Application Report SBEA001 - SEPTEMBER 2001

OPTOELECTRONICS CIRCUIT COLLECTION

By Neil Albaugh

The following collection of analog circuits may be useful in electro- applications such as optical networking systems. This page summarizes their salient characteristics.

AVALANCHE BIAS SUPPLY 1 DRIVERÐ1 Provides an output voltage of 0V to +80V for reverse biasing A single-ended voltageÐcontrolled current source is shown an avalanche photodiode to control its . This circuit can here. This circuit operates from a single +3.3V supply and it also be reconfigured to supply a 0V to Ð80V output. can drive from 0A to 2A into a with a 0V to 2V input from a Digital-to-Analog (D/A) converter. Its input LINEAR TEC DRIVERÐ1 can be also be configured for a 0V to Ð2V input voltage. This is a bridge-tied load (BTL) linear for driving a thermoelectric cooler (TEC). It operates on a single +5V LASER DIODE DRIVERÐ2 supply and can drive ±2A into a common TEC. Similar to the previous circuit except that it operates from ±5V supplies and it is inverting; i.e., drives from 0A to 2A LINEAR TEC DRIVERÐ2 into a laser diode with a 0V to Ð2V input. A very low noise bipolar input op amp allows this circuit to achieve a low This is very similar to DRIVERÐ1 but its power output stage noise output current, an important consideration in dense was modified to operate from a single +3.3V supply in order channel spacing systems. As drawn, neither terminal of the to increase its efficiency. Driving this amplifier from a laser diode is grounded. standard +2.5V referenced signal causes the output transis- tors to have unequal power dissipation. TEMPERATURE UNDERÐAND OVERÐRANGE SENSING WITH A WINDOW COMPARATOR LINEAR TEC DRIVERÐ3 This circuit is useful to monitor the temperature of a TEC This BTL TEC driver power output stage achieves very high , to sense an out-of-range condition, or if the efficiency by swinging very close to its supply rails, ±2.5V. thermistor is shorted or open. Current sources make the This driver can also drive ±2A into a common TEC. Opera- threshold adjustments non-iterative. The supply voltage is tion is shown with the power output stage operating on +5V. ±1.5V supplies. Under these conditions, this linear amplifier can achieve very high efficiency.

www.ti.com AVALANCHE PHOTODIODE parallel with RF. In addition, low-pass filtering can be accomplished by adding a passive RC low-pass filter (LPF) BIAS SUPPLYÐ1 on the DAC output, the OPA445 output, or both. Low noise An avalanche photodiode (APD) is commonly used in opti- on the bias supply is important, noise on an APD bias supply cal detector circuits that require high sensitivity and wide will feed through into the following stageÐusually a bandwidth. A high reverse bias voltage across the photo- transimpedance amplifier. diode junction creates avalanche gain, and varying the re- verse bias voltage can control this gain. Although some 80 APDs require a bias of a few hundred volts, many InGaAs and Si APDs require only 60V to 80V. VOUT vs VIN The circuit shown in Figure 1 can provide a positive bias 60 voltage of up to +80V to an APD. The 0V to +2V input control voltage can be a DAC output or from an analog 40 source. The OPA445 high-voltage op amp is rated to operate with up Output (V) to ±45V supplies and it can provide up to ±15mA if its 20 power dissipation limits are observed. To obtain a high positive output voltage from this op amp, it can be operated 0 from unequal supplies as long as the voltage difference 0 0.5 1.0 1.5 2.0 between the supplies is 90V or less and the amplifier’s VIN (V) common-mode input voltage stays within its specified range. To allow the OPA445’s output voltage to swing to zero, a FIGURE 2. APD Bias Supply DC Output versus Input Voltage. negative supply of Ð5V was chosen. Staying within the 90V supply voltage difference specification, a +85V positive supply was chosen. This allows the OPA445 output voltage 1000 to swing up to +80V.

RIN and RF set the gain of this noninverting op amp circuit to 40V/V. As illustrated in Figure 2 an input of 0V to +2V 100 results in an output of 0V to +80V. Other gains can be calculated by the equation:   10 RF Av =   + 1 Output (mV)  RIN 

Other op amps can be used in this circuit if lower output 1 voltages are desired. An OPA551 or OPA552 can be used 1 10 100 1k 10k 100k 1M for a 0V to +50V bias supply if its positive supply voltage Frequency (Hz) is reduced to +55V. Bandwidth of this circuit is about 60kHz, as seen in FIGURE 3. APD Bias Supply Small Signal Bandwidth. Figure 3. To reduce noise, a can be placed in

Ð V+ 85V + C2 1nF OPA445 U1 0V to +2V Input = 0V to +80V Output 7 R Simplified APD Model 3 1 SERIES 50kΩ V+ 6 2 VÐ C1 5 VIN R 4 1nF IN R C 11.3kΩ SHUNT SHUNT 100kΩ 10pF RF 442kΩ

Ð VÐ 5V +

FIGURE 1. Positive Bias Supply Circuit Diagram.

2 SBEA001 The circuit’s transient response in Figure 4 shows a clean step response with no peaking or overshoot. 100 The OPA445 is now offered in a SO-8 surfaceÐmount 0V to 2V 1kHz Square Wave Input package, so adding a programmable APD bias supply into a 80 corner of a larger PCB layout is now feasible.

