|
Home | Articles | Projects | Products | Bookmarks | About |
Posted at - 15th April 2018
Overcurrent protected ports are useful for remote devices placed in an unsafe area where the power supply cable could be damaged for example. Therefore, having an overcurrent protected port is a must. Just imagine how it will be to have a power supply feeding multiple devices in a star topology and suddenly one device fails or the conductor is damaged by a metallic object which shorts the positive with negative, this in turn will result in a general fault cutting off the entire power supply. One solution for this case could be individually fuse protected ports, but a fuse is more expensive, more lazy than an electronic circuit and will require some maintenance.
In the picture above is ilustrated a board with three triggered ports -from left- and one standalone port -right- all featuring overcurrent protection. The standalone protected port could be used for feeding non-stop devices, while the triggered version could fed intermitent devices on demand. The triggered version is based on the standalone version also.
Overcurrent Protected Port (standalone version):
In the designing process of these circuits, the goal was to be able to supply 12-14Vcc up to 500mA with a low voltage drop. A resistor of 1 Ohm (1%; 3W) was used as a sort of current sensor. When the load sink 500mA, the resistor will develop 500mV across it. This voltage will be read by a transistor between base-emitter jonction, therefore when the current reaches about 650mA the transistor will start to conduct and will switch a P-MOSFET off. The mosfet will provide a low voltage drop due its low Rdson (0.06 ohms) and almost no current controlled gate, being the right device for this job. The whole circuit uses a positive feedback, so it will shutdown itself definitely until the load will be removed from feeding line.
The circuit was so fast at first that it couldn't charge a small capacitor intended as a load. So, a delay was then taken into consideration and the circuit was rethought to ignore shortcircuits shorter than 10-15ms, in order to be able to charge at least one cap. of 1000uF. More details in the schematic section.
Some photos taken during prototyping process:
 |
 |
 |
 |
 |
 |
 |
 |
As it can be seen, on the left side is the continuous power feed, in the right is the output of the circuit. The Q1 is a P-channel MOSFET which acts as a normally closed switch with its Gate connected to GND through R3. Since no current flowing through R3 the voltage drop will be 0V across it. Measuring Vgs will give about -12.75V, therefore the Q1 is in conduction state. The diode D1 is for reverse protection and usually incorporated into Q1 by default.
After the mosfet switch the current flows through precision resistor R1 (1 Ohm, 1%, 3W) which is actually the current sensor. When 500mA flows through R1 the voltage across is direct proportional, in this case will be 500mV (V = IR) and the output voltage across the load decreases to 12.25V. The voltage drop over Q1 and R1 will be read by Q2, so when the load sinks more current than 600mA (650mA typically) the base-emitter jonction of Q2 became polarised and the Q2 starts to conduct between collector-emitter applying positive voltage to the gate of Q1.
So, when the Q2 starts to conduct and the voltage across R3 is about 7-8V the Vgs of Q1 equals about -5V (-12.75 + 7.75 = -5) any further increase in current at load will develop even more voltage drop across R1, thus increasing voltage to gate of Q1 dimming Q1 off even more, therefore the overall voltage drop across the Q1+R1 increases further more and the Q2 conducts more, the voltage across R3 raises again, so when the gate of Q1 reaches about -3.5V Q1 will be turned off, so the entire circuit turn itself off then the voltage drop measured by Q2 is about full rail (12.75V) keeping the voltage of Q1's gate at 0V (-12.75 + 12.75 = 0) until any load is removed from line, that's the positive feedback.
Things are running superfast with electronic components and the probability of having a constant pure resistive load is obviously less realistic. In general loads are highly dynamic and reactive, so at first this circuit was unable to charge a capacitor of 1000uF because it behaves as a shortcircuit when it is fully discharged. In the schematic, C1 is supposed to introduce some delay in the feedback loop, slowing down the protection response. For a shortcircuit at output port, the protection is activated only when C1 reach the Q2 base-emitter jonction voltage (~0.65V). The charging process of C1 is made via R2 at the speed of 12.75V until it reaches approx. 0.65V and Q2 starts to conduct. So, this delay can be roughly calculated as follows:
Replacing V0 with 12.75V; V(t) with 0.65V; t with 0.005 seconds; R with 22000 Ohms; C1 value equals 4.3uF. A good standard value will be 3.3 or 4.7uF.
It was seen that the delay introduced in feedback loop offers a limited response for capacitive loads, but for inductive loads high voltages can develop across output port damaging the Q1 mosfet (tested). So, a protection in this way is mandatory. The zener diode D2 will short voltage spikes coming from inductive loads, also the C2 is intended for filtering.
This type will supply power only while CMD-IN receives a positive voltage between +5 to +15V. This version is based on the standalone version too, but with addition of several components. From CMD-IN, if a >5V is applied Q3 starts to conduct connecting the Gate of Q1 to ground, thus closing the mosfet switch Q1 and the circuit behaves exactly the same as standalone version. When there is no voltage potential at CMD-IN, base of Q3 is locked down by R6 thus leading Q3 open, disconnecting Gate of Q1 from the ground.
Q4 does two things:
The D3 diode will keep current flowing from Q4 into load when Q3 is open. Without that diode the Gate of Q1 cannot be kept near to zero and the Q1 starts to conduct erroneously. While Q4 keep the Gate of Q1 near zero, also from Q4's collector will flow a high impedance voltage through Q2, C1, R2 messing things up, therefore a dummy resistor R7 will keep this branch clean during standby.
Few words about these circuits: easy to build, reliable. They were not tested in a production environment yet, but the following:
|
Home | Articles | Projects | Products | Bookmarks | About |