Physics 160 Lecture 13 R. Johnson May 11, 2015 JFET Amplifiers Common-Source Amp Source Follower Note that Zout does not depend on the output impedance of the driving voltage source. Ci Cin VDD Generally the performance is poor compared with a BJT amp, except for the input impedance. For a JFET it is spectacularly high! This can often be an overwhelming reason to use a FET. VDD JFET Cin RD JFET 1 gm RG RS RG can be a very large resistance! 1 Z out RS gm May 6, 2015 RS CS RG Gain gm RD Physics 160 2 JFET Amplifiers • Typically, the place you want to use a JFET amplifier is where you need very high input impedance, for example because your signal source has a very high impedance (cannot deliver much current). – Probably best used in differential amps or source-followers, not common-source amps. • Another example is where the base current of a bipolar transistor will cause a significant g error. – The LF411 Op-amp that you will soon use in several circuits uses JFETs at its inputs. This is very nice, because the current flowing into the inputs is negligible in all cases. • Otherwise, bipolar transistors will usually give much better performance in terms of gain, predictability, etc. May 11, 2015 Physics 160 3 JFET Amplifiers • You need to keep in mind: – The gate-source junction must be reverse biased at all times (or at worst, zero volts). – The gate does need some bias current, although it can be very small. – The drain-to-source voltage g cannot be too small, especially p y if the drain current is substantial. For a bipolar transistor the collectoremitter voltage can be a fraction of a volt, but for a JFET count on a few volts. – Unlike U lik th the case off a bi bipolar l ttransistor, i t th there iis no simple i l fformula l ffor the transconductance. You must consult the data sheet, and it will be small compared with the bipolar transistor transconductance. • Or equivalently, equivalently the effective dynamic resistance of the source will be a few hundred ohms, not the bipolar transistor value of 25 divided by the current in mA. May 11, 2015 Physics 160 4 Very high impedance source is no problem (except at high frequency) JFET Source Follower JFET has rather high output impedance, so a BJT follower is useful to boost the output drive. Driving source 0V R2 1.382pA V2 1.382pA 15Vdc VS 6.401mA 10Meg J1 J2N5485 1Vac 0Vdc 4.011mA +0.5 V Q1 2.390mA V 13.89uA Q2N3904 -2.404mA 0.1 01V C2 12uF V3 15Vdc 6.401mA May 11, 2015 Voltage division between Zout of the JFET (1/gm) and RS.results in gain less than unity 3.997mA R1 RS Physics 160 3.9k V 2.404mA R3 6.2k RL 1Meg 0A 5 Setting the Bias Current Essentially the same procedure as used for the JFET current source. 15.5V R 3 .9 k 4 A 4mA May 11, 2015 Physics 160 6 JFET Follower Gain G May 11, 2015 1 250 gm RS g m 3.9 4 0.94 1 RS g m 1 3.9 4 Physics 160 7 Improving Follower Performance 4.274mA R2 1.380pA 0V V2 1.380pA 15Vdc VS 6 654mA 6.654mA J1 10Meg J2N5485 1Vac 0Vd 0Vdc V +0.5 V 13.83uA Q1 2 380mA 2.380mA Q2N3904 -2.394mA 0.1 0.1 V 12uF 4.260mA 15 V V3 15Vdc C2 J2 2.394mA -1.464pA J2N5485 R3 14 14.5 5V 6.654mA R1 6.2k RL 1Meg 0A 125 Replace the source resistor by a JFET (or BJT) current source. May 11, 2015 Physics 160 8 Reminder from Lecture 12: 15.00V 6.869mA J3 V1 15Vdc -1.053pA 1 053pA J2N5485 IDSS 4.254mA J3 V1 -1.381pA 1 381 A J2N5485 15Vdc 531.8mV 6.869mA 4.254mA R1 125 R May 11, 2015 Physics 160 0. 