This bias supply circuit can easily be reconfigured to supply 60 a 0V to –80V output by changing the op amp’s supplies to Output Voltage +5V and Ð85V. A 0V to Ð2V input is required, but a 0V to 40 +2V input can be used if the input is connected to R and IN Output (V) the op amp noninverting input is grounded. This changes the 20 op amp to an inverting configuration. Note that the circuit’s input impedance is now lower (equal to R ) if the OPA445 IN 0 is used as an inverter. The familiar inverting op amp gain 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 equation is: Time (ms)   = RF Av  Ð  FIGURE 4. APD Bias Supply Transient Response.  RIN 

measuring the voltage drop across a shunt (R ) LASER DIODE DRIVERÐ1 SHUNT in series with the laser diode. The output current is con- The voltage-controlled current source circuit shown in Fig- trolled by the input voltage (VIN) that may be from an analog ure 5 can be used to drive a constant current into a signal or voltage source or from a voltage-output DAC. As shown, the pump laser diode. This simple linear driver provides a scale factor is 1V input equals 1A output. cleaner drive current into a laser diode than switching PWM The scale factor (V /I ) can be set to other values by drivers. IN OUT choosing appropriate resistor values using the equation: The basic circuit is that of a Howland current pump with a current booster (Q ) on the output of a R-R CMOS OPA350 VIN = R3 1 ¥ RSHUNT and, R1 = R3, R2 = R4 op amp (U2). Laser diode current is sensed by differentially IOUT R4

R 2 + 10kΩ V1 3.3V Ð C1 120pF

INVERTING INPUT R 1 U2 100kΩ 7 2 V+

6 Q1 OPA350 3 FTZ869

4 VÐ R3 100kΩ VIN

RSHUNT 1VIN = 1AOUT 0.1Ω R4 as shown 10kΩ

LASER DIODE NOTE: Bypass are not shown.

FIGURE 5. Laser Diode ConstantÐCurrent DriverÐ1, Circuit Diagram.

3 SBEA001 A PÐSpice simulation (DC sweep) was performed on VIN, sponse on both the rising and falling edges of the pulse. sweeping the input voltage from 0V to 2V. The lower curve, Figure 9 also shows clean pulse response operating on a 5V shown in Figure 6, shows the relationship of the laser diode supply. current to the input voltage. Power dissipation of Q1 is shown in the upper curve. 2.5

3.0 2.0 VIN = 2V

2.0 1.5 VIN = 1.5V (W)

D 1.0 P (A) 1.0 OUT 0 I VIN = 1V

0.5 2.0 VIN = 0.5V

0 1.0 0 100 200 300 400 500 (A) Time (µs) OUT OUT I

0 FIGURE 8. Laser Driver Transient Response with +3.3V 0 0.5 1.0 1.5 2.0 Supply. VIN (V)

FIGURE 6. Output Current and Q1 Power Dissipation ver- 2.5 sus Input Voltage with 3.3V Supply. 2.0 VIN = 2V Operating on a supply voltage of 3.3V, Q1 dissipates only 1.5W at an output current of 1A. This is well within the 1.5 V = 1.5V capability of the FTZ , as its SOT-223 package can IN (A) dissipate heat into the copper traces on a PC board. 1.0 OUT I VIN = 1V Similar curves are shown in Figure 7 for operation on a 5V supply voltage. This clearly shows the power advantage in 0.5 V = 0.5V operating the current source on a low supply voltage. The IN higher 5V supply voltage is advantageous if a higher output 0 compliance voltage is required. A laser diode macromodel 0 100 200 300 400 500 µ was unavailable, so a laser diode junction was simulated by Time ( s) a series connection of three . FIGURE 9. Laser Driver Transient Response with +5V Supply.

6.0 Changing the value of the shunt resistor or the scale factor will necessitate changing the compensation capacitor C1. 4.0 Verify that your circuit is stable before connecting an expen- sive laser diode to the output. (W)

D 2.0 P If a negative output control voltage is available, it can be

0 applied to R1 and R3 is then tied to ground. The amplifier is 2.0 then configured as an inverting amplifier.

The power booster used for Q1 is a very high gain single NPN 1.0 transistor, and is not a Darlington, it has a Beta of over 300 (A) at a collector current of 1A, allowing the CMOS OPA350 op OUT I amp to easily drive it to high currents. Zetex rates its continu- 0 ous collector current as 6.5A but SOA limits are reached 0 0.5 1.0 1.5 2.0 before approaching this current. VIN (V) If a unidirectional output current is acceptable, this circuit FIGURE 7. Output Current and Q1 Power Dissipation versus can be used to drive a TEC for cooling a laser diode or an Input Voltage with 5V Supply. APD. Adding a mechanical or a low on-resistance H-bridge will allow the TEC polarity to be switched and The PÐSpice Probe output of a transient response simulation is changed from heating to cooling. shown in Figure 8. The input voltage pulse amplitude was Satisfactory operation of this circuit should be verified in stepped from 10mV to 500mV to 10mV to 2V, the current your actual application by breadboarding and testing. output waveform was plotted. The circuit displays clean re-

4 SBEA001 LASER DIODE DRIVERÐ2 3.0 The voltage-controlled current source circuit shown in Figure 10 can be used to drive a very low-noise constant 2.0 current into a signal or pump laser diode. This simple linear (W) D 1.0 driver provides a far cleaner drive current into a laser diode P than a switching PWM driver can achieve. 0

The basic circuit is that of an NPN transistor current booster 2.0 LASER CURRENT (Q1) on the output of (U1), a very low noise bipolar op amp OPA227. Laser diode current is sensed by measuring the 1.0 voltage drop across a shunt resistor (RSHUNT) in the emitter (A) OUT of Q1. The output (laser) current is controlled by the input I voltage (VIN) that may be from an analog voltage source or 0 from a voltage-output DAC. As shown, the scale factor is Ð2.0 Ð1.5 Ð1.0 Ð0.5 0 Ð1V input equals 1A output. VIN (V) The scale factor (V /I ) can be set to other values by IN OUT FIGURE 11.Output Current and Q Power Dissipation Ver- choosing appropriate resistor values using the equation: 1 sus Input Voltage with 3.3V Supply. V R IN = 1 ¥R I R SHUNT OUT 2 ance voltage is required. Op amp U1 supplies can be from ±5V to ±15V. A PÐSpice simulation (DC sweep) was performed on VIN, sweeping the input voltage from 0V to Ð2V. The lower curve See Figure 12 for the PÐSpice Probe output of a transient shown in Figure 11 shows the relationship of the laser diode response simulation. The input voltage pulse amplitude was current to the input voltage. Power dissipation of Q1 is stepped from Ð10mV to Ð500mV to Ð10mV to Ð2V, the shown in the upper curve. Operating on a supply voltage of current output waveform was plotted. The circuit displays 3.3V, Q1 dissipates only 1.5W at an output current of 1A. clean response on both the rising and falling edges of the This is well within the capability of the FTZ851 transistor, pulse. The circuit simulation also exhibited clean pulse as its SOT-223 package can dissipate heat into the copper response with Q1 operating on a 5V supply. traces on a PC board. Changing the value of the shunt resistor or the scale factor

A similar DC sweep output current (IOUT) versus VIN curve will necessitate changing the compensation capacitor C1. is obtained with Q1 operating on a 5V supply. The higher 5V Verify that your circuit is stable before connecting an expen- supply voltage is advantageous if a higher output compli- sive laser diode to the output.

