The switch that won't let go
You tell a transistor to switch OFF. For the next hundred nanoseconds it keeps conducting as if nothing happened. Then it collapses.
Wile E. Coyote runs off the cliff and hangs in mid-air, legs still spinning, until he looks down. A saturated transistor does the same thing. Today we measure that hang time, give it a name (storage time) and learn to shorten it.
Guess before we start: for a transistor switch driven hard into saturation, which is slower, turning ON or turning OFF?
Argue it in pairs for two minutes. There may be more than one defensible answer.
Closer to home: an IPS keeping your fan running through load-shedding switches its power devices thousands of times a second. Every slow edge is wasted as heat.
One line, two addresses: the DC load line
KVL around the collector loop of a common-emitter switch gives the only (IC, VCE) pairs the circuit allows:
A straight line on the output characteristics. The transistor (through IB) picks where on it you sit: the Q-point.
| End | VCE | IC | Switch |
|---|---|---|---|
| Cutoff | VCC | ≈ 0 | OFF (open) |
| Saturation | ≈ 0.2 V | (VCC − 0.2)/RC | ON (closed) |
Amplifiers live in the middle of the line; switches live only at its two ends. They should race through the middle as fast as possible, because that is where the transistor is both carrying current and dropping voltage, so it heats.
The ideal intercept is VCC/RC at VCE = 0. The real ON point stops just short of it, at VCE(sat) ≈ 0.2 V. Use IC(sat) = (VCC − 0.2)/RC in calculations.
Slide the Q-point along the line
How hard to push: saturation and overdrive
To reach the ON point, the base must supply at least what the load needs, divided by the worst β you might get:
How far past that minimum you go is the overdrive factor:
Last class: IC(sat) = 4.8 mA, βmin = 50 ⇒ IB(min) = 96 µA; with IB = 0.43 mA, ODF = 0.43/0.096 ≈ 4.5.
In saturation, IC ≠ βIB. β·IB is what the transistor could carry; RC decides what it does carry. The ratio IC/IB in saturation is the forced β, always below β.
Why βmin? One reel of BC547s can range from about 110 to 800 in β, and β drops at low temperature and at high current. Designs use 2–5× overdrive for margin. Why not 50×? Slide 14.
Your turn · 1: read the load line
VCC = 9 V, RC = 1.5 kΩ, β = βmin = 100, VCE(sat) = 0.2 V.
- Find the load-line end points and the ON point.
- For IB = 30 µA find the Q-point. Switch or amplifier?
- Find IB(min) and the ODF at IB = 100 µA.
- Compare the transistor's power at (b) and in saturation.
Work it out first.
Press Solve when the class has an answer.
Worked example: an ESP32 drives a 12 V relay
Given
A rooftop water-pump controller in a Dhaka apartment: an ESP32 pin (VOH = 3.3 V) switches a 12 V relay coil (Rcoil = 240 Ω) through an NPN with βmin = 50. VBE(sat) = 0.8 V, VCE(sat) = 0.2 V.
Find
RB (standard E12 value) for at least 2× overdrive, the resulting ODF, pin current and ON-state dissipation.
Method: saturated-switch design (Boylestad, transistor switching networks; Millman & Taub, transistor switch)
Solution
Predict the answer first. Then reveal one step at a time.
Design lab: one pin, three loads
Four times that steal your speed
- td delay: input edge → iC reaches 10 %. Charging the junction capacitances up to cut-in.
- tr rise: 10 % → 90 % of ICS. Building up base charge through the active region.
- ts storage: input falls → iC drops to 90 %. Removing the excess charge stored in saturation.
- tf fall: 90 % → 10 %. Back through the active region.
In saturation both junctions are forward-biased, so the collector junction injects too and the base fills with minority carriers beyond what ICS needs. That excess QS must drain before the collector junction can reverse-bias; until then iC stays stuck at ICS.
The switching oscilloscope
β = 50, ICS = 10 mA, τ = 25 ns, τs = 50 ns · leaves saturation after 20 ns (ts formula: 20 ns); readout ts is to the 90 % point
Charge control: formulas you can use
While active, iC relaxes toward βiB with time constant τ. From 0 toward βIB1:
Excess charge drains with time constant τs, pulled by the reverse current IB2:
Same idea for the fall, now heading toward −βIB2:
The trade-off
More IB1 (overdrive): βIB1 ≫ ICS, so tr shrinks. But IB1 sits in the numerator of ts: storage grows.
More IB2 (reverse drive): shrinks ts and tf. Costs a negative supply or a clever circuit.
If βIB1 ≤ 0.9 ICS the log's argument goes negative: iC never reaches 90 %. Check βIB1 > ICS (saturation) before using any of these.
