A) with the LWDAQ software's Thermometer Instrument. Here we tell you how the hardware behind the thermometer works.

To read out the temperature sensors, you transmit command words one at a time to the A2053 using your LWDAQ Driver's command job. The A2053 has a current source which you can, by sending the correct command word, direct through one of six resistors. Two of the resistors reside on the A2053 board itself, and are 0.01% precision wire-wound resistors of 1060 Ω and 1100 . The 1060 Ω resistor provides a low-temperature reference point of 15.38°C for standard 1000-Ω RTDs, and the 1100 Ω resistor provides a high-temperature reference point of 25.69°C.

When you select the high-temperature reference, by setting bit DC6 (TT) in the command word, the A2053 sends back an analog voltage that is a linear function of the high-temperature reference resistance. You digitize this voltage using the LWDAQ Driver. The A2037 driver, for example, provides a sixteen-bit ADC which will do the job with sufficient accuracy to obtain 20 mK resolution. When you set bit DC16 (BT), you obtain a voltage from the A2053 that is the same linear function of the low-temperature resistance. Now you can select the four RTDs in turn, and interpolate between the top and bottom reference voltages to obtain the temperature of the RTD.

Example: We have a RTD Head (A2053) connected to an LWDAQ Driver (A2037). We connect a single platinum temperature sensor to the first two pins of P1 on the A2053. We use the A2037 command job to send command $00A0 (hexadecimal) to the A2053. This command wakes up the board and selects the top reference temperature. We allow 1 ms for the A2037's 10 kHz input filter to settle. We execute an adc16 job and retrieve the sixteen-bit result. For an A2053A this will be between 29,000 and 31,000. We send command $8080 to the A2053, which selects the bottom reference temperature. We wait 1 ms, and execute another adc16 job. The result should be between 38,000 and 41,000. We select our temperature sensor, which is T1, with command $0880, wait 1 ms, and execute an adc16 job. The result should be between 0 and 65,500 provided the sensor temperature is between -15°C and +55°C. We know that the top temperature reference is 25.69°C, and the bottom is 15.38°C, so we interpolate between the two to obtain the temperature of our sensor. We are assuming, of course, that the sensor is a 1000-Ω RTD. Note that the high-temperature reference voltage at the A2037's ADC is lower than the low-temperature reference voltage.

If you want to heat a sensor, then use the following command words, which are the same as the sensor read-out select words, but with the HEAT bit set (DC9).

We use a reference socket to test our A2053 boards. The reference socket is an 8-way socket we connect to P1. Pins 1 and 2 of the referenec socket connect to a 1060-Ω 0.01% resistor, pins 3 and 4 connect to a 1100-Ω resistor, pins 5 and 6 are connected directly together with a wire, and pins 7 and 8 are left open-circuit. The temperature we measure on T1, using the procedure given in the above examples, should be within 0.03°C of the bottom temperature reference (15.38°C). The temperature we measure on T2 should be equally close to the top temperature reference (25.69°C). From T3 we get the low-temperature end of the A2053's dynamic range. For an A2053B this should be below -10°C, allowing you to use an ice-water bath to calibrate your temperature sensors. From T4 we get the high-temperature end of the dynamic range. For an A2053B this should be above 50°C, allowing you to measure ambient temperature in most electronic equipment.

The A2053 is asleep when it powers up. It also goes to sleep when you execute a LWDAQ sleep job. To measure the propagation delay of signals travelling from the driver to the A2053 and back again, you execute a LWDAQ loop job and read the loop time out of the driver.

You will find the data acquisition steps required to control and read out voltages from all versions of the A2053 in Gauge.tcl, which is the TclTk script that defines the Gauge Instrument in our LWDAQ Software. You can also look at Thermometer.tcl, which defines the Thermometer Instrument. In Driver.tcl you will find the routines that compose TCPIP messages to communicate with a LWDAQ Driver.

Flowmeter

We describe the Flowmeter instrument in the Flowmeter section of the LWDAQ Manual. You will find details of the data acquisition process in Flowmeter.tcl. The Flowmeter measures the mass flow rate of gas in a pipe by heating up a platinum RTD and measuring how fast it cools. The RTD Head (A2053) provides a heater with which you can heat a sensor in a flow of gas. The time constant of its return to ambient temperature is, for large enough flow rates, proportional to the inverse of the mass flow rate.

