← About PHYS 252

PHYS 252 · Electronics

Lecture 06 · Transistor bias & the CE amplifier

1 Where Lab 6 Sits

Lab 6 has two halves. First you bias a 2N3904 with a voltage divider, as in Lab 5, and check that it sits in the active region. Then you add three capacitors and a signal, and the biased transistor becomes a common-emitter (CE) amplifier. Last, you connect loads to its output and watch the gain change.

Lab 6 in two rows. Bias the transistor: Task 1, DMM pin-to-pin readings of the 2N3904 and 2N3906, then Task 2, divider bias at 5 V: design, measure every current and voltage, find beta dc. Make it amplify: Task 3, a CE amplifier with an input below 50 mVpp at 2 to 20 kHz and an open-load gain above 40, then Task 4, adding C3 and loads of 10 kilohms, 2 kilohms and 300 ohms, with three scope channels at once.
Figure 1. Lab 6 in two rows: bias the transistor (Tasks 1 and 2), then make it amplify (Tasks 3 and 4).

1.1 Task map

Lab 6 taskWhat you doConcept & section
1DMM readings between every pair of pins of a 2N3904 and a 2N3906The junction check (§2.1, and the Lecture 05 notes)
2Design a divider bias at 5 V; measure every current and voltage; find βdc\beta_{dc}The bias rules (§2)
3Build the CE amplifier; measure amplitude, DC offset, phase and the open-load gainCapacitors (§3), the amplifier (§4), gain (§5), input (§6), measuring (§9)
4Add C3C_3 and three loads; show vinv_{in}, vCv_C and vloadv_{load} togetherLoading (§7), swing (§8), measuring (§9)

The examples here use one circuit from start to end: VCCV_{CC} = 9 V, R1R_1 = 39 kΩ, R2R_2 = 8.2 kΩ, RCR_C = 3.9 kΩ, RER_E = 1.0 kΩ and a 2N3904. Lecture 05 designs it step by step. Your lab circuit runs from 5 V, so every number in it is yours to work out. Every task needs an error analysis; Section 10 collects the main error sources.

2 The Bias, in Brief

2.1 Task 1: the junction check

Task 1 is the same DMM table as Lab 5 Task 1. The Lecture 05 notes explain the method and what each pair of leads should show for an NPN and a PNP transistor. You need the pinout to read your table, so here it is again.

From the onsemi 2N3903 and 2N3904 datasheet: the NPN symbol with pin numbers, collector 3, base 2 and emitter 1, and drawings of the TO-92 package with straight leads and with bent leads, the leads numbered 1, 2, 3.
Figure 2. The 2N3904 symbol and the TO-92 package: pin 1 emitter, pin 2 base, pin 3 collector. Source: onsemi 2N3903/D datasheet, Rev. 9, p. 1.

2.2 Five rules for divider bias

Task 2 gives you the design rules. Each one has a reason, and knowing the reason tells you how hard to push it.

RuleShared circuit at 9 V (β = 150)Why
VE≈0.1 VCCV_E \approx 0.1\,V_{CC}0.83 V (aim 0.9 V)IE=VE/REI_E = V_E/R_E. When VEV_E is much larger than the changes in VBEV_{BE} from part to part and with temperature, those changes barely move IEI_E.
VCE≈0.5 VCCV_{CE} \approx 0.5\,V_{CC}4.97 VThe collector can swing up and down from the middle.
R1∥R2<0.1 βdcRER_1 \parallel R_2 < 0.1\,\beta_{dc} R_E6.78 kΩ < 7 kΩ with β = 70The divider is firm: the base current barely loads it, so β barely moves VBV_B.
IR1<ICI_{R1} < I_C, better <0.1 IC< 0.1\,I_C192 µA = 0.23 ICI_CDivider current does no work; it only drains the supply.
every R>300R > 300 Ωsmallest is 1.0 kΩKeeps every current, and the power in each part, small.

The firm-divider rule uses βdc, and βdc varies a lot from part to part. The onsemi datasheet guarantees only minimums at low current: hFE≥70h_{FE} \geq 70 at 1 mA, and 100 to 300 at 10 mA. Design with the minimum at your current. Then any part from the bag works.

2.3 Two rules pull apart

The firm rule wants a small R1∥R2R_1 \parallel R_2, which means small resistors. The current rule wants a small IR1I_{R1}, which means large resistors. You can find how much β you need to meet both.

Worked example: the current gain both rules need at 9 V

Call k=VB/VCCk = V_B/V_{CC}, the fraction the divider passes. Then R2/(R1+R2)=kR_2/(R_1+R_2) = k, and

R1∥R2=R1R2R1+R2=k(1−k)(R1+R2)R_1 \parallel R_2 = \frac{R_1 R_2}{R_1 + R_2} = k(1-k)(R_1 + R_2)

Ignoring IBI_B, the divider current is IR1=VCC/(R1+R2)I_{R1} = V_{CC}/(R_1+R_2). With IC≈VE/REI_C \approx V_E/R_E and VE=0.1 VCCV_E = 0.1\,V_{CC}, the rule IR1<x ICI_{R1} < x\,I_C becomes R1+R2>10 RE/xR_1 + R_2 > 10\,R_E/x. Put that into the firm rule R1∥R2<0.1 βRER_1 \parallel R_2 < 0.1\,\beta R_E:

β>k(1−k)0.01 x\beta > \frac{k(1-k)}{0.01\,x}

At 9 V, VEV_E = 0.9 V and VBV_B = 1.6 V, so kk = 0.178 and k(1−k)k(1-k) = 0.146. With xx = 1 (IR1<ICI_{R1} < I_C), β must exceed 14.6: easy. With xx = 0.1 (IR1<0.1 ICI_{R1} < 0.1\,I_C), β must exceed 146, which is above the datasheet minimum of 70.

Log-log plot of the beta a firm divider needs against the divider-current limit x, where I R1 is below x times I C, for V CC 9 V, V E 0.9 V, V B 1.6 V. The line falls from about 290 at x 0.05 to 14.6 at x 1; at x 0.1 it needs beta above 146. A dashed line marks the datasheet minimum h FE of 70 at 1 mA. A square marks the 39 k and 8.2 k divider at I R1 = 0.23 I C, firm for beta above 68. The region above the line is shaded: both rules met.
Figure 3. The β a firm divider needs against the divider-current limit, at 9 V. Above the line both rules hold. The shared circuit (square) draws IR1I_{R1} = 0.23 ICI_C and is firm for β above 68.

