5VSB Load Tester

[everell]

Pictured below and with drawing is my 5VSB Load Tester. It uses four 10 ohm 5 watt load resistors which can be switched in separately, and a switch to turn the power supply on and off. A 20 pin connector was removed from a “throw away” mother board. Load resistors and switch were soldered to this connector. The power supply then connects to this connector.


When the power supply is plugged into the wall outlet (power switch is in off position) the 5VSB must work, or the power supply won't turn on when the power switch is turned on. This is true regardless of amount of load on the 5VSB rail. So with power switch off, measure the voltage on the 5VSB rail as load resistors are switched in one at a time. Voltage measured should remain at 5 volts +/- 5%.



No resistors “in” is no loading

1 resistor “in” loads to 0.5 amp

2 resistors “in” loads to 1.0 amp

3 resistors “in” loads to 1.5 amp

4 resistors “in” loads to 2.0 amp



I am using only four resistors because any ATX power supply 5VSB should be able to deliver 2.0 amps. Otherwise there is a problem to be resolved. More resistors can be added, but would be destructive testing. My intent was to make sure the 5VSB rail could and would deliver up to 2 amps. I am not as interested in finding at what point the power supply fails and self destructs.

For the second part of the test, switch all four resistors “out”. Then turn the power switch ON. The power supply should come on, and the fan should start turning. Now switch the four resistors in one at a time while measuring the voltage on the 5VSB rail. Once again, the voltage should measure 5 volts +/- 5%.


This is an easy test for a good power supply to pass. Failing either test means trouble. The power supply may stop, or act erratic and unpredictable. This can cause a computer to stop, then restart, or go into blue screen mode. Bulging capacitors and other defects in the 5VSB circuit is a COMMON problem with computer power supplies. Using this Load Tester as a first test will flush out a lot of problems and save a lot of time.

[everell]

Re: 5VSB Load Tester

Although the 5VSB Load Tester can be used for general troubleshooting, I designed the tester to be used in converting 5VSB circuits from “Two Transistor” circuit to PWM chip circuit. As a first example, I recapped an Antec SP450 and did the DM311 mod. With the new DM311 5VSB circuit and no load resistor on the 5vsb rail, it measured 5.08 volts. Great.


Next, I hooked up the 5vsb load tester. Here is how the 5vsb rail looked when loaded:

1 resistor 3.81 volts

2 resistors 3.38 volts

3 resistors 2.47 volts

4 resistors 2.02 volts

Not so good. So what happened? The key is watching how the feedback winding powering the DM311 is reacting to load changes. As more resistors are added, the DM311 has to deliver more power into the transformer. The 5vsb rail is regulated BUT the feedback winding powering the DM311 is not. So as more resistors are added, the feedback voltage powering the DM311 increases. Looking at the DM311 data sheet, when this voltage reaches or exceeds 20 volts, the chip goes into overvoltage shutdown. So here are the feedback voltage measurements:

No resistor 16.02 volts

1 resistor 15.45 volts

2 resistors 19.78 volts

3 resistors 19 to 23 volts varying

4 resistors 12.1 volts

So what is happening? With no resistor, the 5vsb output is 5.08 volts and the feedback power is 16.02 volts. As resistors are added, the 5vsb output is dropping because the feedback power has reached the 20 volt limit, the DM311 stops, then tries to restart, stops, restarts, stops, restarts…..and so on.

I cannot yet say that EVERY Antec SP series power supply will not act properly with DM311 circuit, but this power supply did not work well with it.


Next I tried a TNY277 chip. Here are the results. The 5vsb measurements:

No resistor 5.07 volts

1 resistor 5.04 volts

2 resistors 5.01 volt

3 or more resistors dead

The feedback voltage measurements powering the TNY277 chip:

No resistor 16.80 volts

1 resistor 22.9 volts

2 resistors 25.4 volts

Once again this feedback voltage has gone too high and caused the overvoltage circuit to kick in.

Next I tried the VIPer22A chip. This one worked with all load resistors. For 5vsb rail:
No resistor 5.08 volts

1 resistor 5.04 volts

2 resistors 5.01 volts

3 resistors 4.98 volts

4 resistors 4.94 volts

For feedback winding power to the VIPer22A chip:

No resistor 14.8 volts

1 resistor 20.0 volts

2 resistors 24.0 volts

3 resistors 26.7 volts

4 resistors 29.6 volts

According to the data sheet, 14.5 volts is needed to turn this chip on, 14.6 volts was measured at no load. The maximum voltage to overvoltage protection is 42 volts. With 4 resistors loading, 29.6 volts was measured. So each requirement was met. This chip should be able to handle another resistor or two, but this was not checked.