The flowmeter sensor is immersed in the gas flow it is supposed to measure. The flowmeter measures the temperature of the sensor, and assumes this temperature to be the ambient temperature of the gas, to which the sensor will be cooling. The flowmeter applies +15 V to the sensor by selecting the sensor with the HEAT bit set. It waits for a second or two, and then removes the +15V. It waits for another second to allow certain initial second-order cooling effects to settle down, and then begins recording the exponential cooling of the sensor towards ambient temperature. It calculates the time constant and reports the inverse-time constant to its text window.

To convert the inverse time constant into a gas flow, you need a calibration curve for the sensor in that particular gas and in a passage of that particular cross section. You will find a study of the Flowmeter's performance here. For best performance, the Flowmeter requires the A2053F (F as in Flowmeter), which uses OPA2277UA op-amps instead of our customary EL2244CS op-amps.

Strain Guage

Below is a picture of an Strain Guage (A2053S) with a Reference Plug. The Reference Plug provides four resistances by which you can check the operation of your A2053S.

Figure: A2053S and Reference Plug. Marked are (1) 109.1±0.1Ω resistor between pins 1 and 2, (2) 0 Ω short-circuit between pins 3 and 4, (3) A 100-Ω RTD between pins 5 and 6, and (4) 120±0.1Ω resistor.

If you insert the reference plug into socket 1 on the A2053S, you will be able to refer to each resistance by the element number given in the figure above. Set ref_bottom to 115 Ω and ref_top to 125 Ω to measure each reference resistor in Ohms. The short circuit allows you to see the low end of the A2053S dynamic range. If you read an empty channel on the A2053S, you will get the top of the dynamic range. The 109.1 Ω resistor corresponds to a perfect 100-Ω RTD at 23.6°C. The 120 Ω resistor corresponds to 0% strain on a perfect 120-Ω strain guage. We provide the RTD so you can test the A2053S temperature measurement. Enter 38.96 for ref_bottom and 64.94 for ref_top, and channel 3 will act as a thermometer.

Glue

When we use RTDs to measure the temperature of a piece of metal, we glue it to the metal in two stages. First we secure the sensor to the metal with cyanoacrylate (super-glue), then we cover the sensor with non-conductive epoxy. The sensor has two sides to it, one of which has a lump that secures the sensor leads to the platinum film. It is the flat side that you must glue to the metal, because the flat side is a non-conducting ceramic surface. The platinum film is on the other side, and must be kept away from any conducting surface.

Paul Mocket of the University of Washington describes the effect of glue upon the temperature measured by the sensors here. We have never noticed any effect of glue upon the sensors, but we have never measured the temperature before and after gluing in a reliable way. Some RTD users found that they had problems when they glued the wrong side of the sensor down to the metal surface. We are not surprized, because in this case the platinum film itself is in contact with the metal.

Power Consumption

We picked an A2053A at random and measured its power consumption both asleep and awake, as shown in the following table.

The A2053 consumes additional power when you neglect to select one of its reference resistors or connected RTDs. When no sensor is selected, the returned analog voltage saturates, and the A2053 op-amps drive thirty milliamps into the 100-Ω load at the end of the cable on the multiplexer or driver to which the A2053 is connected. To obtain the lower, designed, waking power consumption, we selected the top reference resistor.

Electronics

Here are the six pages of the A2053 circuit diagram.

You will find printed circuit board files here.

Examples

Here we give some examples of experiments we have performed with the help of the A2053.

Coctail Cooling

We discuss with our friend Eben Klemm the manner in which ice cools a cocktail. We claimed that the melting point of ice depends upon the alcohol content of the fluid it's floating in. To measure the temperature of an alcohol-ice mixture, we need water-proof temperature sensors and an A2053 with dynamic range well below 0°C. The A2035A provides −15°C to +55°C. But we took out a version of the A2053 in which R24 and R25 are 5.6 KΩ instead of 2 kΩ, so that the dynamic range of the thermometer with our 1000-Ω RTD sensors is −50 to +115°C.

We put two RTDs in the fingers of a latex glove and immersed them in a beaker of ice with water. We recorded temperature, and found that both sensors settled to 0±0.05°C. We poured out the water and replaced it with pure ethanol. We agitated the mixture and recorded temperature. We continuted to agitate the mixture at random times over the next half an hour.

We use the Acquisifier Tool with this script to record temperature from the two sensors inside the beaker, and another sensor outside the beaker, at intervals. We present the temperature of the mixture in the graph below.

Figure: Temperature with Time in Alcohol and Ice Mixture.

As you can see, the arrival of the alcohol results in a water-alcohol mixture that is well below freezing. The temperature of the mixture appears to reach equillibrium for several minutes at around −8°C. We know the ice we started with as at 0°C and the alcohol was at room temperature. And yet the mixture is colder.