So the "better" half of the current rule often has to give. The shared circuit keeps the firm rule for the worst-case β and accepts a divider current of 0.23 ICI_C. Work out kk for your own 5 V design and decide the same way; say in your report which rule you relaxed and why.

2.4 The Q point against β

The base sees the divider as a Thévenin source: VTH=VCCR2/(R1+R2)V_{TH} = V_{CC} R_2/(R_1+R_2) behind RTH=R1∥R2R_{TH} = R_1 \parallel R_2. The base current flows through RTHR_{TH}, and the emitter current through RER_E, so

IE=VTH−VBERE+RTH/(β+1)I_E = \dfrac{V_{TH} - V_{BE}}{R_E + R_{TH}/(\beta+1)}(1)
Worked example: the shared circuit for three current gains

VTH=9×8.2/47.2=1.564V_{TH} = 9 \times 8.2/47.2 = 1.564 V and RTHR_{TH} = 6.775 kΩ. With VBEV_{BE} = 0.7 V:

βIEI_E (mA)ICI_C (mA)IBI_B (µA)VEV_E (V)VCV_C (V)VCEV_{CE} (V)
700.7880.77711.10.7885.975.18
1500.82650.8215.470.8265.804.97
3000.84450.8422.810.8455.724.87
ideal divider0.8640.86400.8645.634.77

A fourfold change in β moves ICI_C by 8 %. The divider current is IR1=(9−1.526)/39 kΩI_{R1} = (9 - 1.526)/39\ \text{k}\Omega = 192 µA at β = 150, which is 0.23 ICI_C: it meets IR1<ICI_{R1} < I_C and not IR1<0.1 ICI_{R1} < 0.1\,I_C, as §2.3 predicted.

Two panels. Left: simulated 2N3904 collector curves for base currents of 2 to 12 microamps, I C against V CE from 0 to 9 V, with the DC load line from 1.84 mA at V CE 0 to 9 V at zero current. The hand-calculated Q points for beta 70, 150 and 300 sit almost on top of each other near V CE 5 V and 0.8 mA. Right: I C against beta from 30 to 500. Divider bias stays nearly flat (0.777, 0.821 and 0.842 mA at beta 70, 150, 300), while base bias set for the same current at beta 150 rises in proportion to beta.
Figure 4. Left: simulated 2N3904 collector curves (onsemi model) with the DC load line of the shared circuit; the hand-calculated Q points for β = 70, 150 and 300 sit almost on top of each other. Right: ICI_C against β for divider bias, and for base bias set to the same current at β = 150.

The DC load line joins IC=VCC/(RC+RE)I_C = V_{CC}/(R_C + R_E) = 1.84 mA at VCEV_{CE} = 0 and VCEV_{CE} = 9 V at ICI_C = 0. Every Q point of this circuit lies on it. To check that the transistor is active, not saturated, compare VCEV_{CE} with the datasheet's VCE(sat)≤0.2V_{CE(sat)} \leq 0.2 V (at 10 mA, IBI_B = 1 mA). At 4.97 V the shared circuit is far from it.

The simulated curves come from onsemi's model, whose βdc is about 114 near 0.9 mA. That is one particular part; yours will differ, and Task 2 asks you to measure it.

Practice 2

(a) Design for VCCV_{CC} = 15 V with RER_E = 2.2 kΩ and VE=0.1 VCCV_E = 0.1\,V_{CC}. Find IEI_E, the RCR_C that puts VCEV_{CE} near 7.5 V, and the largest R1∥R2R_1 \parallel R_2 that is firm with β = 70.

(b) For the design in (a), find kk and the β that the firm rule and IR1<0.1 ICI_{R1} < 0.1\,I_C need together.

(c) For the shared 9 V circuit, find ICI_C and VCEV_{CE} with β = 100.

3 Coupling and Bypass Capacitors

3.1 A capacitor passes AC and blocks DC

A capacitor's opposition to AC is its reactance,

XC=12πfCX_C = \dfrac{1}{2\pi f C}(2)

As ff rises, XCX_C falls, and at high enough frequency the capacitor acts as a short. As ff falls, XCX_C rises; at DC (ff = 0) no current flows through it at all, so it is an open circuit. The amplifier uses both ends of this: the capacitors pass the signal and block the bias voltages.

Log-log plot of capacitive reactance X C equals 1 over 2 pi f C against frequency from 10 Hz to 100 kHz for 1, 10 and 100 microfarads: three parallel falling lines. At 1 kHz they read 159, 15.9 and 1.59 ohms. A note says low f gives large X C (an open circuit at DC) and high f gives small X C (a short for AC).
Figure 5. XCX_C against frequency for 1, 10 and 100 µF. At 1 kHz: 159, 15.9 and 1.59 Ω.

3.2 The coupling capacitor is a high-pass filter

C1C_1 joins the generator to the base. The base sits at a DC voltage set by the divider, and the generator sits at 0 V DC. Wired straight together, the generator would pull the base away from its bias. In series, C1C_1 blocks the DC and lets the signal through. Seen from the signal, C1C_1 and the resistance after it form a high-pass filter:

fc=12πRCf_c = \dfrac{1}{2\pi R C}(3)

where RR is the whole resistance in the loop: for C1C_1, the generator's RGR_G plus the stage's input impedance ZinZ_{in} (§6). At fcf_c the signal is down to 70.7 %. Well above fcf_c it passes almost whole.

A source v in drives a series capacitor C into a resistor R to ground; v out is taken across R. Notes: at high f, X C is small and v out is about v in; at DC, X C is open and v out is 0.
Figure 6. A series capacitor and a resistor to ground: a high-pass filter. At high frequency vout≈vinv_{out} \approx v_{in}; at DC voutv_{out} = 0.

C3C_3, between the collector and the load, does the same job at the output. The collector sits near 5.8 V DC, and the load should see only the signal.