5V 125 4mA 9 Gain with current source = 0.99! The lack of high frequency performance is related to the high driving source impedance. We will see how to improve this with a “b t t ” “bootstrap”. May 11, 2015 Physics 160 10 Output Impedance Zout is independent of source resistance!! JFET Zin>>10 Meg 4.274mA R2 1.380pA V2 15Vdc 1.380pA VS 6.654mA 10Meg J1 JFET Zout=1/gm=250 ohms J2N5485 1Vac 0Vdc V Q1 13.83uA BJT : re 25 / 2.38 10.5 2.380mA Q2N3904 BJT Zin=22.5170=3800 ohms BJT Zout=250/170 + 25/2.4=12 -2.394mA C2 100uF 4.260mA J2 V 2.394mA V3 -1.464pA J2N5485 R3 15Vdc 6.2k 6.654mA R1 RL 12 0A 12 ohm load ( t (extreme case)) 125 2380/13 83 172 =2380/13.83=172 May 11, 2015 Physics 160 11 Voltage Division of Impedances 3800 0.95 3800 200 JFET Output BJT Output May 11, 2015 0.95 Physics 160 12 0.48 11.7 12 12 Bootstrapping the Drain of the Follower Make the drain follow the gate to minimize the voltage changes between Gate and Drain. This almost eliminates the low low-pass pass filter formed by the source impedance and the G-S capacitance. High pass filter CGS Low pass filter +7.8 V 0V +0.5 V 4 mA 0.1 V 15 V This trick can greatly increase the response at high frequency. 14.5 V Don’t make RD too much larger or else the VDS will become too small. May 11, 2015 Physics 160 13 No Bootstrap on Drain The low low-pass pass filter formed by the source impedance and gate-drain parasitic iti capacitance kills the gain above about 10 kHz 10 Meg-ohm Source Impedance 1 k-ohm load Note that the gain is less than unity above because of charge division between the output impedance of the BJT emitter emitter-follower follower and the 1 kk-ohm ohm load resistor. May 11, 2015 Physics 160 14 With Bootstrapped Drain 10 Meg-ohm Source Impedance 1 k-ohm load Bandwidth is now nearly 1 MHz Note that the gain is less than unity above because of charge division between the output impedance of the BJT emitter emitter-follower follower and the 1 kk-ohm ohm load resistor. May 11, 2015 Physics 160 15 Bootstrapped Drain with Heavy Load Roll-off due to AC coupling to the load 10 Meg-ohm Source Impedance 100 ohm load Voltage division between Zout and the 100 ohm load has reduced the gain by about 10%. The frequency response has moved down from 1 MHz to 100 kHz With the load greatly increased (100 ohm load resistor), then the gain is so much less than unity that the bootstrap no longer works very well well. We lose a decade of frequency response. May 11, 2015 Physics 160 16 AC Coupled Input The gate doesn’t need much current, but about 1pA still has to flow, so a bias resistor is essential! The capacitor can be very small, since the bias resistor (input impedance) is large. Zin is now completely dominated by the bias resistor ((well above the 3dB point). May 11, 2015 Physics 160 17 Bias resistor lowered Zin to 100 M Note the 10% loss of gain g due to voltage division between the 10 M source impedance and the 100 M input p impedance. May 11, 2015 Physics 160 18 AC Coupling with Bias Bootstrap We can bootstrap the bias network to raise it to even higher impedance at signal frequencies frequencies, just as for a BJT follower. Only a small, inexpensive capacitor is needed, since the bias network impedance is so high! May 11, 2015 R10 at signal frequencies looks like about 600 M. Physics 160 19 Input impedance with bootstrap 0.983 V Voltage at JFET gate Reffective May 11, 2015 10 M 588 M 1 - 0.983 Physics 160 20 DC-coupled follower with nearly zero offset By using a matched JFET pair (must be on the same IC chip) and matched resistors R1 and R4, we can achieve nearly zero DC offset between the input and output output. 