Contents:

Repair and improve broken APC UPS RS-500

Not charging

Symptoms

Warning

Easy fix

Complex fix

Understanding the switching power supply

Technical description of the modification

Credits

Related: Repair instructions for APC Back-UPS RS 500 on Heime, APC UPS units - troubleshooting on Badcaps Forums

Repair and improve broken APC UPS RS-500

Not charging

Broken resistorOne day my UPS stopped to charge the battery. After first look inside I found two resistors that were apparently broken (R150 and R151). Guessing from the relics of the parts, I replaced them by two 10Ω resistors. I think that these resistor pass in fire if your CRT is starting just in the moment of battery test or change of power type.

But it did not help, so I started to analyze the switching power supply and considered that the oscillator based on UC3842B is not working properly. First I suspected that the IC is broken, but later considered that it works, but the 22µF capacitor C7 seems to be underestimated and not sufficient to keep power for oscillator running in a standard conditions.

Symptoms

You can easily verify that your device is broken in the same way: Measure voltage on R139. If you get more than 11V, the oscillator is probably OK. Less than 10V means breakage mentioned here (my had about 8V). 0V means total breakage of the oscillator or blown resistor R139. Warning: There is a dangerous voltage on R139! Use an insulated voltmeter!

What is happening here?

  1. voltage reaches 12V, oscillator starts
  2. transformer oscillator slowly starts
  3. transformer oscillator still not reached fully working state
  4. voltage drops to 8V, oscillator stops
  5. voltage reaches 12V, and the loop repeats

Warning

Don't try to “fix” broken UPS by a plain battery replacement. Oscillator failure causes permanent discharge resulting in a battery fault. If you connect a new battery to a broken UPS without a fix, your battery will permanently fail in a few days or weeks and you will have to purchase new one again!

Some UPSes use high voltage electrolytic C22 capacitor. There can be a dangerous voltage several seconds after disconnecting it from mains. In case of blown resistor R139, dangerous voltage may remain there for hours or days!

Easy fix

Far the simplest fix of this problem: Replace C7 by a capacitor with a higher value, e. g. 47µF. It should start to work.

If the oscillator works properly, there should be about 13.65V on the secondary part. If it does not, the voltage is zero or very low.

Note that if charging was broken for more that few days, your battery is very probably dead and needs to be replaced as well. You may see Replace battery signal.

If you are not familiar with electronic repair work, you may consider to follow Repair instructions for APC Back-UPS RS 500 on Heime. It describes easy fix of the same problem in more detailed steps.

Complex fix

Degradation of the cover plasticConsulting the data sheet of UC3842B, I found yet another broken design: At least in RS500 230V version REV05 one of resistor around (R28) eats permanently about 2 Watts and generates a lot of heat. The temperature there is permanently about 90°C and it even causes degradation of the cover plastic (see the image).

That is why I decided to do a more radical change:

This modification decreases power consumption of the bare UPS by 2 Watts and the temperature of the resistor by 40°C at cost of 1–2 seconds delay between connecting the power plug and oscillator start.

Side view of the oscillator Top view of the oscillator

Understanding the switching power supply

The FET Q6 is chopping the DC to T1 (winding 4-6) by grounding the pin 6 by some frequency. T1 then provides square AC voltage on winding 1-2, which is fed to U2 as DC again by D7/R50. Ferrites B8/B9 provides RFI shielding.

C7 also provides filtering for a smoother DC for U2 so a bigger capacitor would be better for that too. Winding 7-9 on T1 actually provides the charging current for the battery.

Technical description of the modification

Original design used a very hard divisor (R28 and R139) that divides the plug voltage and can supply about 10mA. It is nearly enough for running the oscillator, so 22µF was sometimes enough.

But the the resistor divisor should be used just to charge capacitor C7 and then use its energy to start oscillation. Once the oscillator runs, it is powered from the transformer (which gets its energy through 2SK2608 power FET) and the divisor is not needed for the regular work. You can verify this fact by disconnecting the large divisor resistor, touch, and remove. Oscillator runs until you disconnect the power.

The UC3842B data sheet says, that if there is not enough voltage to run oscillator, the IC eats less than 0.5mA.

It means, that the divisor could be designed to provide only 0.5mA, not 10mA. Additionally, low current of the divisor allow to omit resistor to the ground completely: The IC has a protection Zener diode (well, it should be never activated with the design used in this UPS – once the voltage reaches about 14V, regulation feedback (using opto-couplers) will prevent further raise of the voltage.

In this new design, capacitor must be larger than in the simple fix – e. g. 220µF. The complete start of the oscillator is powered nearly exclusively from it. It must have enough juice to keep voltage above the minimal voltage until the transformer starts to give regular power.

Computing the resistor R28 value:

Credits

Thanks to George Helas for sending a fixes and providing a photo of broken resistor.