3.3 The bypass capacitor grounds the emitter for AC

C2C_2 sits across RER_E. Seen alone, a resistor feeding a capacitor to ground is a low-pass filter: at high frequency the capacitor shorts the output to ground. At the emitter, that is the point. RER_E stays in the circuit for DC and holds the bias steady, while for the signal the emitter is tied to ground. Section 5.4 shows what happens to the gain without it.

Two circuits. Left: a source v in, a series resistor R, and a capacitor C from the output to ground: at high frequency C shorts v out to ground. Right: a transistor emitter with R E to ground and the electrolytic capacitor C2 across it: for AC, v E is held at its DC value.
Figure 7. Left: the low-pass view, a series RR and a shunt CC. Right: the same idea at the emitter. C2C_2 across RER_E holds vEv_E at its DC value.

3.4 Sizing the capacitors

A simple rule makes each capacitor a short at the lowest frequency you care about: make XCX_C less than a tenth of the resistance in its loop at that frequency. That puts the corner frequency a decade below the band.

Worked example: sizing C1, C2 and C3 for 1 kHz

Take 1 kHz as the lowest frequency, a 50 Ω generator and, for the worst case, a 470 Ω load.

C1C_1. The loop is RG+ZinR_G + Z_{in} = 50 + 2717 = 2767 Ω. XC<276.7X_C < 276.7 Ω at 1 kHz needs C1>1/(2π×1000×276.7)C_1 > 1/(2\pi \times 1000 \times 276.7) = 0.58 µF. With 1 µF, fcf_c = 57.5 Hz; with 10 µF, 5.75 Hz.

C2C_2. The emitter sees re′r'_e (§5), 30.25 Ω. XC<3.025X_C < 3.025 Ω needs C2>52.6C_2 > 52.6 µF. More exactly, the emitter sees RE∥(re′+(RTH∥RG)/(β+1))R_E \parallel (r'_e + (R_{TH} \parallel R_G)/(\beta+1)) = 29.7 Ω, so 100 µF gives a corner at 53.6 Hz and 470 µF at 11.4 Hz.

C3C_3. The loop is RC+RLR_C + R_L = 3900 + 470 = 4370 Ω. XC<437X_C < 437 Ω needs C3>0.36C_3 > 0.36 µF. With 1 µF, fcf_c = 36.4 Hz; with 10 µF, 3.64 Hz.

C2C_2 is by far the largest, because the resistance it must beat is the smallest. The simulations in this unit use C1C_1 = 10 µF, C2C_2 = 470 µF and C3C_3 = 10 µF. For your lab, size them for the lowest frequency of the band Task 3 gives.

3.5 Electrolytic capacitors have a + lead

Values of 10 µF and up are usually aluminium electrolytic capacitors. They are polarized: the + lead must sit at the higher DC voltage, and the voltage across the part must stay below its rating. Put one in backwards and it leaks, heats up and can fail. The case carries a stripe of − signs next to the negative lead, and on a new part the + lead is the longer one.

A purple radial aluminium electrolytic capacitor marked 50 V, 100 microfarads, with a light stripe of minus signs along one side marking the negative lead.
Figure 8. A 100 µF, 50 V aluminium electrolytic capacitor; the light stripe marks the negative lead. Photo: Willtron, CC BY-SA 3.0.

In the shared circuit at β = 150: C1C_1 has 0 V on the generator side and 1.53 V on the base side, so + goes to the base. C2C_2 has 0.83 V on the emitter and 0 V on ground, so + goes to the emitter. C3C_3 has 5.80 V on the collector and 0 V on the load, so + goes to the collector. A 16 V rating covers all three in a 9 V circuit.

Practice 3

(a) A stage has ZinZ_{in} = 1.5 kΩ and is driven from a 50 Ω generator. The lowest frequency is 200 Hz. What is the smallest C1C_1 by the XC<0.1RX_C < 0.1R rule? What is fcf_c with the next larger standard value of 1, 2.2, 4.7 or 10 µF?

(b) The same stage has re′r'_e = 20 Ω. Find the smallest C2C_2 for 200 Hz.

(c) Why does a value of C2C_2 that is too small hurt the gain at low frequency, but not the DC bias?

4 The Common-Emitter Amplifier

4.1 The circuit

Add three capacitors to the biased transistor and it becomes an amplifier. C1C_1 couples the signal into the base, C2C_2 bypasses RER_E, and C3C_3 couples the collector to the load. The emitter is tied to AC ground by C2C_2, so for the signal it is common to the input and the output. That gives the circuit its name.

The common-emitter amplifier. A generator v g of 10 mVpp at 1 kHz with R G 50 ohms feeds v in, then C1 10 microfarads into the base node v B. The divider R1 39 kilohms from V CC 9 V and R2 8.2 kilohms to ground biases the base of a 2N3904. R C 3.9 kilohms runs from V CC to the collector v C; R E 1.0 kilohm runs from the emitter v E to ground, bypassed by C2 470 microfarads. C3 10 microfarads couples the collector to the load R L 47 kilohms at v out. The electrolytic capacitors have their plus side toward the base, the emitter and the collector.
Figure 9. The CE amplifier on the shared bias circuit: a 10 mVpp, 1 kHz generator with RGR_G = 50 Ω, C1C_1 = 10 µF, C2C_2 = 470 µF, C3C_3 = 10 µF and a 47 kΩ load.

4.2 DC plus AC at every node

Every node in the amplifier carries a DC voltage from the bias and, on top of it, the signal. Figure 10 shows both at each node, from a simulation with onsemi's 2N3904 model.

Simulated waveforms over two periods at five nodes, stacked. v in: 9.85 mVpp around 0 V. v B: the same 9.85 mVpp riding on 1.51 V DC. v E: flat at 0.90 V, no AC. v C: 0.78 Vpp around 5.54 V, upside down relative to v B. v out: the same 0.78 Vpp centered on 0 V.
Figure 10. Simulated node voltages over two periods. vinv_{in}: 9.85 mVpp on 0 V. vBv_B: the same signal on 1.51 V. vEv_E: 0.90 V and no AC. vCv_C: 0.78 Vpp on 5.54 V. voutv_{out}: 0.78 Vpp on 0 V.