I conclude that the VIPer22A is the best chip to be used for a 5vsb mod on this Antec SP450, and possibly on most of the SP series. Here is the drawing using the VIPer22A in the Antec SP450 and pictures.

[mariushm]

Re: 5VSB Load Tester

Not surprised the TNY277 didn't make it:

[momaka]

Re: 5VSB Load Tester

Interesting. I will defnitely need to keep this thread in mind when doing the DM311 mods on various PSUs. I still have 2x DM311 chips that I bought ages ago to put in power supplies (but never got around to do it).
Thanks for sharing Everell .

[mockingbird]

Re: 5VSB Load Tester

Pics?

[c_hegge]

Re: 5VSB Load Tester

I usually just replace the 'killer' 22uF 50V cap with a Panny or Nichicon. I've never had an antec overvolt with a good cap in there before.

[rogfanther]

Re: 5VSB Load Tester

Couldn´t a zener diode be used here to somehow regulate the voltage going to the chip ?

[momaka]

Re: 5VSB Load Tester

Couldn´t a zener diode be used here to somehow regulate the voltage going to the chip ?

I was wondering the same thing.
First need to measure the current draw, though. If it's too high, a zener will not work.

[c_hegge]

Re: 5VSB Load Tester

Th3 Un1que says a 1W zenner is the best option. Remember, though, that all a Zenner diode will do in the event of an overshoot is blow up the main transistor. It will save the board, but destroy the PSU.

[momaka]

Re: 5VSB Load Tester

Th3 Un1que says a 1W zenner is the best option. Remember, though, that all a Zenner diode will do in the event of an overshoot is blow up the main transistor. It will save the board, but destroy the PSU.

That's a completely different thing.
The 5.1V, 1W Zener that the Th3 Un1que talks about/uses is placed on the 5vsb output. If 5vsb goes higher than 5.1V, the Zener conducts. So in the even that the 5vsb circuit decides to go over voltage, the Zener conducts, shorts out, and kills the 5vsb circuit but ends up saving the board from damage.
This is most applicable to 2-transistor self-oscillating 5vsb circuits since they are the ones that have the tendency to go over voltage when some component(s) goes out of spec.

The Zener rogfanther and I are referring to is to be placed on the primary side to limit the voltage to whatever smart chip is generating the 5vsb (DM311, TNY277, VIPer22A, etc.). But like I said, I'm not sure if that's possible. Will need to check current draw on those chips.

[c_hegge]

Re: 5VSB Load Tester

of course

[tom66]

Re: 5VSB Load Tester

I would go for a 5.6V zener because 5Vsb is often high with low load. For example one PSU I have outputs 5.2V with no load.

Such a zener will probably melt under the might of an angry power supply though. There are a few posts on these forums showing badly damaged PCBs with those zeners completely toasted.

[budm]

Re: 5VSB Load Tester

You can implement the circuits as shown in the app notes here to protect it from over voltage on the output side.

[c_hegge]

Re: 5VSB Load Tester

I would go for a 5.6V zener because 5Vsb is often high with low load. For example one PSU I have outputs 5.2V with no load.

Such a zener will probably melt under the might of an angry power supply though. There are a few posts on these forums showing badly damaged PCBs with those zeners completely toasted.

If it's a small one, then yes, it will melt. If it's a bigger one, rated for over a watt, then it will just short internally and blow up the switching transistor.

[rogfanther]

Re: 5VSB Load Tester

My suggestion was to use a zener to control the supply voltage to the IC, not to regulate the 5VSB. If the psu changes the feedback voltage when under load, it is out of the control of the IC. Hence the zener, to try to keep it regulated and the IC working without going into shutdown mode.

[PeteS in CA]

Re: 5VSB Load Tester

In principle, you can zener-regulate anything, especially if you use a bipolar transistor or MOSFET to get current gain. And your output ripple can be extremely low. It's called a linear regulator. Two problems, though. First (and less serious), your typical zener is a 5% or 10% accuracy part (2% can be found, but are less common). Second, and more important, linear regulators are inefficient. If you want decent headroom for low-line conditions, the efficiency will be well under 50%. Even if you only do it for the +5V Stby, that's going to hit the power supply's over-all efficiency. Inefficiency also means heat, and you don't want extra heat inside your power supply or computer.

Zener or TVS diode clamps can work reasonably well, as long as you can tolerate the clamp voltage tolerance (5%?). TVS diodes don't have the cumulative damage issue MOVs have, but are faster (but handle less energy).

[rogfanther]

Re: 5VSB Load Tester

Well, as the idea was just to keep the voltage under 20V, something around 15V +-10% should work ok, provided it could dissipate enough wattage