Worked example: rise, storage and fall
Given
β = 50, τ = 20 ns, τs = 40 ns, ICS = 10 mA. Turn-on drive IB1 = 0.5 mA; turn-off reverse drive IB2 = 0.25 mA.
Find
tr, ts and tf. Then: what happens if you double IB1?
Method: charge-control model (Taub & Schilling; Millman & Taub, switching times)
Solution
Predict the answer first. Then reveal one step at a time.
Your turn · 2: how much reverse drive?
β = 40, τ = 15 ns, τs = 60 ns, ICS = 20 mA, IB1 = 1.0 mA, IB2 = 0.5 mA.
- Confirm saturation and find the ODF.
- Find tr and ts.
- What IB2 would cut ts to 10 ns?
Work it out first.
Press Solve when the class has an answer.
Two cheap fixes for storage time
(a) At each edge C is momentarily a short: a big current spike pours charge in on the rising edge and yanks it out on the falling edge. In steady state C is an open circuit, so RB alone sets a modest IB1 and little charge is stored. Both edges get faster. Rule of thumb: RBC ≈ τ.
(b) A Schottky diode (VF ≈ 0.3–0.4 V) from base to collector turns on before the collector junction (≈ 0.5 V) can. Extra base current is diverted into the collector, so the transistor sits at the edge of saturation instead of deep in it.
No deep saturation, so almost no excess charge: ts ≈ 0.
This is the “S” in 74S and 74LS TTL, where every transistor that could saturate carries a built-in Schottky clamp. Week 3.
Critical thinking · two arguments
“More base current is always safer. Let's drive the base with 50× overdrive.” Argue against it.
Argue it in pairs for two minutes. There may be more than one defensible answer.
A Schottky diode is still a diode. Why doesn't the clamp diode itself suffer storage time?
Argue it in pairs for two minutes. There may be more than one defensible answer.
Your turn · 3: the dim LED
An Arduino (5 V pin) drives a red LED (VF = 2.0 V) with a 150 Ω series resistor from +5 V through an NPN (βmin = 100). A student picks RB = 100 kΩ “because the base only needs microamps”.
Is the LED fully on? Find IC, VCE and the transistor's power. Then fix RB.
Work it out first.
Press Solve when the class has an answer.
Where the switch gets hot
ON: tiny voltage, full current. OFF: full voltage, no current. Both are cheap:
Every edge passes through the middle of the load line where V and I are both large. For a resistive load with linear edges each transition costs VCCICt/6:
Example: 12 V LED strip, 0.5 A, PWM at 50 % duty
VCE(sat) = 0.2 V, tr + tf = 1 µs (a slow power BJT).
PON = 0.2 × 0.5 × 0.5 = 50 mW
At 20 kHz: Psw = ⅙ × 12 × 0.5 × 10−6 × 20 000 = 20 mW
At 200 kHz: 200 mW. Switching loss now dominates.
Faster edges = cooler switch. That is why tr and tf matter even in power electronics, not just in logic.
The valve that killed its transistor
A dialysis machine or infusion pump opens and closes a 12 V solenoid pinch valve on the tubing (coil 24 Ω, L ≈ 50 mH, I = 0.49 A). A transistor switches it. On the bench it works perfectly. After about a week in the ward, the transistor is dead.
At turn-off the coil insists on keeping its current: v = L di/dt. Interrupting 0.5 A in 1 µs asks for 0.05 × 0.5/10−6 = 25 kV. The transistor avalanches at its breakdown voltage instead and absorbs the coil's energy:
Why does it survive the bench test but die after a week? And what is the fix?
Argue it in pairs for two minutes. There may be more than one defensible answer.
Pop quiz
Question 1 / 5
On the DC load line, where does a BJT switch sit when it is ON?
Three things to remember
A switch lives at two ends of the load line: OFF (VCC, 0) and ON (0.2 V, IC(sat)). Saturate with IB ≥ IC(sat)/βmin, times a 2–5× overdrive.
ton = td + tr, toff = ts + tf. Overdrive speeds the rise but stores charge and lengthens ts; reverse drive drains it.
Fixes: speed-up capacitor, Schottky clamp (74LS, Week 3), or a device with no stored charge: the MOSFET, next class. Always put a diode across a coil.
β = 60, τ = 25 ns, τs = 50 ns, ICS = 12 mA, IB1 = 0.6 mA, IB2 = 0.3 mA, td = 5 ns. Find tr, ts, tf, ton, toff and the rough maximum switching frequency 1/(ton + toff).
Work it out first.
Press Solve when the class has an answer.