15.00V 4.254mA R2 1.381pA V2 15Vdc 10Meg VS 1Vac 0Vdc J1 J2N5485 13.81uV 0V 531 8mV 531.8mV R4 0V 125 0V 13.69uV 0V J2 -1.381pA V3 V J2N5485 13.69pA RL 1Meg 15Vdc -14.47V 4.254mA We do pay the price that now the JFET output impedance of 1/gm is in series with a 125 ohm resistor. R1 0V 125 0V -15.00V May 11, 2015 Physics 160 21 Output of matched follower There is no bootstrap of the drain, hence the poor frequency response with high source impedance. A BJT emitter follower could be added to drive a bootstrap if higher frequency response is needed (the JFET output impedance is too high to drive a bootstrap by itself effectively). May 11, 2015 Physics 160 22 JFET Common-Source Amplifier Gain is gm times RD: Looks like ~2.3 mmho from graph, predicting G=2.37.5=17 1.015mA RD 7.5k Observe only ~15 from PSpice A BJT would have gm~1/25 mho, giving a gain of ~300! C1 RVS V2 10u J1 J2N5485 15Vdc VS 1Meg -1.215pA 1Vac 0Vdc V 1.015mA RS C2 1.5k 100u RL 1Meg 0A Without t out bypassing bypass g RS S with t capac capacitor to C2 the gain will be less than unity! May 11, 2015 Physics 160 23 JFET Common-Source Amplifier Only the high input impedance is impressive. Better to use the FET as a follower and follow it b a BJT amp. by (Note: when FETs are used as voltage amplifiers, usually the load on the drain is a current source rather th th than a resistor). i t ) S See HW 4 4. May 11, 2015 Physics 160 24 JFET Variable Resistor R1 INPUT Voltage Divider OUTPUT 1.000V 433.2mV 1k V R3 2.717uA V 1Meg V3 V1 = 0 V2 = .5 5 TD = 0 TR = 0.5m TF = 0.5m PW = 0.001m 1.000V PER = 1.002m V4 R2 -2.283V 2.283V 1Meg 564.0uA J1 J2N3819 2.717uA -1.082pA p -5.0Vdc 1Vdc 566.8uA V1 A lower voltage in V1 increases the current through the 1 Meg resistors and lowers the gate voltage of J1, causing its resistance to increase, thus increasing the voltage at the OUTPUT. 0V A circuit that you will study in the lab. V1 controls the gate bias and hence the resistance of the JFET (at low drain current). The JFET then forms a voltage divider together with R1. It looks quite linear as long as we stay well below saturation and don’t have too large a voltage swing. R3 provides some negative feedback that helps stabilize the output. May 11, 2015 Physics 160 25 JFET Non-Saturation Region Resistance is inverse of the slope May 11, 2015 Physics 160 26 Divider output with V1=-5V In= green O t red Out= d May 11, 2015 Physics 160 27 Divider output with V1=-7V The output is slightly curved (nonlinearity) May 11, 2015 In= green Out= red Physics 160 28 Op-Amps and Feedback • DC-coupled differential amp – Amplify from DC up to some limiting frequency • Very high differential gain 8 – To a first approximation, just consider the gain to be infinite 3 + + power V+ Non-inverting input - 4 Inverting input 2 1 Output V- OUT power Without an external feedback network, the transistors do not have a stable bias point. They are either in saturation or turned off, and the op-amp output is at one limit or the other. This can be useful as a comparator. The first operational amplifiers were constructed with vacuum tubes, and later with discrete t transistors. i t Expensive! E i ! To use the op-amp as a linear device, an external feedback network, with negative feedback, is essential! May 11, 2015 Physics 160 29 Ideal Op-Amp Such an ideal op-amp of course does not exist, but a first analysis of op op-amp amp circuits can be done to a good approximation usually by ignoring the non-ideal behavior. May 11, 2015 Physics 160 30 3 V+ Vin In 8 Non-Inverting Amplifier + 2 - 4 Virtual Vin Note: Zin is very large (transistor base or gate). R2 90k R1 10k Out 1 V- OUT Vin Vout Identical to the Lecture Lecture10 example, except that the Op Amp is a highperformance IC. R1 R1 R2 0 Simplified analysis of an op-amp with negative feedback: - Assume infinite gain, so the negative feedback always has to keep the two inputs equal in order to have a finite output. - Assume zero current flow into the op-amp inputs. - Then calculate the relationship between input and output from the feedback network. Vout R2 G 1 1 9 10 Vin R1 May 11, 2015 Physics 160 31