C 1 LASER 100pF DIODE + VS 3.3V Ð

U1 V+ 7 2 V+ VÐ 6 Q1 OPA227 3 FTZ 851 + 4 VÐ VÐ 5V VÐ Ð

R1 1kΩ

V+ R1 10kΩ R + NOTE: Bypass capacitors are not shown. SHUNT V+ 0.1Ω 5V Ð VIN

FIGURE 10. Laser Diode Constant-Current DriverÐ2 Circuit Diagram.

5 SBEA001 A laser diode macromodel was unavailable so a laser diode 2.5 junction was simulated by a series connection of three silicon diodes. This driver circuit requires both the 2.0 VIN = Ð2V and of the laser diode to be floating.

The power booster used for Q1 is a very high gain single 1.5 VIN = Ð1.5V NPN transistor, and is not a Darlington, it has a Beta of over

(A) 100 at a collector current of 2A, allowing the OPA227 op 1.0 OUT I VIN = Ð1V amp to easily drive it to high currents. Zetex rates its continuous collector current as 6A, but SOA limits are 0.5 reached before approaching this current. VIN = Ð0.5V Satisfactory operation of this circuit should be verified in 0 0 100 200 300 400 500 your actual application by breadboarding and testing. Time (µs)

FIGURE 12.Laser DriverÐ2 Transient Response with +3.3V Supply.

LINEAR TEC DRIVERÐ1 EMF” on current through the TEC under dynamic tempera- ture control conditions. The linear thermoelectric cooler (TEC) driver circuit is capable of driving ±2A into a TEC (see Figure 15). The 4.0 circuit operates on a single +5V supply and drives the TEC 1Ω TEC in the highly desirable “constant-current” mode. A bridge- 1.5Ω TEC 2Ω TEC tied load (BTL) amplifier topology achieves bidirectional 2.0 current output. This type of amplifier drives its load differ- entially, so the TEC must not be grounded on either end. An input offset of 1/2 supply voltage (in this case, an offset of 0.0 ±2.5V) is used to allow the to swing in both directions and to interface with a single-supply input voltage TEC Current (A) Ð2.0 source. This is represented in the circuit by VOS.

In Figure 15, voltage VIN is amplified by a R-R CMOS op amp Ð4.0 U1, with a class B power output stage (formed by the addition Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 of a complimentary power transistor pair Q1 and Q3) that drives VIN (V) one end of a TEC load through shunt resistor R4. A CMOS instrumentation amplifier (IA) U senses the voltage drop 3 FIGURE 13. TEC Current Versus Input Voltage VIN. across the shunt resistor and amplifies it by 50, it then feeds it back to the input of U . This feedback approach forces the 1 To determine the circuit’s power dissipation and the require- output TEC current to be a function of V . Shunt resistance and IN ments for heat sinking the SOT-223 output power transis- IA gain determines the scale factor of the circuit. tors, a simulation was run by sweeping the DC input voltage V as before, and using the same size TECs. The results are IN = A¥R V4 where AV is the IA gain in V/V shown in Figure 14. IOUT

A P-Spice simulation (DC sweep) was performed on VIN, sweeping the input voltage from Ð2.5V to +2.5V. This is 3.0 VOPA = +5V Ω equivalent to an input voltage of 0V to +5V from an external 1 TEC VS = +5V 1Ω TEC voltage source. 2.5

The P-Spice Probe output current to the TEC is shown in 2.0 Figure 13. TEC current is shown for three different sizes; 1.5Ω TEC 1.5Ω TEC

Ω Ω Ω (W) 1.5

1 , 1.5 , and a 2 TEC. When operating on a single +5V D supply, this driver is capable of driving a 1Ω or 1.5Ω TEC P 2Ω TEC 2Ω TEC to 2A. Output voltage compliance limits the 2Ω TEC current 1.0 to about 1.6A. 0.5 As Figure 13 illustrates, this TEC driver amplifier is a voltage-controlled current source. Constant-current drive 0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 assures that TEC drive current is independent of production V (V) variations in TEC junctions or long-term aging. Constant- IN current drive also eliminates the effect of thermal “back FIGURE 14. Output Transistor Power Dissipation.

6 SBEA001

3

R

10k

OS

V

2

3

2

U

2

C

10nF

4

7

OPA

Ð

V

V

V+

OPA353

are not shown. not are

3

6

4

470pF

C

, and U and ,

2

, Q ,

1

2

6

R

10k

R

15

S

V

4

2

Q

Q

FTZ968

FTZ869

(2) Bypass capacitors on Q on capacitors Bypass (2)

NOTES: (1) Optional: use one dual op amp OPA2353. amp op dual one use Optional: (1) NOTES:

OS

V

6

OPA

TEC

7V+

REF

V

INA155

Ð

V

4

200k

200k

22.2k

22.2k

5k

5k

Ð

3

IN

IN+

U

2

1

8

3

4

R

0.02

1

R

10k

3

1

FTZ968

Q

FTZ869

Q

S

V

OPA

5V

V

OPA

+

V

5

3

R

15

4.7nF

C

6

S

5V

V

1

C

Ð

S

10nF

V

V+

V

ÐÐ

+

OPA

4

7

V

OPA353

1

IN

3

2

OS

U

V

2.5V

V

Ð

+

OS V

FIGURE 15. TEC DriverÐ1 Circuit Diagram

7 SBEA001 Power dissipation of one NPN Q1 and one PNP Q3 power output transistor is shown in this sweep. The dissipation of 2.0 Ω the devices in the other half of the bridge (NPN = Q2 and 1 TEC PNP = Q4) will be the same, as shown in Figure 16. 1.5Ω TEC Depending on whether the TEC is in its cooling or heating 1.5 2Ω TEC mode, power is dissipated in Q1 and Q4 or in Q2 and Q3. Driver efficiency is usually a concern in large multi-channel 1.0 systems due to limitations on the total power dissipation of

the system. Linear amplifiers do not reach the efficiencies of TEC Current (A) PWM switching types, but they do offer important advan- 0.5 tages, principally their very low noise. Switching noise interference in laser and APD circuits is not a concern with 0 linear drivers. 100 1k 10k 100k 1M 10M 100M The DC simulation data was used to plot the efficiency of the Frequency (MHz) driver. To simplify the calculation, only the power output FIGURE 17. TEC Driver Amplifier Frequency Response. stage was considered. The CMOS OPA353 op amp power dissipation is only about 26mW, so deleting it contributes Some peaking is shown in the frequency response curves of the little error to the overall efficiency calculation. In this calcu- 1Ω and 1.5Ω TEC but increasing the capacitance of the com- lation, efficiency is considered to be the ratio of the power pensation capacitors C and C can eliminate this. Notice that delivered to the load TEC to the power supplied to the driver. 3 4 Ω the capacitance of C3 and C4 are not the same. Amplifier U1 For example, a current of 1A into a 1 TEC represents a requires more capacitance than U because of the presence of 2 2 power POUT of 1W dissipated in the load (POUT = I R). The feedback gain provided by the instrumentation amplifier, U3. power supplied to the driver PIN is 1A from the = 5V supply, or 5W (P = E ¥ I). Similar results are seen in Figure 18, a PÐSpice transient IN simulation of this TEC driver amplifier. Slight peaking is noted Therefore: for heavy loads, as predicted by the frequency response curves. P I¥R2 1A2 ¥1Ω Also evident in the transient response waveform is a small Eff(%) ===OUT ¥100, or Eff(%) ¥100 ¥%100= 20 crossover distortion “glitch” around 0A output current due to PIN V¥ISS 5V ¥1A the delay between turning off one transistor before its compli- Swinging the output voltage close to the supply rails mini- mentary transistor turns on. For example, the op amp driving mizes the voltage across the output transistor, thus reducing its the power must slew quickly between the voltage power dissipation. Likewise, it maximizes the voltage across at which the NPN transistor turns off and the PNP transistor the load. As can be seen from the equation above, this turns on. Due to the fact that this amplifier uses a Class B increases the efficiency of a linear driver. In fact, as seen in output stage, this region is 2VBE . A fast op amp, such as the Figure 16, this circuit can reach an efficiency of over 60% OPA353, minimizes the crossover distortion and enhances under favorable conditions. The discussion of TEC Drivers 2 stability about the crossover region. For applications such as and 3 investigates this further. an audio amplifier, the output stage transistors could be biased slightly into conduction (class AB1) which would eliminate the crossover region altogether. Applications for driving 100 loads such as a thermoelectric cooler do not warrant the V = +5V OPA increased complexity of a Class AB1 output stage. VS = +5V 80

60 1.5 1Ω TEC 1.5Ω TEC 1.0 40 Ω Efficiency (%) 2 TEC 2Ω TEC 0.5 1.5Ω TEC 20 1Ω TEC 0 0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 Ð0.5 TEC Current (A)

Ð1.0 FIGURE 16. Driver Efficiency with 1Ω, 1.5Ω, and 2Ω Loads.

Ð1.5 DriverÐamplifier loop stability was investigated by running 0246810 an AC and Transient simulation. Results of the AC simulation Time (ms) are shown in Figure 17. FIGURE 18. TEC Driver Transient Response.

8 SBEA001 LINEAR TEC DRIVERÐ2 While this is the optimum offset to allow the op amp to swing symmetrically to its supply rails, it creates a problem The linear TEC driver circuit (see Figure 21) is capable of for the output transistors. To swing optimally to its rails, it driving +1.5A and Ð1A into a TEC. The small-signal ampli- would be biased to 1/2VS to half its supply of only +3.3V or fier circuit operates on a single +5V supply and the power an offset of 1.65V. The 2.5V offset that is used creates an output stage operates on a single +3.3V supply to minimize its unbalance in the output transistor bridge, causing some power dissipation. This circuit drives the TE cooler in the problems as we shall see. highly desirable “constant-current” mode. A BTL amplifier As Figure 19 illustrates, this TEC driver amplifier is a topology achieves bi-directional current output. This type of voltage-controlled current source. Constant-current drive amplifier drives its load differentially, so the TEC must not be assures that TEC drive current is independent of production grounded on either end. variations in TEC junctions or long-term aging. Constant- An input offset of 1/2 the op amp supply voltage VOPA (in this current drive also eliminates the effect of thermal “back case, an offset of +2.5V) is used to allow the amplifiers to swing EMF” on current through the TEC under dynamic tempera- in both directions and to interface with a single-supply input ture control conditions. voltage source. This is represented in the circuit by V . OS To determine the circuit’s power dissipation and the require- In the schematic, voltage VIN is amplified by a R-R CMOS ments for heat sinking the SOT-223 output power transis- op amp U1 with a class B power output stage (formed by the tors, a simulation was run by sweeping the DC input voltage addition of a complimentary power transistor pair Q1 and as before, using the same sizes of TECs. The results are Q3) that drives one end of a TEC load through shunt resistor shown in Figure 20. R4. A CMOS INA155 instrumentation amplifier (IA) U3 senses the voltage drop across the shunt resistor and ampli- 3.0 fies it by 50; it then feeds it back to the input of U1. This VOPA = +5V feedback approach forces the output TEC current to be a V = +3.3V 2.5 S function of VIN. Shunt resistance and IA gain determines the scale factor of the circuit. 2.0 1Ω TEC VIN = A¥RV4 (W) 1.5 where AV is the IA gain in V/V. D IOUT P 1.5Ω TEC 1.0 A PÐSpice simulation (DC sweep) was performed on VIN, 2Ω TEC sweeping the input voltage from Ð2.5V to +2.5V. This is 0.5 equivalent to an input voltage of 0V to +5V from an external voltage source. 0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 The PÐSpice Probe output current to the TEC is shown in V (V) Figure 19. IN FIGURE 20. Output Transistor Q1 and Q3 Power Dissipation.