Read it from top to bottom. The signal passes C1C_1 and rides on the base's DC, so vBv_B is the input on top of 1.51 V. C2C_2 holds the emitter still, so the whole input sits across the base-emitter junction. The junction turns that small voltage into a change of collector current, and RCR_C turns the current back into a large voltage at the collector. C3C_3 then removes the 5.54 V DC, so the output is centered on zero.

The simulated DC values differ a little from the hand values of §2.4 (VBV_B 1.51 against 1.53 V, VCV_C 5.54 against 5.80 V). The model's VBEV_{BE} is 0.615 V instead of 0.7 V and its βdc is 114, so it runs at 0.888 mA. The same spread shows up between your own calculation and your breadboard.

4.3 The output is inverted

When the base voltage rises, more collector current flows. More current means a larger drop across RCR_C, so the collector voltage falls. The output is upside down compared with the input: a phase shift of 180°.

Simulated v in (left axis, millivolts) and v out (right axis, volts) over two periods: when v in reaches its peak, v out is at its trough. v in 9.85 mVpp, v out 0.78 Vpp with R L 47 kilohms, shifted 179 degrees.
Figure 11. Simulated vinv_{in} and voutv_{out} on separate scales. When the input peaks, the output is at its trough. The shift is 179° at 1 kHz.

The last degree comes from the coupling and bypass capacitors, whose corners lie between 0.3 and 11 Hz. Each adds a small phase lead well above its corner. At lower frequencies the shift moves further from 180°.

4.4 The AC equivalent circuit

To find the gain, keep only what the signal sees. Two steps turn the amplifier into its AC equivalent:

  1. Short every large capacitor. Above their corners, XCX_C is negligible.
  2. Short the DC supply to ground. A good supply holds its voltage, so it has no AC voltage on it.
The AC equivalent circuit: v g with R G drives v in at the base. R1 and R2 both go from the base to ground. The emitter goes straight to ground. The collector v C connects R C and R L, both to ground, and v out. Notes: V CC to ground, C1, C2 and C3 shorted; r c equals R C in parallel with R L.
Figure 12. The AC equivalent: R1R_1 and R2R_2 both run from the base to ground, the emitter is grounded, and RCR_C and RLR_L are in parallel at the collector.

With VCCV_{CC} at AC ground, R1R_1 joins R2R_2 from the base to ground, and RCR_C joins RLR_L from the collector to ground. Their parallel value is the AC collector resistance, rc=RC∥RLr_c = R_C \parallel R_L.

Practice 4

(a) Sketch vBv_B, vEv_E and vCv_C for the shared circuit with C2C_2 removed. Which node gains an AC signal it did not have before?

(b) Draw the AC equivalent of an amplifier whose emitter has two resistors in series, with C2C_2 across only the lower one.

5 r′e and the Voltage Gain

5.1 One current through two resistors

For small signals, the base-emitter junction acts like a small resistor, re′r'_e, in the emitter. The transistor then works as a current source that copies the emitter current into the collector. This picture is the T model.

The T model of the transistor: v in drives the top of r prime e, which goes to ground and carries i e. A current source i c, about equal to i e, flows from the output node down into the same point. The output node v out connects to r c to ground. Equations: v in = i e r prime e, v out = minus i c r c, A v = i c r c over i e r prime e = r c over r prime e.
Figure 13. The T model. The input sits across re′r'_e; the collector current, almost equal to the emitter current, flows through rcr_c.

With the emitter at AC ground, the whole input is across re′r'_e: vin=iere′v_{in} = i_e r'_e. The same current, less the tiny base current, flows through rcr_c: vout=−icrcv_{out} = -i_c r_c. Since ic≈iei_c \approx i_e,

Av=voutvin=icrciere′=rcre′A_v = \dfrac{v_{out}}{v_{in}} = \dfrac{i_c r_c}{i_e r'_e} = \dfrac{r_c}{r'_e}(4)

The minus sign is the 180° inversion; gains are quoted as magnitudes.

5.2 Where the 25 mV comes from

The emitter current of a forward-biased junction rises exponentially with its voltage:

IE=IS eVBE/VT,VT=kTqI_E = I_S\, e^{V_{BE}/V_T}, \qquad V_T = \frac{kT}{q}

VTV_T is the thermal voltage: 25.7 mV at 25 °C. A small change of VBEV_{BE} changes IEI_E along the slope of this curve, and the slope of an exponential is the function divided by its scale:

dIEdVBE=IEVT⇒re′=dVBEdIE=VTIE≈25 mVIE\frac{dI_E}{dV_{BE}} = \frac{I_E}{V_T} \qquad\Rightarrow\qquad r'_e = \frac{dV_{BE}}{dI_E} = \frac{V_T}{I_E} \approx \frac{25\ \text{mV}}{I_E}

So re′r'_e is not a part you can buy. It is the slope of the junction curve at the Q point, and it depends only on the current through it: twice the current, half the re′r'_e. Books round VTV_T to 25 or 26 mV; the difference is 4 %.

The exponential junction curve I E = I S e to the V BE over V T with V T 25 mV, from 0.62 to 0.735 V. Two dashed tangents: at 0.826 mA the slope gives r prime e = 25 mV over I E = 30.2 ohms; at 0.413 mA it gives 60.5 ohms, a flatter tangent.
Figure 14. The junction curve with VTV_T = 25 mV. At IEI_E = 0.826 mA the tangent gives re′r'_e = 30.2 Ω; at half that current the curve is flatter and re′r'_e = 60.5 Ω.

5.3 The gain of the shared circuit

Worked example: the open-load gain

With β = 150, IEI_E = 0.8265 mA (§2.4), so re′r'_e = 25 mV / 0.8265 mA = 30.25 Ω. With no load, rc=RCr_c = R_C = 3.9 kΩ:

Av=390030.25=128.9A_v = \frac{3900}{30.25} = 128.9

With 26 mV instead, re′r'_e = 31.46 Ω and AvA_v = 124.0. A 10 mVpp input that reaches the base as 9.82 mVpp (§6.3) gives 1.27 Vpp at the collector.

