Using UC3638 PWM controller to realize the design of semiconductor thermoelectric cooler temperature control drive scheme

semiconductor thermoelectric cooler (Thermo-E1ectric Cooler, TEC for short) has the characteristics of small size, no noise, no pollution, etc. Widely used in aerospace, military, optoelectronics, electromechanical, medical, automotive, communications and other fields. In the development process of a certain type of instrument, the metal block needs to be quickly heated, cooled and kept at a constant temperature, and the thermal cycle process control is performed. TEC can well meet this requirement.


Semiconductor thermoelectric cooler (Thermo-E1ectric Cooler, TEC for short) has the characteristics of small size, no noise, no pollution, etc. Widely used in aerospace, military, optoelectronics, electromechanical, medical, automotive, communications and other fields. In the development process of a certain type of instrument, the metal block needs to be quickly heated, cooled and kept at a constant temperature, and the thermal cycle process control is performed. TEC can well meet this requirement. TEC relies on heat exchange to absorb heat from one side and release heat from the other side to achieve heating and cooling. Since almost all TEC electric power is converted into heat energy QF=UI, only when the heat exchange amount QJ of the TEC heat-absorbing surface is greater than the calorific value QF/2 (assuming that the heat generated by the electric power is equally distributed on both sides), and when the heat-emitting surface effectively dissipates heat, Only the TEC heat-absorbing surface can be cooled (the heat-absorbing surface heat absorption QIN=QJ-QF/2, the heat-radiating surface heat dissipation QOUT=QJ+QF/2). The TEC heat-absorbing surface or the radiating surface is determined by the direction of the TEC current (TEC can be regarded as a non-linear resistance load, the direction of the current changes with the direction of the voltage). The heat absorption is affected by the temperature difference and the magnitude of the current at the same time; and in order to obtain different temperature differences and heat absorption at the same voltage, the current changes greatly. Therefore, using TEC for heating and cooling, the control current is better than the voltage; because of the need to adjust the size and direction of the current, it is called bipolar current control. In order to realize the characteristics of thermal cycling process control, it is necessary to design a TEC drive controller.

1. Determine the TEC temperature control drive scheme

The temperature control objects have various shapes, and the TEC cannot be made too large under the influence of thermal stress, so in many cases, multiple TECs must be used. TEC electrical connection methods include series, parallel and series-parallel. Parallel use is more difficult because of the difficulty of wiring and large conduction loss during driving. Therefore, according to factors such as device capacity and temperature control objects, we choose series and series-parallel connections. Although the existing TEC analog control IC and circuit have high temperature control accuracy at a certain temperature point, they are mainly for low-power applications and have a low operating voltage, which is not suitable for rapid temperature changes under high power, and the chip price is relatively expensive. However, the full digital control method of the microcontroller is used to generate PWM, which is not conducive to the miniaturization of the system due to the low PWM frequency. UC3638 is an enhanced motor control IC of TI, which can form a high-performance DC motor PWM drive circuit. It has a differential current amplifier, which can be used with an error amplifier to form an average current feedback, so it can also be used for unipolar or bipolar adjustable current occasions. Using DC stabilized power supply, bipolar current can be obtained by driving the full bridge through UC3638, and it is simple and easy to form a TEC drive (Of course, OV can also be designed to adjust the DC steady current power supply, and then the solid state relay is used to form a full bridge to change the direction of the current, but the design It is more complicated, which is not conducive to the rapid construction of the system in the early stage of development to evaluate the entire instrument). Appropriately increase the operational amplifier, UC3638 can form two control methods with direct current setting and temperature setting (the current setting signal is generated by analog PID), and no microcontroller intervention is required. The structure of the TEC temperature control system is shown in Figure 1. The PC sends out the temperature points and controls the time, and then the 89C52 completes real-time temperature monitoring, temperature upload and digital control (optional temperature setting or direct current setting), after D/A conversion, and then through analog PID or direct delivery To the analog isolation amplifier; finally, the bipolar current controller composed of UC3638 plays the role of inner loop current adjustment, drives the TEC, and heats and cools the temperature control object. The temperature of the object is detected by a temperature sensor and amplified by signal conditioning. On the one hand, it is sent to the analog PID temperature adjustment, and on the other hand, it is converted into a digital quantity by A/D for the 89C52 to complete the digital temperature control.

 Using UC3638 PWM controller to realize the design of semiconductor thermoelectric cooler temperature control drive scheme