3.0 Power dissipation of one NPN (Q ) and one PNP (Q ) power 1Ω TEC VOPA = +5V 1 3 V = +3.3V output transistor is shown in this sweep. Note the difference in 1.5Ω TEC S 2.0 2Ω TEC power dissipation between the NPN and PNP transistor. This is a direct result of the unbalance caused by the offset voltage 1.0 problem previously noted. The dissipation of the devices in the other half of the bridge (NPN = Q2 and PNP = Q4) are similar 0 but a “mirror image” as seen in Figure 4. Depending on whether

TEC Current (A) the TEC is in its cooling or heating mode, power is primarily Ð1.0 dissipated in the PNP transistor Q3 or in Q4. The power dissipated in NPNs, Q1 and Q2, is relatively minor. Ð2.0 The effects of limited output compliance voltage are also Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 clearly seen in the curve of Figure 22. V (V) IN This compliance limit can also be seen in the power transis- FIGURE 19. TEC Current Versus Input Voltage VIN. tor output voltage driving the each end of the TEC load (see Figure 23). Note that the amplifier’s output voltage polarity TEC current is shown for three different sizes; 1Ω, 1.5Ω, crossover point is at the 2.5V input bias point. Biasing the and a 2Ω TEC. With the output transistors operating on a output transistors at their optimum 1.65V point would allow single +3.3V supply, this driver is capable of driving a 1Ω the output voltage to swing symmetrically to the supply TEC to over 2A or 2Ω TEC to over 1A in one direction but rails, +3.3V and ground. its output current capability is lower in the opposite direc- Driver efficiency was not simulated for this TEC driver tion. Output voltage compliance limits are asymmetrical due amplifier. to the 1/2VOPA offset voltage applied to the op amp input.

9 SBEA001 Ω

3

R

10k

OS

V

are not shown.

2

3

2

3

U

2

C

10nF

7

4

, and U

OPA

2

V

V+

OPA353

, Q

1

6

4

C 470pF

2

6

R

10k

R

15

(2) Bypass capacitors on Q

S

V

2

4

Q

Q

NOTES: (1) Optional: Use one dual op amp OPA2353

FTZ869

FTZ968

OS

V

6

OPA

TEC

7V+

REF

V

INA155

Ð

V

4

200k

200k

22.2k

22.2k

5k

5k

Ð

3

IN

IN+

U

3

8

1

2

4

R

0.02

1

R

10k

1

3

Q FTZ869

Q FTZ958

S

V

OPA

V 5V

OPA

+

V

5

3

R

15

C

4.7nF

6

S

V 5V

1

C

Ð

S

10nF

V+

V

V

ÐÐ

+

OPA

4

7

V

OPA353

1

IN

2

3

OS

U

V

V 2.5V

+

Ð

OS

V

FIGURE 21. TEC DriverÐ2 Circuit Diagram.

10 SBEA001 Linear amplifiers do not reach the efficiencies of PWM Also evident in the transient response waveform is a small switching types, but they do offer important advantages, crossover distortion “glitch” around 0A output current due principally their very low noise. Switching noise interfer- to the delay between turning off one transistor before its ence in laser and APD circuits is not a concern with linear complimentary transistor turns on. For example, the op amp drivers. driving the power transistors must slew quickly between the Driver amplifier loop stability was investigated by running voltage at which the NPN transistor turns off and the PNP an AC and Transient simulation. Results of the AC simula- transistor turns on. Due to the fact that this amplifier uses a tion are shown in Figure 24. Class B output stage, this region is 2VBE . A fast op amp such as the OPA353 minimizes the crossover distortion and Some peaking is shown in the frequency response of the 1Ω enhances stability about the crossover region. For applica- and 1.5Ω TEC but increasing the capacitance of the compen- tions such as an audio amplifier, the output stage transistors sation capacitors C and C can eliminate this. Notice that 3 4 could be biased slightly into conduction (class AB1) which the capacitance of C and C are not the same. Amplifier U 3 4 1 would eliminate the crossover region altogether. Applica- requires more capacitance than U because of the presence 2 tions for driving loads such as a thermoelectric cooler do not of feedback gain provided by the instrumentation amplifier, warrant the increased complexity of a Class AB1 output U3 . stage. Similar results are seen in Figure 25, a PÐSpice transient Due to the asymmetry of output stage of this driver, it may simulation of this TEC driver amplifier. Slight peaking is not be the best choice for a TEC driver amplifier. For better noted for heavy loads as predicted by the frequency response efficiency, see the last section of “Linear TEC Driver–3.” curves.

3.0 2.0 VOPA = +5V 1Ω TEC VS = +3.3V 2.5 1Ω TEC 1Ω TEC 1.5 Q4 2.0 1Ω TEC Ω 1.5 TEC 1.0 (W) 1.5 D Q4 P 2Ω TEC 1.0 Q TEC Current (A) 4 0.5 Q2 0.5

0 0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 100 1k 10k 100k 1M 10M 100M Frequency (mHz) VIN (V)

FIGURE 22. Output Transistor Q2 and Q4 Power Dissipation. FIGURE 24. TEC DriverÐ2 Amplifier Frequency Response.

3.5 2.0 VOPA = +5V V = +3.3V 3.0 Ω S Q and Q Q and Q 1 TEC 1 3 2 4 1.0 1.5Ω TEC 2.5 VOPA = +5V 2Ω TEC VS = +3.3V

(V) 2.0 0 (A) OUT OUT V 2Ω TEC I 1.5 Ð1.0 1.5Ω TEC 1.0 1Ω TEC

0.5 Ð2.0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 0246810 Time (ms) VIN (V)

FIGURE 23. TEC DriverÐ2 Amplifier Output Voltage Swing. FIGURE 25. TEC DriverÐ2 Amplifier Transient Response.