Put re′=25 mV/IEr'_e = 25\ \text{mV}/I_E into Equation (4) with no load, and use IC≈IEI_C \approx I_E:

Av=RCIE25 mV≈VRC25 mVA_v = \dfrac{R_C I_E}{25\ \text{mV}} \approx \dfrac{V_{RC}}{25\ \text{mV}}(5)

The gain of a fully bypassed stage depends only on the DC voltage across RCR_C. In the shared circuit VRCV_{RC} = 3.20 V, so Av≈A_v \approx 128, matching the worked example. With the bias rules of §2.2, VRCV_{RC} is about VCC−0.1 VCC−0.5 VCC=0.4 VCCV_{CC} - 0.1\,V_{CC} - 0.5\,V_{CC} = 0.4\,V_{CC}: 3.6 V at 9 V, a ceiling near 144. Work out the same estimate for your 5 V design before you build it, and compare it with the gain Task 3 asks for.

Two panels against collector current from 0 to 1.8 mA. Left: A v approximately V RC over 25 mV, a straight line with R C 3.9 kilohms; the point at I C 0.821 mA has V RC 3.20 V and A v about 128. Right: the largest output peak. Room up, I C R C, rises; room down, V CE minus 0.2 V with R E 1.0 kilohm and V CC 9 V, falls. They are equal at 1.00 mA; a dashed line marks 0.82 mA.
Figure 15. Left: Av≈VRC/25A_v \approx V_{RC}/25 mV against ICI_C with RCR_C = 3.9 kΩ. Right: the largest output peak. The room up, ICRCI_C R_C, grows with current; the room down, VCE−0.2V_{CE} - 0.2 V, shrinks. They are equal at 1.00 mA.

You cannot raise the gain for free. More current means more drop across RCR_C, so VCEV_{CE} shrinks and the output has less room to swing down before the transistor saturates (§8).

5.4 Without the bypass capacitor

Take C2C_2 out and RER_E is back in the emitter for AC. The input now sits across re′r'_e and RER_E in series, and

Av=rcre′+REA_v = \dfrac{r_c}{r'_e + R_E}(6)
Worked example: the shared circuit without C2

Open load: Av=3900/(30.25+1000)=3.79A_v = 3900/(30.25 + 1000) = 3.79. With the 47 kΩ load, rcr_c = 3601 Ω and AvA_v = 3.50.

Simulated v out for the same 10 mVpp input with R L 47 kilohms: with C2 the output is 0.78 Vpp, a gain of 79; without C2 it is 34 mVpp, a gain of 3.4.
Figure 16. Simulated output for the same 10 mVpp input and a 47 kΩ load: 0.78 Vpp (gain 79) with C2C_2, 34 mVpp (gain 3.4) without.

The low gain has a use. It hardly depends on re′r'_e, so it barely changes with temperature or with the signal's size, and the output is less distorted. Many amplifiers leave part of RER_E unbypassed for this reason. In Lab 6 the bypassed stage is the one to build.

5.5 The model and the formula disagree

The simulated gain with the 47 kΩ load is 79, while Equation (4) gives 3601/30.25 = 119. The difference is in the model. onsemi's 2N3904 model has an emission coefficient NF = 1.5, which makes its junction curve less steep than the ideal one: its current follows eVBE/(1.5 VT)e^{V_{BE}/(1.5\,V_T)}. At its Q point of 0.888 mA, the simulator reports gmg_m = 22.0 mS, so re′=1/gmr'_e = 1/g_m = 45.4 Ω, against 29.1 Ω for an ideal junction at the same current and 27 °C.

Simulated I C against V BE for the onsemi 2N3904 model at V CE 5 V, and the curve of an ideal junction (emission coefficient 1) through the same Q point at 0.888 mA. The ideal curve is steeper: r prime e = 25.9 mV over I C = 29 ohms. The model, with NF 1.5, gives r prime e = 1 over g m = 45 ohms.
Figure 17. Simulated ICI_C against VBEV_{BE} for the onsemi model at VCEV_{CE} = 5 V, and an ideal junction through the same Q point. The model's flatter curve gives re′r'_e = 45 Ω instead of 29 Ω.

A model is a fit to measurements over a wide range of currents, and this one's slope near 1 mA is not that of an ideal junction. re′=25 mV/IEr'_e = 25\ \text{mV}/I_E is itself an approximation for an ideal junction. Treat both as estimates: report the formula's gain and the simulated gain next to your measured gain, and say which one your transistor is closer to.

Practice 5

(a) A bypassed stage runs at IEI_E = 1.5 mA with RCR_C = 2.2 kΩ. Find re′r'_e and the open-load gain, with 25 mV and with 26 mV.

(b) The same stage with RER_E = 680 Ω left unbypassed: what is the gain?

(c) Using Equation (5), estimate the open-load gain of a fully bypassed stage at VCCV_{CC} = 12 V that follows the rules of §2.2.

6 Input Impedance and Source Loading

6.1 Two current gains: DC and small-signal

βdc is the ratio of the DC currents, IC/IBI_C/I_B. For signals, what matters is the ratio of the small changes,

βac=icib\beta_{ac} = \dfrac{i_c}{i_b}(7)

Datasheets call it hfeh_{fe}, with lowercase letters for small-signal values. It is close to βdc but not the same number, because ICI_C against IBI_B is not a straight line through the origin. Figure 18 gives the DC gain and Figure 19 the small-signal one. For the 2N3904, onsemi gives hFE≥70h_{FE} \geq 70 at 1 mA and hfeh_{fe} = 100 to 400 at 1 mA, 10 V and 1 kHz.

DC current gain rows from the onsemi 2N3903 and 2N3904 datasheet, at V_CE = 1.0 V. For the 2N3904, hFE is at least 40 at 0.1 mA, 70 at 1.0 mA, 100 to 300 at 10 mA, 60 at 50 mA and 30 at 100 mA; the 2N3903 rows are lower.
Figure 18. DC current gain hFEh_{FE} of the 2N3903 and 2N3904 at five collector currents, VCEV_{CE} = 1.0 V: for the 2N3904, at least 70 at 1 mA and 100 to 300 at 10 mA. Source: onsemi 2N3903/D datasheet, Rev. 9, p. 2.
Small-signal rows of the onsemi 2N3903 and 2N3904 datasheet, measured at I_C = 1.0 mA, V_CE = 10 V and 1.0 kHz, under the table header. Input impedance h_ie: 1.0 to 8.0 kilohms for the 2N3903 and 1.0 to 10 kilohms for the 2N3904. Voltage feedback ratio h_re: 0.1 to 5.0 and 0.5 to 8.0 times 10 to the minus 4. Small-signal current gain h_fe: 50 to 200 for the 2N3903 and 100 to 400 for the 2N3904.
Figure 19. Small-signal current gain hfeh_{fe} of the 2N3903 and 2N3904 at ICI_C = 1.0 mA, VCEV_{CE} = 10 V, ff = 1.0 kHz. Source: onsemi 2N3903/D datasheet, Rev. 9, p. 2.

