YO3BN - Backup Power Supply 13V/500mA


Home Articles Projects Products Bookmarks About

Posted at - 28th February 2018

Uninterruptible Power Supply Module - 13V/500mA


This module was intended to be used with a 16Vac transformer with some functions managed by a microcontroller (MCU). It is a simple analogue module which provide uninterruptible power supply feeding other low power equipments, thus by using a backup lead-acid accumulator.

Objectives:


Prototyping


Firstly, the transformer is rated at 16VRMS, therefore its peak value will be VRMS * √2, roughly 22.6Vpeak in this case. Taking into consideration the voltage drop over rectifying diodes, there will be some chances that Vcc spikes could reach 30Vcc. With this assumption in mind, I also know that my MCU will use 2.56V as e reference voltage with an optimum input impedance of approx. 10KOhms, therefore a resistive divider of 10% was calculated in order to monitor the main Vcc, 25.6 * 10% = 2.56V (above image, upper left corner).

Then the accumulator's charging circuit calculus in upper right corner. Two transistors operating in saturation mode and a power resistor as a buffer. Schottky diodes were used because they had low voltage drop, SR550 have 0.1V for smaller currents and about 0.4V for 1A.




In the image above I've done some calculations about using a resistive divider in order to activate the backup accumulator, but it didn't convinced me entirely. Also, the discharging curve of capacitors delays the MOSFET triggering. Therefore something like a schmitt trigger should be used.




I tried to keep the things as simple as I can, so after drawing the schmitt trigger I was wondering if I could design something simpler. So I used that zener trigger drawn at the bottom of the page.


The Schematic


Taking it into functional blocks, first, the rectifying block is composed by D1-D4; C1-C2; followed by the resistor divider formed from R2-RV2 which will prepare the main voltage to be monitored by MCU. D5 blocks the current flowing from accumulator side when main supply is off. Then follows the LM317 regulator with its adjacent components, RV3 is used to adjust main's output voltage (13V in this case).

The charging circuit is activated by a positive voltage (+5V) coming from MCU GPIO in CHRG CMD. The D8 forms some sort of protection for MCU allowing current flowing only from MCU side, but it can be omitted since R7 will provide enough resistance to keep larger currents away. R7 is calculed to bring Q2 in saturation region which is dictated by R4 and Q1 jonction voltage drop. R6 actually keeps the base of Q2 closed to ground when CHRG CMD is floating. Q1 is a PNP type acting as a voltage amplifier, connecting the battery to main Vcc through R3 power resistor. Thus, Q1 works in saturated region limited by battery voltage and R3, therefore power dissipation through Q1 will be negligible, so no heatsink is required. Further, the charging and discharging voltages of the accumulator battery will be monitored by MCU through R1-RV1 divider. Also, the two zener diodes D12, D14 and R12-R13 will cut off the voltage used by MCU's ADC acting like a protection, as it was already mentioned, the MCU will use 2.56V as a reference.

Now comes the uninterruptible switch circuitry. This is composed by the zener trigger Q3, D9, R8-R10 and the switch part R11, D10-D11, Q4, D13. The LM317 needs a room of +3V over its output voltage, so 13 + 3V = 16V. Therefore, the zener trigger should be turned off by a voltage less than 16V. Using a zener diode of 15V and adding the internal transistor voltage drop of 0.7V the threshold will be around 15.7V which is perfect. This stage will lock the switch Q4 into an open state by giving a positive voltage at its gate. Note that Q4 is a P channel MOSFET, so it will enter in conduction only when its Gate to Source becomes negative, i.e. -5V.

Let's recap the switching process!
Case one:
Main Vcc >15.7V, Q3 closed - positive collector, Q4 opened - positive gate, the battery is isolated.
Case two:
Main Vcc <15.7V, Q3 opened - floating collector, Q4 closed by accumulator through R11 - negative gate, the battery powering the rest.

Other things happens at the gate of Q4 which they're important. When the Q3 is closed, the Vgs of Q4 is reverse polarised around +8V. That's because the Vcc - Vbatt = Vgs, that means 22V - 14V = 8V, using KVL we can find this. But if there is no battery and the MCU accidentally turns on the charging process, Vgs will be equal to Vcc, which is too much (+22V). In IRF5305 datasheet it is specified a Vgs maximum of +/- 20V. So, as a protection two zener diodes D10-D11 in common cathode configuration were used as a Gate protection preventing swings bigger than +/-15.7V. Now if Q3 is opened, no Vcc, the Vgs will be negative Vbatt, i.e. -14V so Q4 will fully conduct. Rdson will be so small that its power dissipation is negligible, it doesn't need a heatsink. R10 is a sort of protection preventing high currents to develop if Q4 or Q3 fails.


Testing


With RV3 set to 13.4V, current consumption of 0.2A, after 5 minutes, the heatsink of LM317 gets warm, from 20°C to 55°C.

Sourcing 1A, the power dissipated by LM317 is around 7W, too much for a such small heatsink, therefore after 4 minutes the heatsink reaches 127°C starting from 20°C and the LM317 shuts itself off by activating the thermal protection.

Testing the charging process, using a 12V 7Ah lead acid battery with a 12.7 initial voltage, R3 dissipate approx. 1W, the temperature reaches 133°C into 5 minutes.


Other Notes


KiCAD Files


Update 17th March 2018

Some mistakes I did designing this board:

73 de YO3BN


Home Articles Projects Products Bookmarks About