11 SBEA001 LINEAR TEC DRIVERÐ3 Current limiting in a voltage-controlled current source be- comes a simple matter of clamping the maximum input volt- The linear TEC driver circuit (see Figure 28) is capable of age to the amplifier. If driven by a rail-to-rail op amp, a voltage ± driving over 2A into a typical TEC. The circuit operates on divider or pot will set the current limit. When the R-R op amp ± a bipolar 2.5V supply and drives the TE cooler in the highly hits its rail, that voltage is divided down to an appropriate desirable “constant-current” mode. A BTL amplifier topology voltage that represents the maximum desired TEC current. The achieves bidirectional current output. This type of amplifier R-R op amp can’t swing past its rail, so this clamps the input drives its load differentially, so the TEC must not be grounded voltage to the TEC driver amplifier. on either end. To determine the circuit’s power dissipation and the require- With bipolar supplies, an input offset voltage is not necessary ments for heat sinking the SOT-223 output power transistors, to allow the amplifier outputs to swing in both directions. A a simulation was run by sweeping the DC input voltage as level shift circuit will be necessary to interface with a single- before using the same sizes of TECs. The results are shown in supply input voltage source that is biased to 1/2VCC. Figure 27. In the schematic, voltage VIN is amplified by a R-R CMOS op amp U1 with a class B power output stage (formed by the addition of a complimentary power transistor pair Q1 and Q3) 4.0 ± VOPA = 2.5V that drives one end of a TEC load through shunt resistor R4. A ± VS = 2.5V CMOS instrumentation amplifier (IA) U3 senses the voltage 3.0 drop across the shunt resistor and amplifies it by 50, it then feeds 1Ω TEC Ω it back to the input of U1. This feedback approach forces the 1 TEC Ω output (TEC) current to be a function of V . Shunt resistance 1.5 TEC

IN (W) 2.0 D and IA gain determines the scale factor of the circuit. P 2Ω TEC 1.5Ω TEC V 2Ω TEC IN = A¥R 1.0 V4, where AV is the IA gain in V/V. IOUT

A PÐSpice simulation (DC sweep) was performed on VIN, 0 sweeping the input voltage from Ð2.5V to +2.5V. The ampli- Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 fier output current to the TEC is shown in Figure 26. This is VIN (V) a PÐSpice Probe file output. FIGURE 27. Output Transistor Power Dissipation.

5.0 Power dissipation of one NPN (Q ) and one PNP (Q ) power ± 1 3 1Ω TEC VOPA = 2.5V ± output transistor is shown in this sweep. The dissipation of the 1.5Ω TEC VS = 2.5V 2Ω TEC devices in the other half of the bridge (NPN = Q and PNP = Q ) 2.5 2 4 will be the same as those shown in Figure 28. Depending on whether the TEC is in its cooling or heating mode, power is 0.0 dissipated in Q1 and Q4 or in Q2 and Q3. Driver efficiency is usually a concern in large multi-channel TEC Current (A) systems due to limitations on the total power dissipation of the Ð2.5 system. Linear amplifiers do not reach the efficiencies of PWM switching types but they do offer important advantages, Ð5.0 principally their very low noise. Switching noise interference Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 in laser and APD circuits is not a concern with linear drivers. V (V) IN The DC simulation data was used to plot the efficiency of the driver. To simplify the calculation, only the power output stage FIGURE 26. TEC Current Versus Input Voltage VIN. was considered. The CMOS OPA353 op amp power dissipa- TEC current is shown for three different TEC sizes; 1Ω, 1.5Ω, tion is only about 26mW, so deleting it contributes little error and a 2Ω TEC. Operating on a bipolar ±2.5V supply, this to the overall efficiency calculation. In this calculation, effi- driver is capable of driving a 1Ω or 1.5Ω TEC to 2A. Output ciency is considered to be the ratio of the power delivered to voltage compliance limits the 2Ω TEC current to about 1.6A. the load TEC to the power supplied to the driver. For example, Ω This output capability is the same as TEC DriverÐ1, essentially a current of 1A into a 1 TEC represents a power POUT of 1W 2 the same circuit operating on a single +5V supply. dissipated in the load (POUT = I R). The power supplied to the ± As Figure 26 illustrates, this TEC driver amplifier is a voltage- driver PIN is 1A from each 2.5V supply, or 5W controlled current source. Constant-current drive assures that (PIN = E ¥ I). TEC drive current is independent of production variations in Therefore: TEC junctions or long-term aging. Constant-current drive also P I¥R2 1A2 ¥1Ω eliminates the effect of thermal “back EMF” on current through Eff(%) ===OUT ¥100, or Eff(%) ¥100 ¥%100= 20 the TEC under dynamic temperature control conditions. PIN V¥ISS 5V ¥1A

12 SBEA001 Ω

3

R

10k

2

2

3

U

6

2

C

C

10nF

10nF

7

4

OPA

OPA

are not shown.

-V

3

+V

V+

V-

OPA353

6

4

, and U

C

470pF

4

, Q

1

2

6

R

10k

R

15

S

S

V

2

4

-V

Q

Q

(2) Bypass cpacitors on Q

FTZ869

FTZ968

NOTES: (1) Optional: Use one dual op amp OPA2353

6

OPA

TEC

7V+

REF

+V

INA155

Ð

V

OPA

4

V

200k

200k

Ð

22.2k 22.2k

5k

5k

Ð

3

IN

IN+

U

3

8

1

2

4

R

0.02

1

R

10k

OPA

1

3

-V -2.5V

Q Q FTZ869

FTZ968

S

S

-V

+V

OPA

V

+

Ð

OPA

+V +2.5V

5

3

R

OPA

15

C 4.7nF

+

+V

5

S

6

C

10nF

+V +2.5V

1

C

S

10nF

V+

V-

OPA

+V

OPA

+

4

7

V

+V

Ð

OPA353

1

S

IN

2

3

U

V

-V -2.5V

S

V

Ð

ÐÐ ÐÐ

+

FIGURE 28. TEC DriverÐ3 Circuit Diagram.