6.2 The π model: the base sees βr′e

The T model puts re′r'_e in the emitter. Seen from the base, that resistor looks larger. The input voltage drives iei_e through re′r'_e, but the base supplies only ib=ie/(β+1)i_b = i_e/(\beta+1) of it:

vinib=ie re′ie/(β+1)=(β+1) re′≈β re′\frac{v_{in}}{i_b} = \frac{i_e\, r'_e}{i_e/(\beta + 1)} = (\beta + 1)\, r'_e \approx \beta\, r'_e

The π model draws exactly this: a resistor βre′\beta r'_e from the base to ground, and a current source β ib\beta\, i_b from the collector to ground. It gives the same gain as the T model and makes the input side easy to read.

The pi model of the stage, redrawn from the handwritten notes with the base resistor labeled beta r prime e. v g and R G drive v in. R1, R2 and beta r prime e, which carries i b, go from the input node to ground. Inside a dashed transistor box, a current source beta i b, labeled i c, flows from the output node to ground. R C and R L also go from the output node to ground; v out is at the top.
Figure 20. The π model of the stage: R1R_1, R2R_2 and βre′\beta r'_e from the input to ground; the current source β ib\beta\, i_b in parallel with RCR_C and RLR_L at the output.

6.3 The input impedance of the stage

The generator sees three resistances from the base to ground in parallel:

Zin(stage)=R1∥R2∥βre′Z_{in(stage)} = R_1 \parallel R_2 \parallel \beta r'_e(8)
Worked example: the input impedance of the shared circuit

With βac = 150, βre′\beta r'_e = 150 × 30.25 Ω = 4537 Ω, and R1∥R2R_1 \parallel R_2 = 6775 Ω. So

Zin=6775 Ω∥4537 Ω=2717 ΩZ_{in} = 6775\ \Omega \parallel 4537\ \Omega = 2717\ \Omega
Resistance in kilohms against beta ac from 50 to 450, with r prime e 30.25 ohms. Dashed: beta r prime e, a rising line. Dotted: R1 parallel R2 = 6.78 kilohms. Solid: Z in = R1 parallel R2 parallel beta r prime e, which rises slowly: 2.09 kilohms at beta 100, 2.72 at 150, 4.34 at 400. The datasheet h fe range of 100 to 400 is shaded.
Figure 21. ZinZ_{in} against βac for the shared circuit: 2.09 kΩ at β = 100, 2.72 kΩ at 150 and 4.34 kΩ at 400. The bias divider caps it at 6.78 kΩ.

Because βac is uncertain by a factor of four, so is βre′\beta r'_e, and ZinZ_{in} follows. Measure it if you need it: put a known resistor in series with the input and see how much of the signal reaches the base.

6.4 The source divides with the input impedance

A function generator is an ideal source vgv_g behind its output resistance RGR_G (50 Ω on most lab generators). RGR_G and ZinZ_{in} form a voltage divider:

vin=ZinRG+Zin vgv_{in} = \dfrac{Z_{in}}{R_G + Z_{in}}\, v_g(9)
A function generator drawn as v g behind R G inside a dashed box, connected at v in to the amplifier's input impedance Z in to ground.
Figure 22. The generator, vgv_g behind RGR_G, loaded by the amplifier's ZinZ_{in}.
Worked example: how much of 10 mVpp reaches the base

With RGR_G = 50 Ω: vinv_{in} = 2717/2767 × 10 mVpp = 9.82 mVpp, 98.2 %. With a 600 Ω source: 2717/3317 = 81.9 %, 8.19 mVpp. With the 50 Ω generator, the open-load output is 128.9 × 9.82 mV = 1.27 Vpp.

The gain AvA_v is defined from vinv_{in} at the amplifier, not from the number on the generator's display. Measure vinv_{in} with the scope, at the input of the amplifier, every time you measure voutv_{out}.

Practice 6

(a) For the shared circuit, find βre′\beta r'_e and ZinZ_{in} with βac = 250.

(b) A 1 kΩ source drives the shared circuit at βac = 150 with vgv_g = 5 mVpp. What reaches the base, and what is the open-load output?

(c) Why does raising R1R_1 and R2R_2 by the same factor raise ZinZ_{in}, and what does it cost the bias?

7 Loading the Output

7.1 A load sits in parallel with the collector resistor

C3C_3 connects the load to the collector. For DC it is open, so the load does not change the bias. For AC it is a short, so the load sits in parallel with RCR_C, and the AC collector resistance drops to

rc=RC∥RL=RCRLRC+RLr_c = R_C \parallel R_L = \dfrac{R_C R_L}{R_C + R_L}(10)
The output side for AC: a current source i c from v C to ground, R C from v C to ground, then C3 in series to v out and R L to ground. For AC, C3 is a short, so r c = R C parallel R L = R C R L over R C plus R L.
Figure 23. The output side for AC: the collector current source drives RCR_C, and through C3C_3 the load RLR_L.

7.2 Gain falls with the load

Worked example: four loads on the shared circuit

With re′r'_e = 30.25 Ω and 9.82 mVpp at the base:

RLR_Lrcr_cAvA_vvoutv_{out}
47 kΩ3601 Ω119.11.17 Vpp
4.7 kΩ2131 Ω70.5692 mVpp
470 Ω419 Ω13.9136 mVpp
100 Ω98 Ω3.231.7 mVpp
A v = R C parallel R L over r prime e against load resistance from 30 ohms to 1 megohm on a log axis, rising to the open-load value of 129. Points: 47 kilohms, r c 3601 ohms, A v 119.1, 1169 mVpp out; 4.7 kilohms, 2131 ohms, 70.5, 692 mVpp; 470 ohms, 419 ohms, 13.9, 136 mVpp; 100 ohms, 98 ohms, 3.2, 32 mVpp. Input 9.82 mVpp at the base. A dotted line at R L = R C marks half the gain.
Figure 24. Gain against load resistance for the shared circuit. A load equal to RCR_C halves the gain; small loads leave almost none.