2. Introduction to UC3638 chip

The UC3638 enhanced DC motor PWM controller is suitable for various types of DC motor PWM drive control, and can also be used to design power amplifiers that require unidirectional or bidirectional current drive. Its internal structure is shown in Figure 2. It contains an analog signal error amplifier and a PWM modulator; according to the error to amplify the polarity and size of the input signal, the PWM modulator outputs two pulse trains with different polarities and widths. Therefore, it can be used in two-way speed control systems and other occasions that require unipolar or bipolar adjustable voltage or current. Due to improved circuit design and increased integration, UC3638 reduces many peripheral circuit components. It also has the following features: the circuit speed has been significantly improved, there is a programmable high-frequency triangle wave generator, a high conversion rate error amplifier, a high-speed PWM comparator, and the PWM switching frequency can reach 500kHz. The increased 5 times fixed gain differential current amplifier and error amplifier can form an average current feedback control circuit to improve the system response speed (current-type control). Two 60V/50mA open collector outputs can drive the upper tube of the full bridge, and two 500mA totem pole outputs can drive the lower tube of the full bridge. The programmable pin AREFIN allows single power supply or dual power supply operation. Oscillator ramp amplitude and PWM dead zone are divided by 5V, which are set by setting pins PVSET (oscillator ramp amplitude setting) and DB (PWM dead zone setting). In addition, it also includes an accurate 5V reference voltage output, under-voltage lockout, bi-directional cycle-by-cycle peak current protection, and a remote control shutdown pin (which can also be used as a soft start).

Using UC3638 PWM controller to realize the design of semiconductor thermoelectric cooler temperature control drive scheme

3. Principle and design of bipolar current controller

As shown in Figure 3, the current control signal Vc from the isolation amplifier passes through R1. , And the current feedback signal from the external differential current amplifier A1B is added through R3, and the current PT is reduced by the internal error amplifier of UC3638 (R2, C1, C2 constitute a compensation network), and then the internal PWM comparator forms the PWM Control signal Aoutl. Aout2, B0utl, Bout2, form drive signals AoutL, AoutH, BoutL, BoutH through VQ1~VQ8, drive the full bridge of VT1~VT2 in Fig. 4, output PWM power signal, and then through L1, L2, C11. C12 filters the bipolar current (voltage) to supply power to the TEC (RTEC in Figure 4). The bridge current signal is detected by resistors RS1 and RS2 (actually multiple low-resistance resistors in parallel), amplified by an inverting amplifier composed of A1A and A1B, and then sent to A1B in Figure 3 to form an average current feedback.

Using UC3638 PWM controller to realize the design of semiconductor thermoelectric cooler temperature control drive scheme

3.1 UC3638 peripheral circuit design

1) As shown in Figure 3 and Figure 5, assuming that the power supply of UC3638 is ±12V and the amplitude of the triangle wave generator is 10Vp-p, according to the design requirements of UC3638, the level of the pin PVSET is set as follows: VPK-VVLY=5VIPVSET, VIPVSET =VR6=10/5=2 V, take R6 as 10 kΩ, IR5=IR6=2/(10×lO3)=O. 2 mA, R5=(VAREF-VR6)/IR6=(5-2)/(O.2×lO-3)=15 kΩ. R5 approval is subject to actual debugging.

2) If take 1V dead zone voltage, 5-VDB=1V, VDB=4V, take R12 as 10 kΩ, IR1=IR12=VDB/R12=4/R12=4/10×103=O. 4mA. R11=(5-VDV)/IR11=1/(O. 4×10-3)=2.5 kΩ. The value of R11 is subject to actual debugging. Parallel C6 on R12 can obtain soft-start characteristics (that is, the dead zone gradually decreases from large after power-on).

3) Take VCC-VSD=8V (less than 2.5 V to enter the soft-start state), R14=10 kΩ, IR14=IR15=8/10×l0-3=O 8 mA, R15=(2×VCC-VR14)/H15 =(24-8)/(O.8×10-3)=20 kΩ. In order to obtain the chip delay enable feature: connect the capacitor C8 in parallel with R14.

4) Take the external capacitor C4 of the frequency generator as 1000 pF, since f/=l/ (5×RT×CT), f=35 kHz, R13=RT=1/(5×f×GT)=1/( 5×35kHz×1000 pF) ≈5.8kΩ. Take R13 as 6.2 kΩ. When RT=6.2 kΩl, the charging current of foot RT is limited to 2.4 V/6. 2 kΩ=O 387 mA. Less than the specified maximum 1mA limit.

5) Check and calculate the dead time tDB=VDB/[(VPK-VVLY)×=(5-VDB)×RT×CT/VPVSET=1×68kΩ×1000pF/2V=34μsItcanbeseenthatthedeadtimeismuchlongerthantheswitchingtimeoftheMOSFETandR11canbeadjustedaccordingtotheoutputwaveforminpracticalapplications

3.2 Current detection circuit design

In practical applications, because multiple TEC modules need to be used in series and parallel to increase the heating and cooling power, the drive output voltage and current are relatively large (maximum design value ±24V/20A, actual measurement ±25V/17.5A). In order to improve efficiency and reduce heat generation, we use multiple low-resistance resistors in parallel as current detection sampling resistors. In this way, the current detection differential amplifier inside the original UC3638 is not amplified enough, and an amplifier must be added. But the experiment found that, when the PWM driver is working, the high-speed switching of the circuit produces very large dv/di and di/dt, resulting in a common-mode peak voltage, which causes a large ground interference. After taking various measures to reduce interference, because the current limit threshold inside the controller is low (±2.5V), it is easy to cause the drive circuit to self-lock without output. Therefore, we realized high-rate differential current amplification and level shifting by adding operational amplifiers (A1B in Figure 3 and A1A and A1B in Figure 4) according to its internal circuit (short-circuit CS+ and CS to ground to shield the internal differential amplifier). ), and the final differential current amplification output is connected to CSOUT through the resistor R4 in Figure 3 to achieve the maximum current limit function. At the same time, a small capacitor is added to the signal path to filter out high-frequency interference. In addition, a capacitor is added between the gate and source of the MOSFFT to reduce the adverse effects of circuit switching and Miller capacitance on the gate drive signal. This measure, together with the dead zone forming circuit, enables the driver to work reliably, and there will be no through phenomenon of upper and lower tubes. And effectively improve the stability of the control circuit.

