I am building an adjustable power supply based on a cheapo ATX PSU board. I'll be rolling my own controller, and it's 2-transistor forward with MOSFETs rather than half-bridge with bipolars, but with a little rewiring of the power section and a daughterboard for the new controller i can use most of the old board. If i have spots for the power switches, capacitors, heatsinks, diodes, transformers, and a free auxiliary supply, why reinvent the wheel?
To feed the new controller i needed to get 15 volts out of the standby supply (yes it's possible). I found the feedback point and wired a pot. When i powered it up, i got 0.something volts instead of more than 5v. The rail got shorted. I probed everything but the source eluded me for more than 15 minutes. Sandwiched between the minimum load resistor and the big 2200uF 16v cap i had installed in preparation for the mod, was a little 5v1 zener. As the voltage went past 5.1v it went short to protect the motherboard (if one were connected).
With the zener removed, a new optocoupler with a higher value resistor for the LED and a higher value minimum load resistor, i got my 15 volts (it goes up to 20v before losing regulation btw) and i also got rid of the oscillation, the transformer is now quiet. The standby supply has no problem providing 1.7A continuous and the voltage only drops to 14.6v with this load. Should be more than enough for my new controller as it runs fine off a 7915 which has a maximum of 1.5A.
Background: This power supply came from a computer i got for free a couple years ago and it was made specifically for Torent Computers Romania, it's not just a rebadge. I guess it was a custom order because apart from the Fuhjyyu caps (all good!), the 4-diode treatment (but good enough diodes) and the missing transient filter, it had complete filtering on the secondary, including the 5vsb rail, properly sized heatsinks, and decent rectifiers. Apart from a loud whine from the standby supply transformer (it was oscillating due to worn out optocoupler, the LED current had been set too high) it still worked and all voltages were fine.
Morale: An unprotected 5vsb supply can deliver 20 volts if the feedback loop fails. Are you really sure you want to keep using that gutless wonder?
To feed the new controller i needed to get 15 volts out of the standby supply (yes it's possible). I found the feedback point and wired a pot. When i powered it up, i got 0.something volts instead of more than 5v. The rail got shorted. I probed everything but the source eluded me for more than 15 minutes. Sandwiched between the minimum load resistor and the big 2200uF 16v cap i had installed in preparation for the mod, was a little 5v1 zener. As the voltage went past 5.1v it went short to protect the motherboard (if one were connected).
With the zener removed, a new optocoupler with a higher value resistor for the LED and a higher value minimum load resistor, i got my 15 volts (it goes up to 20v before losing regulation btw) and i also got rid of the oscillation, the transformer is now quiet. The standby supply has no problem providing 1.7A continuous and the voltage only drops to 14.6v with this load. Should be more than enough for my new controller as it runs fine off a 7915 which has a maximum of 1.5A.
Background: This power supply came from a computer i got for free a couple years ago and it was made specifically for Torent Computers Romania, it's not just a rebadge. I guess it was a custom order because apart from the Fuhjyyu caps (all good!), the 4-diode treatment (but good enough diodes) and the missing transient filter, it had complete filtering on the secondary, including the 5vsb rail, properly sized heatsinks, and decent rectifiers. Apart from a loud whine from the standby supply transformer (it was oscillating due to worn out optocoupler, the LED current had been set too high) it still worked and all voltages were fine.
Morale: An unprotected 5vsb supply can deliver 20 volts if the feedback loop fails. Are you really sure you want to keep using that gutless wonder?