13 SBEA001 Swinging the output voltage close to the supply rails mini- Similar results are seen in Figure 31, a PÐSpice transient mizes the voltage across the output transistor, thus reducing simulation of this TEC driver amplifier. Slight peaking is its power dissipation. Likewise, it maximizes the voltage noted for heavy loads as predicted by the frequency response across the load. As can be seen from the previous equation, curves. this increases the efficiency of a linear driver. In fact, as seen in Figure 29, this circuit can reach an efficiency of over 60% under favorable conditions. The dissipation of TEC DriverÐ3 2.0 on ±2.5V is exactly the same as TEC DriverÐ1 on +5V. VOPA = +5V VS = +3.3V 1Ω TEC 1.0 1.5Ω TEC 100 VOPA = +5V 2Ω TEC V = +5V S 0

80 (A) 2Ω TEC OUT I 1.5Ω TEC 60 Ð1.0 1Ω TEC

40

Efficiency (%) Ð2.0 0246810 20 Time (ms)

0 FIGURE 31. TEC Driver Amplifier Transient Response. Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0

FIGURE 29. DriverÐ1 (+5V supply) and DriverÐ3 (±2.5V sup- Also evident in the transient response waveform is a small plies) efficiency with 1Ω, 1.5Ω and 2Ω TEC loads. crossover distortion “glitch” around 0A output current due to the delay between turning off one transistor before its complimentary transistor turns on. For example, the op amp Driver amplifier loop stability was investigated by running an driving the power transistors must slew quickly between the AC and Transient simulation. Results of the AC simulation voltage at which the NPN transistor turns off and the PNP are shown in Figure 30. transistor turns on. Due to the fact that this amplifier uses a Class B output stage, this region is 2VBE . A fast op amp, such as the OPA353, minimizes the crossover distortion and 2.0 V = ±2.5V enhances stability about the crossover region. For applica- OPA Ω ± 1 TEC VS = 2.5V tions such as an audio amplifier, the output stage transistors 1.5Ω TEC 1.5 could be biased slightly into conduction (class AB1) which would eliminate the crossover region altogether. Applica- 2Ω TEC tions for driving loads such as a thermoelectric cooler do not 1.0 warrant the increased complexity of a Class AB1 output stage.

TEC Current (A) ± 0.5 Reducing the power output transistor supplies from 2.5V to ±1.5V increases the overall efficiency of the driver amplifier by reducing the power dissipation in the output transistors. 0 As expected, lower transistor VCE results in lower power 100 1k 10k 100k 1M 10M 100M dissipation at the same collector current. Running the output Frequency (MHz) stage on as low voltage as possible increases efficiency— FIGURE 30. TEC Driver Amplifier Frequency Response. especially if the amplifier output voltage can swing close to the rail. Although the low-level circuits are operating on different Some peaking is shown in the frequency response curves of supply voltages (±2.5V) than the output stage (±1.5V), this the 1Ω and 1.5Ω TEC but increasing the capacitance of the does not cause the same asymmetry problems as discussed compensation capacitors C and C can eliminate this. No- 3 4 in “TEC Driver–2.” This is because the inputs are ground- tice that the capacitance of C and C are not the same. 3 4 referenced as are all of the supplies; both the op amp and the Amplifier U requires more capacitance than U because of 1 2 output stage can swing symmetrically both positive and the presence of feedback gain, that is provided by the negative. instrumentation amplifier, U3. See Figure 32 for output current into three sizes of TECs.

14 SBEA001 Smaller transistor heat sinks are required and reductions in 4.0 1Ω TEC ± cooling capacity are possible in large systems. VOPA = 2.5V Ω ± 1.5 TEC VS = 1.5V Figure 34 shows the very high efficiency that can be achieved 2Ω TEC 2.0 by a linear TEC driver when it swings very close to its supply rails. Driving a 2Ω TEC to its maximum current of about 1.3A, this amplifier achieves an efficiency of about 0 90%. Driving a 1Ω TEC to its maximum current of about 1.3A, this amplifier achieves an efficiency of about 80%. TEC Current (A) Ð2.0 Lower output currents achieve lower efficiencies but, at the same time, power dissipation is also lower.

Ð4.0 Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 V (V) 100 IN ± VOPA = 2.5V V = ±2.5V ± S FIGURE 32. TEC Driver Output Current with 1.5V Output 80 Stage Supplies.

60 Operation on low voltage supplies does sacrifice output 2Ω TEC voltage compliance somewhat. This reduces the maximum 40 drive current into a 2Ω or 1.5Ω TEC but lower resistance 1Ω Efficiency (%) 1.5Ω TEC

TECs are unaffected. 20 1Ω TEC Worthwhile gains in power dissipation are also realized by operating the output stage on ±1.5V supplies, as shown in 0 Figure 33. Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 FIGURE 34. Improved Amplifier Efficiency with Output Stage ± 1.5 Operating on 1.5V Supplies. ± VOPA = 2.5V ± VS = 1.5V The key to achieving good efficiency with a linear driver is 1Ω TEC 1.0 to match the TEC driver amplifier characteristics with ap- 1.5Ω TEC 2Ω TEC propriate power supplies for your TEC. Thermoelectric (W)

D coolers are available with a wide range of voltage and P current characteristics; it is important to choose a TEC that 0.5 requires a drive voltage that is very close to the voltage(s) available from your existing power supply. In new designs, it may be possible to chose a TEC and then specify a power 0 supply voltage that optimizes the performance of the TEC Ð3.0 Ð2.0 Ð1.0 0 1.0 2.0 3.0 and driver amplifier. V (V) IN Satisfactory operation of these circuits should be verified in FIGURE 33.Output Transistor Power Dissipation with ±1.5V your actual application by breadboarding and testing. Supplies.

