7805 Supply That Drops Out and Shuts Down After Two Minutes
A 230 V/50 Hz to 9 V, 1 A transformer feeds a bridge rectifier (four 1N4007), a 1000 uF/25 V reservoir capacitor and a 7805 in TO-220 with no heatsink. The load is a logic board plus a relay drawing 0.60 A continuously. Measurements: 11.3 V peak across the reservoir, about 8.3 V average, and the 5 V rail shows a 0.6-0.8 V sawtooth at 100 Hz. After two to three minutes the output collapses to about 4.5 V until the unit is switched off and allowed to cool. Ambient inside the enclosure is 45 degrees C. The 7805 datasheet gives dropout 2.0 V typical (2.5 V maximum), theta_JC = 5 degrees C/W, theta_JA = 65 degrees C/W without a heatsink, T_J maximum 125 degrees C, and 35 V absolute maximum input. Available: 4700 uF/25 V capacitors, a 15 degrees C/W heatsink and thermal compound.
1. Calculate the ripple, the minimum input voltage and decide whether the regulator is dropping out.
Read the worked answer
- Peak input: V_pk = 9 x 1.414 - 2 x 0.7 = 12.7 - 1.4 = 11.3 V, which matches the measured value.
- Ripple: delta_V = I_L / (2 f C) = 0.60 / (2 x 50 x 1000 x 10^-6) = 0.60 / 0.100 = 6.0 V
- Minimum input: 11.3 - 6.0 = 5.3 V
- The regulator needs V_in at least V_out + dropout = 5 + 2.0 = 7.0 V (7.5 V for the 2.5 V worst case).
5.3 V is far below 7.0 V, so for part of every 10 ms cycle the pass transistor is fully enhanced and the output simply follows the collapsing input. That produces exactly the observed 100 Hz sawtooth on the output rail and an average output below 5 V. The ripple formula is an approximation (it assumes the capacitor supplies a constant current for about 8 ms and ignores the diode conduction angle), but the margin is so large that the conclusion does not depend on it.
Note that the capacitor voltage rating is not an issue here: 11.3 V peak on a 25 V part is fine.
2. With the 4700 uF capacitor fitted, calculate the regulator dissipation and decide whether the 15 degrees C/W heatsink is adequate at 45 degrees C ambient.
Read the worked answer
- New ripple: delta_V = 0.60 / (2 x 50 x 4700 x 10^-6) = 1.28 V
- Minimum input: 11.3 - 1.28 = 10.0 V; average input about 10.6 V. Dropout is solved.
- Regulator dissipation: P = (V_in,avg - V_out) x I = (10.6 - 5) x 0.60 = 3.4 W
Without a heatsink: delta_T = 3.4 x 65 = 221 degrees C, so the junction hits thermal shutdown within minutes. That is the two-to-three minute collapse.
With the 15 degrees C/W heatsink and compound (theta_CS about 1): theta_JA = 5 + 1 + 15 = 21 degrees C/W, giving delta_T = 3.4 x 21 = 71 degrees C and T_J = 45 + 71 = 116 degrees C against a 125 degrees C limit. That works, but the margin is only 9 degrees C, and a warmer day, a closed box or a slightly higher mains will trip it again. A 10 degrees C/W heatsink gives T_J about 99 degrees C, which is the right engineering choice; the tab must also be isolated if the heatsink touches the chassis ground.
3. Would replacing the transformer with a 12 V, 1 A type solve the problem? Compare the alternatives and recommend one.
Read the worked answer
12 V transformer: peak becomes 15.6 V and the average with 4700 uF about 15.0 V, so dropout is comfortably solved. But dissipation rises to (15.0 - 5) x 0.60 = 6.0 W, nearly double the heat, and the 15 degrees C/W heatsink would give a 127 degree rise, far beyond the limit. It also creates a 7805 input of about 17 V peak, which is fine for the 35 V limit but wasteful. So the change cures the symptom and worsens the real problem.
Alternatives:
- 9 V transformer + 4700 uF + 10 degrees C/W heatsink: T_J about 99 degrees C, simple, about 3.4 W wasted. Acceptable.
- A buck converter (for example LM2596) after the reservoir: at 80-90% efficiency the dissipation falls to 0.3-0.7 W, no heatsink, but switching noise appears on the rail and the part count and layout care rise.
- A 5 V switching adapter feeding the board directly: the cheapest solution if the 5 V rail has no analogue requirement.
Recommendation: for a project, option 1 is the safe minimum fix (bigger capacitor plus adequate heatsink), but note that the 0.6 A DC load means the secondary RMS current is around 0.9-1.1 A for a capacitor-input rectifier, so the 1 A transformer is already at its limit; a 1.5-2 A transformer would run cooler and hold its voltage better at low mains.