Controlling your World the Easy Way

Hi guys
I've bought quite a few PSU5's and now PSU5 no noise to use in Pinball WPC89 repairs. In 99% of cases they work flawlessly but I've had 2 or 3 stubborn boards where the regulator won't start up or is very intermittent. When these intermittent boards fire up they run flawlessly for hours but if I turn it off and come back an hour later it refuses to start.
The bridge and cap's are new in all cases and I make sure the regulator is well grounded. I've noticed if I discharge the main 15000uf cap the regulator will start every time but when it doesn't I can see the 5v LED dimly lit and pulsing very fast.
Any help would be appreciated.

This is new to me on the PSU5. This is a generic issue with switch mode regulators with built in current limits and I haven't seen it on the PSU5, only on the PSU3-12 and PSU3-5.

The mechanism is a little hard to explain so I will provide the fix first and then try to explain what is happening. The easiest way to fix the issue is to add a resistor across the 15,000uF cap to ensure that it discharges in the time it takes to turn off the machine and turn it back on. This will work around the issue.

Switching regulators do not always behave well when there's a voltage present on the output when the regulator is turned on. This is due to the built in slow ramp circuit also called the soft start circuit. The soft start circuit is a circuit that ramps the output voltage at a controlled rate when the supply is powered up. It is required to prevent the output voltage from over shooting the desired voltage when the regulator is turned on with only a light load and without the soft start circuit. The control loop will see the output voltage is too low and it will charge the output capacitor as fast as the current limit will allow. When the voltage reaches 5V (in the case of the PSU5) the inductor current is now as large as the circuit will allow and there's little or no load current. The energy stored in the inductor will now be dumped into the output capacitor for the time taken for the control loop to get to the correct duty cycle. This causes the voltage to exceed the desired voltage and can break attached circuits.

To prevent this the PSU5 ramps the output voltage at a controlled rate by changing the voltage reference during startup. Instead of immediately going to 0.8V it is ramped from 0 to 0.8V in a fixed amount of time. This allows the control loop to ramp the output voltage from 0 to 5V in a controlled time without going into current limit. When the output voltage reaches 5V the inductor current is low enough to prevent meaningful overshoot of the output voltage. This is the plan and generally it is a good plan.

There's a wrinkle in the plan when the output is short circuited or overloaded. Lets assume the regulator is working normally and the output is short circuited. The output cap dumps all the charge into the load and the control loop sees that the output voltage is too low. It now commands maximum duty cycle in an attempt to raise the output voltage to the correct value. The current gets large and the current limit kicks in preventing the regulator from blowing up. If the short is removed the output voltage will increase as fast as the current limit will allow resulting in voltage overshoot, exact as we had with uncontrolled startup. To fix this the regulator will detect when the output voltage is too low and reset the soft start circuit so that when the short is removed the output voltage will ramp up in a nice and controlled way.

When there's a voltage present on the output of the regulator when the regulator is turned on things can go wrong and it your case it does. Lets assume that for normal startup the regulator needs a 60% duty cycle to ramp the output voltage without running into the current limit. As the reference voltage increases the regulator sees the output voltage a little below the reference and commands a 30% duty cycle. Before the output voltage settles the reference voltage has increased further and the duty cycle stays at 60%. When the reference stops changing the duty cycle quickly settles to the correct value. Lets say the input is 12V and 5V output. The correct duty cycle is 5/12% or about 42% and the regulators will operate at 42% duty cycle. If the output increases the duty cycle will decrease to maintain regulation.

Finally we get to the issue: Lets say there's 3V on the output. During soft start 3V will normally be reached when the reference voltage is at 3/5 of the final value of 0.8V or 0.48V on the reference during soft start. Lets see what happens when the regulator turns on and we pick up the story where the reference is at 0.1V. Under normal circumstances the duty cycle will be 60% but now the output voltage is already way above the desired voltage. The control loop recognizes this and reduces the duty cycle to essentially 0% until the reference gets to 0.48V at which time it tries to increase the duty cycle to 60%. This takes a bit of time and by the time the duty cycle is 60% the output voltage is below the desired voltage commanded by the soft start circuit. In your case what happens is that the difference between the output voltage is big enough to trigger the short circuit detection and the reference ramp is reset. If the duty cycle got high enough before the short circuit detection gets tripped there's enough energy in the inductor to charge up the output voltage high enough so that during the next attempt the voltage on the output is high enough for the behavior to recur. This leads to a low and pulsating output voltage as the regulator repeatedly tries to restart.

Two fixes are available. Discharge the output or slow down the ramp. I recommend discharging the cap because this is easily done with a resistor across the output cap. The reason why it occurs on only some machines is because it depends on the exact value of the output cap, its leakage and the current drawn by the boards when the voltage is low.

Slowing down the ramp is possible by changing a capacitor on the PSU5 but it may have unintended consequences. The logic boards and microprocessors need a ramp that is fast enough for the reset circuits to work correctly. Often there's a wide range of ramp times that work with the reset generators but most of them will fail if the ramp time is too long. I will add a photo of which cap to change later when I had a chance to prepare it.

I hope this helps.
Daniel

Thanks for the fantastic response Daniel. Any recommendations on resistor size?