15 SBEA001 TEMPERATURE UNDERÐAND The PÐSpice Probe output is shown in Figure 36. As the thermistor voltage crosses each threshold, U1 and U2 indi- OVERÐRANGE SENSING WITH cate their status with a TTL HIGH or LOW output. A WINDOW COMPARATOR The window comparator circuit shown in Figure 35 has been 6.0 used to monitor a TEC operating temperature and indicate U2 Output an out-of-range condition. Separate outputs are provided for (V) Temp Too Low indicating an over-temperature or under-temperature condi- 2 OUT Temp O.K. tion. A logic HI at the output of U1 or U2 indicates that the Ð1.0 thermistor temperature is outside of the set points of the window comparator. 6.0 U1 Output (V) Temp Too A logic LOW at both outputs indicates that the TEC tem- High 1 OUT

perature is within its safe operating range. A dual compara- U Temp O.K. tor with open-collector or drain outputs can be used in a Ð1.0 “wired-OR” configuration but the single OPA340 or dual 2.0 OPA2340 CMOS op amp offers higher accuracy. HI Threshold U1 and U2 Inputs

In the schematic, voltage VIN is used to simulate the voltage (V) U Delta IN LO Threshold Thermistor Voltage appearing across a thermistor that measures the laser/TEC V temperature. If a 10kΩ at 25°C thermistor is excited by 0 100µA from a REF200 Current Reference, it will read 0 0.5 1.0 1.5 2.0 V (V) 1.000V across the thermistor at 25°C. By using this con- IN stant-current source, the thermistor’s output voltage is a FIGURE 36. Input Voltage, HIGH and LOW Thresholds, and direct function of its temperature and this is easily converted Ouput Voltages. to °C by using the thermistor’s calibration chart.

A PÐSpice simulation (DC sweep) was performed on VIN, To prevent loading of the thermistor, a low bias current sweeping the thermistor voltage from 0V to 2V. The lower precision CMOS op amp was used as a comparator. Operat- Ω temp limit was set to 500mV with RLOW, a 5k resistor. A ing on a single +5V supply, the R/R op amp outputs are Ω 10k resistor, RDELTA, set the difference between the low TTL/CMOS compatible. A dual OPA2340 is ideal for lowÐ and high thresholds to 1V; therefore, the high threshold was speed window comparator applications. 500mV + 1.000V = 1.500V.

+5V +5V

U3 + +5V VS REF200 5V 100µA U Ð 1 7 V+ 2 6 OPA340

3 VÐ 4 RH 20kΩ HI = Temp Error: Temp Too High Thermistor +5V Output RDELTA U2 + 10kΩ 7 VIN V+ 1V 2 6 Ð OPA340

3 VÐ 4 RL R 20kΩ LOW HI = Temp Error: 5kΩ Temp Too Low

NOTES: (1) Optional: Use one dual op amp OPA2340

(2) Bypass capacitors on U1 through U3 are not shown.

FIGURE 35. Precision Window Comparator Circuit Diagram.

16 SBEA001 One cautionÐthe REF200 Current Reference requires about thermistor directly. This can monitor the temperature control 2.5V of headroom to operate properly. This means that the loop error and a second window comparator can monitor the upper threshold should not exceed 2.5V if this circuit is thermistor temperature as described above. operated on the recommended +5V supply. Ordinarily, accurately setting both thresholds of a window By setting the thresholds appropriately, a thermistor can also comparator is a tedious process. Voltage excitation of the be monitored for an open or shorted condition. commonly-used three-resistor divider guarantees interaction This circuit can also perform a continuous check of whether between thresholds. To adjust thresholds, first one pot is the temp control system is operating within specification. adjusted—then the other. The second adjustment changes Setting the window low temperature trip point to the lower the first threshold point, etc. Threshold trimming is an iterative process for the standard window comparator cir- desired operating temperature limit with RLOW and its upper temperature trip point to the upper desired operating tem- cuit. perature limit by RDELTA will allow this circuit to indicate By using current excitation, only two fixed or pots when the TEC is within the desired control range. A digital are needed. RLOW sets the lower threshold and RDELTA sets HIGH appears at the output of U1 or U2 if the temperature the “width” of the “window”. Adjusting the RDELTA does exceeds either the upper or lower preset error bounds. not change the lower threshold and adjusting the lower Additional sensitivity can be obtained by sensing an instru- threshold does not change the voltage difference between mentation amplifier (IA) output (this is usually the amplified thresholds. Thus the adjustment procedure is greatly simpli- error signal that is used by the control loop) instead of fied. connecting the window comparator directly to the ther- Noisy environments may require a small amount of hyster- mistor. Since the IA usually has gain, a larger signal is esis () to prevent “chatter” on the outputs available to drive the comparator. In this case, the compara- at the comparator switching points. tor is monitoring the amplified difference between the ther- Satisfactory operation of this circuit should be verified in mistor voltage and the temp set voltage rather than the your actual application by breadboarding and testing.

17 SBEA001 IMPORTANT NOTICE

Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, modifications, enhancements, improvements, and other changes to its products and services at any time and to discontinue any product or service without notice. Customers should obtain the latest relevant information before placing orders and should verify that such information is current and complete. All products are sold subject to TI’s terms and conditions of sale supplied at the time of order acknowledgment.

TI warrants performance of its hardware products to the specifications applicable at the time of sale in accordance with TI’s standard warranty. Testing and other quality control techniques are used to the extent TI deems necessary to support this warranty. Except where mandated by government requirements, testing of all parameters of each product is not necessarily performed.

TI assumes no liability for applications assistance or customer product design. Customers are responsible for their products and applications using TI components. To minimize the risks associated with customer products and applications, customers should provide adequate design and operating safeguards.

TI does not warrant or represent that any license, either express or implied, is granted under any TI patent right, copyright, mask work right, or other TI intellectual property right relating to any combination, machine, or process in which TI products or services are used. Information published by TI regarding third–party products or services does not constitute a license from TI to use such products or services or a warranty or endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual property of the third party, or a license from TI under the patents or other intellectual property of TI.

Reproduction of information in TI data books or data sheets is permissible only if reproduction is without alteration and is accompanied by all associated warranties, conditions, limitations, and notices. Reproduction of this information with alteration is an unfair and deceptive business practice. TI is not responsible or liable for such altered documentation.

Resale of TI products or services with statements different from or beyond the parameters stated by TI for that product or service voids all express and any implied warranties for the associated TI product or service and is an unfair and deceptive business practice. TI is not responsible or liable for any such statements.

Mailing Address:

Texas Instruments Post Office Box 655303 Dallas, Texas 75265

Copyright  2001, Texas Instruments Incorporated