PC ATX Power Supply turn off when driving Boost Converter

Hello Guys,

Hope you are doing great.

I was looking for a supply for my boost converter with high current capacity and came across an ATX Computer Power supply. I connected the dummy load of 4.7ohms between +5V and GND and on +12V rail i connected a Car Headlight bulb. After turning on the turning the bulb lit and voltage and current were 11.90V and 7.26A. After ensuring that the supply was able to drive a high current bulb i disconnected the bulb and connected the boost converter across 12V rail and started from the minimum duty cycle. Upon increasing to 60% duty i observed that the current through boost converter using DMM was about 2A and voltage went down to 10V. Upon further increasing the duty cycle the voltage went down to 8.5V and the power supply turned off. I switched off the supply and turned it on after half a minute and same happened.

I am confused now that the supply easily drove the headlight bulb which was taking 7.25A and the voltage were 11.98V but in case of boost converter the voltage dropped to 8.5V at just 2A. Is this because of the pulsating current that boost converter takes as input? (Inductor being switched) If yes then how this can problem can be solved?

Try increasing the load on the 5V.

You seem to have a problem with the choice of boost converter surge current ripple voltage interfering in the feedback loop. You may need to reduce the ripple voltage by series L or low ESR shunt C at converter input. Ensure power leads are twisted pairs to reduce immunity issues. Then show your layout and circuit.

I connected a capacitor of 22000uF in across the output terminals of power supply but same problem still exist.

Hard to say without knowing the actual current and voltage waveforms.

3.3V and 5V nodes have both a minimum current specification of 0.3 A to keep their voltage limits, I don't even expect that too low load causes a shutdown. It's rather a problem of exceeding the 12V peak current specification.

Considering that the ATX supply has already considerable output capacitance at the 12V nodes, an external 22 mF capacitor must not necessarily absorb the current peaks completely. It depends on the actual wiring and capacitor ESR. A RC or LC filter might help, as suggested.

Here are the schematics of the two circuits:

Have you tried a diode or zener diode across the 12V supply?

Hello basit701,
Try using a 27 ohm 5 Watt resistor for the 5 volt line as a dummy load as some ATX
PSU's require a minimum load as this rail is usually the regulated one.
Try using a higher value capacitor/s on the output of your converter.
Regards,
Relayer

I have a question regarding Current ratings and dynamic load ESR.

What are the 12V current ratings?
What is the headlamp cold resistance?
What is the dynamic load impedance of the boost converter avearge load referred to source? ( using output load R divided by voltage boost ratio squared) Since minimum load depends on ESR of the switched reactive load and duty cycle, can you measure the ESR of this load and compare it to the headlamp cold resistance?

If none of this helps, can you insert another choke and low ESR caps on 12V to boost regulator to reduce load regulation ripple that may OVP faults in the 12V regulation error sense circuit.

Where are you taking your v/I measurements. At output of psu or at the load. As others have said you are maybe exceeding the capabilities of the psu. Your converter is not 100% efficient and you will have losses. Need to know what psu output power Is. Hopefully there's a label on it. Also input and output v/i of both sides of converter to get and idea of its efficiency.

basit701 said:

Here are the schematics of the two circuits:

Click to expand...

Input current ripple is very small with 2 mH (millihenry) inductor and 100 kHz switching frequency. About 50 mA input current ripple can't be a problem for ATX power supply. Perhaps the inductor saturates causing high current peaks. I assume DMM can't measure 100 kHz ripple current and shows more like an average current value instead.