A CE stage has an output resistance of about RCR_C: seen from the load, it is a source behind 3.9 kΩ, like the loaded divider of Lecture 01. A load much larger than RCR_C takes nearly the full output. A load much smaller than RCR_C shorts most of it. That is why a CE stage is often followed by a stage with a low output resistance when it has to drive a small load.

C3C_3 also forms a high-pass filter with RC+RLR_C + R_L. A smaller load lowers that resistance and raises the corner, so check C3C_3 against your smallest load (§3.4).

Practice 7

(a) Find rcr_c, AvA_v and voutv_{out} for the shared circuit with loads of 22 kΩ and 1 kΩ, for 9.82 mVpp at the base.

(b) With C3C_3 = 1 µF, find the corner frequency for each load in (a).

(c) Which load in (a) gives the most output power, vout(rms)2/RLv_{out(rms)}^2/R_L?

8 Swing, Clipping and Distortion

8.1 How far the collector can swing

The collector sits at its Q point and swings both ways. It cannot go up past the point where ici_c reaches zero (cutoff), and it cannot go down past the point where the transistor saturates. For a bypassed stage the emitter stays at VEV_E, so the collector can fall to about VE+VCE(sat)V_E + V_{CE(sat)}.

Worked example: swing limits of the shared circuit

Open load, β = 150: the collector can rise by ICRCI_C R_C = 0.821 mA × 3.9 kΩ = 3.20 V before ici_c = 0. It can fall by VCE−0.2V_{CE} - 0.2 V = 4.97 − 0.2 = 4.77 V before saturation. The smaller one, 3.20 V, limits a clean output to 6.4 Vpp. With a load, the upward limit becomes ICrcI_C r_c.

Simulated v C for a 100 mVpp generator signal with open load, over two periods. The output runs from 1.00 to 8.19 V around the Q point of 5.54 V. Its bottom is flat at about 1.0 V, near the V E plus 0.2 V = 1.1 V line, where the transistor saturates; its top is rounded and stays below the 9 V cutoff line. A dashed sine shows what a linear amplifier would give: from 1.64 to 9.43 V, past the 9 V line.
Figure 25. Simulated vCv_C for 100 mVpp from the generator, open load. The bottom flattens at about VEV_E + 0.2 V (saturation); the top rounds off below 9 V. Dashed: the output of a perfectly linear amplifier.

In Figure 25 the bottom is flat: the transistor saturates for part of each cycle. The top is rounded but never reaches 9 V. Before the transistor cuts off, the exponential junction has already squeezed that half of the wave (§8.2).

8.2 Distortion: the junction is not a straight line

re′r'_e is the slope at one point of an exponential curve. A large signal walks along the curve, where the slope changes. With vbev_{be} the AC voltage across the junction,

iC=IC evbe/VTi_C = I_C\, e^{v_{be}/V_T}

so equal positive and negative swings of vbev_{be} give unequal swings of current.

Worked example: three input sizes

With VTV_T = 25.7 mV, a sine of 2 mVpp at the base (±1 mV) makes iCi_C swing from −3.8 % to +4.0 %. At 10 mVpp: −17.7 % to +21.5 %. At 50 mVpp: −62 % to +164 %, far from a sine.

i C over I C during one period of a sine at the base, from i C = I C e to the v be over V T with V T 25.7 mV. 2 mVpp: plus 4.0 and minus 3.8 percent, almost symmetric. 10 mVpp: plus 21.5 and minus 17.7 percent. 50 mVpp: plus 164.6 and minus 62.2 percent, a tall peak and a shallow trough.
Figure 26. iC/ICi_C/I_C over one period for 2, 10 and 50 mVpp at the base. The larger the input, the taller the peak and the shallower the trough.

Because the circuit inverts, the tall current peaks pull the collector down, toward saturation, and the shallow troughs let it rise less. Even the 10 mVpp of the worked examples shows a visible difference between the two halves.

8.3 The small-signal rule

A common rule keeps the AC emitter current below a tenth of the DC current: ie(pp)<0.1 IEi_{e(pp)} < 0.1\,I_E. The voltage across the junction is iere′i_e r'_e, so

vin(pp)<0.1 IE re′=0.1×25 mV=2.5 mVv_{in(pp)} < 0.1\, I_E\, r'_e = 0.1 \times 25\ \text{mV} = 2.5\ \text{mV}

For a fully bypassed stage this holds whatever the bias, because IEre′I_E r'_e is always 25 mV. On the bench, start with a few millivolts at the base. Raise the input only while the top and bottom halves of vCv_C stay the same height. If they differ, the input is too large, and the gain you measure is not the small-signal gain.

Practice 8

(a) A bypassed stage has ICI_C = 1.2 mA, RCR_C = 3.3 kΩ and VCEV_{CE} = 4.0 V. Find the room up and down for the open-load output. Which limit sets the largest clean output?

(b) The same stage drives a 10 kΩ load. What is the room up now?

(c) Find the positive and negative swing of iCi_C, in percent, for 5 mVpp at the base.

9 Measuring with the Oscilloscope

9.1 Offset, amplitude and phase

Task 3 asks for three things about the output: its amplitude, its DC offset and its phase relative to the input. The scope's input coupling decides which you can see.

Two scope screens drawn from the worked example. Left, DC coupling at 2 V per division and 200 microseconds per division: v C sits 5.80 V, 2.9 divisions, above the center line, and its AC part is only 0.58 divisions tall. Right, AC coupling: CH1 at 5 mV per division shows 9.8 mVpp, 2.0 divisions; CH2 at 200 mV per division shows 1.17 Vpp, 5.8 divisions. Where CH1 peaks, CH2 is at a trough: 180 degrees.
Figure 27. Drawn from the worked example, not measured. Left: DC coupling at 2 V/div shows vCv_C 5.80 V above ground, with an AC part only 0.58 division tall. Right: AC coupling, CH1 (vinv_{in}, 9.8 mVpp) at 5 mV/div and CH2 (vCv_C, 1.17 Vpp) at 200 mV/div; CH1's peak lines up with CH2's trough.

DC offset. With DC coupling, the trace shows the whole voltage, DC plus AC. Read the offset as the height of the trace's center above the ground marker. At vCv_C it is the collector's Q-point voltage; at voutv_{out}, after C3C_3, it should be close to zero.

Amplitude. With AC coupling, the scope blocks the DC with a capacitor of its own, so you can use a sensitive scale and fill the screen with the swing. Read VppV_{pp} as in Lecture 01. AC coupling adds a high-pass filter of its own; check in your scope's manual that its corner is far below your signal frequency.

Phase. Put vinv_{in} and the output on two channels with the same time base. Measure the time Δt\Delta t between matching points, such as a positive peak of one and the nearest negative peak of the other, and convert it: phase = 360° × Δt/T\Delta t/T. An inverting stage gives close to 180°.

9.2 Three waveforms at once

Task 4 asks for vinv_{in}, vCv_C and vloadv_{load} on the screen together. Each probe has a ground clip, and all of them must go to the circuit's ground: on most bench scopes the clips are tied together through the instrument, so a clip on any other node shorts that node.

The CE amplifier with three scope probes: CH1 on v in at the generator side of C1, CH2 on v C at the collector, and CH3 on v load across R L after C3. Every probe ground clips to the circuit ground.
Figure 28. Probe points: CH1 on vinv_{in}, CH2 on the collector vCv_C, CH3 on the load vloadv_{load}. Every ground clip goes to circuit ground.

Trigger on CH1, the input, so the picture is stable whatever the load does. Give each channel its own volts per division. Above the corner of C3C_3, vloadv_{load} is the AC part of vCv_C without its DC. A smaller load makes both smaller, because it lowers rcr_c.

9.3 Small signals are noisy

A few millivolts at the input is close to the noise of a scope channel on its most sensitive range. Use the scope's averaging or its bandwidth limit if it has them, read the amplitude with the built-in measurement rather than by eye, and say in your report which you used. Another way is to measure the generator's larger, cleaner signal and compute vinv_{in} from a known divider, but then the divider's tolerance joins your error budget.

Practice 9

(a) On a screen at 10 mV/div, vinv_{in} spans 1.6 divisions; at 0.5 V/div, vCv_C spans 3.4 divisions. Find the gain.

(b) At 200 µs/div and 1 kHz, the input's peak and the output's nearest trough are 0.1 division apart from where 180° would put them. How many degrees is that?

10 Error Analysis

Every task needs an error analysis. For each value you report, state its uncertainty, name the main sources of error, and compare it with the value you expected, simulated or took from the datasheet. Two cases in this lab need care.

10.1 β from the base current

Task 2 asks for IBI_B and βdc. It is tempting to compute IB=IE−ICI_B = I_E - I_C from two currents you already measured. That subtracts two nearly equal numbers, and their errors add.

Worked example: two ways to find the base current

Suppose your meter reads each current to ±0.5 % and you find IEI_E = 0.8265 mA and ICI_C = 0.8210 mA. Then IBI_B = 5.50 µA, with an uncertainty of

δIB=(0.005×826.5)2+(0.005×821.0)2 μA=5.82 μA\delta I_B = \sqrt{(0.005 \times 826.5)^2 + (0.005 \times 821.0)^2}\ \mu\text{A} = 5.82\ \mu\text{A}

That is 106 % of the value (8.24 µA in the worst case). The difference tells you nothing about β.

Measure IBI_B directly instead, on the µA range in series with the base. If that reading is 5.47 ± 0.2 µA, then

βdc=0.8210 mA5.47 μA=150.1,δββ=(0.25.47)2+0.0052=3.7 %\beta_{dc} = \frac{0.8210\ \text{mA}}{5.47\ \mu\text{A}} = 150.1, \qquad \frac{\delta\beta}{\beta} = \sqrt{\left(\frac{0.2}{5.47}\right)^2 + 0.005^2} = 3.7\,\%

so βdc = 150.1 ± 5.5.

Use your own meter's specification for the ranges you used. You can also find ICI_C and IEI_E without breaking the circuit: measure the voltage across RCR_C or RER_E and divide by the resistor's measured value. Then the resistor's measured value joins the error.

10.2 The gain

Worked example: the uncertainty of the gain

Suppose you read vinv_{in} = 9.8 ± 0.4 mVpp (noise and trace width at a sensitive scale) and voutv_{out} = 1.27 Vpp ± 3 % (the scope's vertical accuracy). Then AvA_v = 129.6, and

δAvAv=(0.49.8)2+0.032=5.1 %\frac{\delta A_v}{A_v} = \sqrt{\left(\frac{0.4}{9.8}\right)^2 + 0.03^2} = 5.1\,\%

so AvA_v = 129.6 ± 6.6. The small input is the larger error.

10.3 What to compare with

MeasurementMain error sourcesCompare with
Bias voltages and currentsMeter specification; resistor tolerance; VBEV_{BE} and β of your partYour design calculation; a simulation
βdcThe IBI_B reading above allDatasheet hFEh_{FE} at your current
Open-load gainReading vinv_{in}; scope accuracy; input too large (distortion)RC/re′R_C/r'_e; VRC/25V_{RC}/25 mV; a simulation
Loaded gainThe same, plus the load's tolerance(RC∥RL)/re′(R_C \parallel R_L)/r'_e
PhaseReading Δt\Delta t; capacitor corners near your frequency180°

Temperature also moves the gain. VT=kT/qV_T = kT/q is 25.26, 25.69, 26.12 and 26.55 mV at 20, 25, 30 and 35 °C, so re′r'_e, and with it the bypassed gain, changes by about 1.7 % for every 5 °C. A transistor warmed by your fingers, or a bench lamp, can shift a reading by that much. A difference between measured and expected only means something when it is larger than the uncertainty. When it is, say which source most likely explains it, and check that the size fits.

Practice 10

(a) You read vinv_{in} = 4.0 ± 0.2 mVpp and voutv_{out} = 0.50 Vpp ± 3 %. Find AvA_v and its uncertainty.

(b) Your measured open-load gain is 20 % below RC/re′R_C/r'_e. Name two causes from this unit that could explain it, and a measurement that would tell them apart.

← All course materials