3.3 Drive and full bridge circuit

NPN and PNP complementary switch tubes are used to amplify and level shift the PWM output control signal. The P-channel tube IRF14905 is used as the upper tube, and the N-channel tube IRF3205 is used as the lower tube to form a full-bridge circuit. It is worth noting that when the maximum duty cycle is required to reach l, ordinary pump circuits (usually used to drive the N-channel upper tube) cannot be used to drive the upper tube.

3.4 Design of output LC filter circuit

As shown in Figure 6 and Figure 7, suppose the voltage Uo at both ends of the TEC is approximately constant within a period, UAB is the voltage between the midpoints of the bridge arm, US is the bridge arm supply voltage, IL is the Inductor current, and IOMAX is the maximum output current of the driver. , D is the duty cycle, TS is the switching period, fs is the switching frequency, L=L1=L2 is the filter inductance, C=C1=C2 is the filter capacitor, RTEC is the equivalent resistance of the TEC, the LC output filter circuit and The voltage across the bridge arm and the filter inductor current waveform can be obtained


=(Us+Uo)9(l―D)Ts/(2L) (1)

Because Uo=(2D-1) Us,

So △IL=D (1-D) TsUs/L.

When D=0. 5, △IL has the maximum value △ILMAX=Us/(4fsL). L can be selected to make the ripple current △IL, not more than (10-20)% of LOMAX.Since the temperature difference that TEC can reach will decrease with the increase of ripple current, the approximate derating formula is

dθ/dθMAX=1/(1+N2) (2)

In the formula: dθ is the temperature difference that the TEC can reach under the current ripple;

dθMAX is the maximum temperature difference that TEC can reach under DC current;

N is the current ripple factor, and usually manufacturers require that the current ripple factor should not be greater than 10%.

Because the derivation of TEC ripple current is more complicated, here is an estimation formula in the literature for reference.

Where: fs is the frequency of the oscillator and is also equal to the switching frequency of the PWM voltage;

L=L1=L2 is the inductance of the filter;

C=C1=C2 is the capacitance of the filter;

The equivalent resistance of RTECTEC;

VTEC is the DC voltage drop of TEC;

ESR is the equivalent series resistance of the filter capacitor;

VS is the power supply voltage of the full bridge.

Equation (3) shows that when the filter inductance, power supply voltage, module, etc. are determined, increasing the switching frequency, increasing the capacity of the filter capacitor, and reducing the equivalent series resistance of the filter capacitor can reduce the current ripple. Of course, it is necessary to weigh the increased circuit board area, power consumption and other factors to finally determine the values ​​of L, C, and fs.

4. Experimental results

Experiments were conducted on the direct current control method (PID is completed by 89C52) and the temperature setting control method (PID is completed by analog circuit). It is found that the direct current control method and the digital control algorithm are more difficult to design, and the control effect is not as good as that of the analog PID control method. Good (only up to ±O. 3℃). Only use a proportional regulator with appropriate gain (difference control), and appropriately attenuate the high-frequency gain, use 89C52 to set the temperature without any adjustment (ie, digital controller open loop), and the short-term temperature stability can reach ±O. About 15℃, but the digital set temperature open-loop control analog adjustment exists. The temperature may not be at the set point (on the one hand, relative to the set temperature of the host PC, the initial D/A set value is not very accurate, on the other hand, the analog The circuit has drift), and the long-term stability is not good. For this reason, we use 89C52 for digital target control. The purpose is that even if the initial D/A digital setting value is not very accurate, the difference is 1℃~2℃. It is judged by the program and automatically Adjust the digital quantity sent from 89C52 to D/A to make the final control temperature reach the given value of the upper PC. Experiments have found that the temperature stability can even be improved if the algorithm is reasonable. Taking the above measures, the temperature stability is less than ±O. 15°C (if the algorithm is improved, there should be a possibility of improvement, this has been observed in the experiment, especially if the temperature control range is not large), the temperature offset (temperature accuracy) is less than ±O. 5°C. Figure 8 and Figure 9, Table 1 and Table 2 are the experimental results.

5 Conclusion

A bipolar current driver with an output of ±24V/20A (measured ±25V/17.5A) is designed to control the heating and cooling module composed of TEC in the instrument, and 89C52 is used for PID temperature control, the maximum heating and cooling speed is “l℃ /S, temperature stability “±O. 15°C, temperature deviation “±O. 5℃, the temperature control range is 45℃~100℃. Practice has proved that it is feasible to use UC3638 to form a bipolar current drive controller for TEC temperature control.

The Links:   DMF682AN CM150DC1-24NFM

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *