Super 5vsb mod for Bestec ATX-250 12E

[everell]

I have been wanting to try building a super 5vsb mod which will do 5 amps or so output (and survive!). I figured what better psu to use than the motherboard killer Bestec ATX-250 12E. Found one cheap on ebay, so here we go. First step is to remove both heat sinks and clean all that resistive glue off the pc board. So here is what it looks like stripped down.

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

Next step was to gut the old 5vsb parts from the pc board, make the replacement mod board using the NCP1216 pwm chip, attach the mod board to the heat sink, and reinstall the heat sink. My thanks to ldsishere for helping by making the pc board for this project. The original 5vsb power transistor which was attached to the heat sink was replaced with a power FET 20N60. I figured it would probably handle any amount of current this circuit could throw at it! Besides, it was a part out of my junk box.

[LDSisHere]

Re: Super 5vsb mod for Bestec ATX-250 12E

I was curious as to why you decided to go with the NCP1216 instead of the NCP1200. Is there something special about the Bestec that makes the NCP1216 a better fit?

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

According to the data sheets, the NCP1216 is just an "enhanced" NCP1200 with some better features. They are supposedly pin for pin compatable. So I decided to try the later version chip.

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

Next step was to build a add-on 5vsb indicator LED. You can see the mod on a small piece of perf board mounted to the other heat sink with all the output diodes. Then I reinstalled this remaining heat sink. This gives me a visual indication when the 5vsb is on.

[mockingbird]

Re: Super 5vsb mod for Bestec ATX-250 12E

Very nice...

I got an ATX-300-12EB3 that was made in early 2005. They actually put a PCIe connector on it but more importantly, it's already got the A6351A which I assume is for the 5VSB.

Any chance you can do a Fortron unit in the near future? (THA or THN specifically. Early revisions have no DM311 integration).

Respect to you LDSisHere on your custom etched boards... I'm still working on improving my technical understanding of electronics, then maybe I can buy some boards off you for various PSUs I have that need this mod.

[Pentium4]

Re: Super 5vsb mod for Bestec ATX-250 12E

That is awesome everell, thanks for sharing it looks great Keep us updated! this is probably a stupid question....how come you left those 3 optocouplers, aren't those part of the 5VSB circuit? or are those used for the voltage regulation on the output rectifiers?

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

Time to fire it up and see what it will do. I have an assortment of two 2 ohm resistors and four 10 ohm resistors. By using various combinations, I was able to do the following test:
Load resistance 5vsb measured voltage
no resistor 4.98 volts
10 ohms 4.95 volts
5 ohms 4.93 volts
3.3 ohms 4.90 volts
2.5 ohms 4.88 volts
2 ohms 4.86 volts
1.67 ohms 4.83 volts
1.43 ohms 4.81 volts
1.25 ohms 4.79 volts
1.11 ohms 4.76
1 ohm 4.74 volts
0.91 ohms 4.72 volts
0.83 ohms 4.70 volts
0.77 ohms 4.68 volts
0.71 ohms started at 4.59 volts but quickly began dropping to zero

No smoke, so turned it off and let it cool off for a few minutes. Turned it back on and nothing. Uh-oh......

So what components failed? Is it fixable?

[LDSisHere]

Re: Super 5vsb mod for Bestec ATX-250 12E

According to the data sheets, the NCP1216 is just an "enhanced" NCP1200 with some better features. They are supposedly pin for pin compatable. So I decided to try the later version chip.

When you get a chance, I hope you try a NCP1200 on this supply. I tried a NCP1216 on a supply (I do not remember which one.) and it did not seem to work as well on that supply as the regular NCP1200.

When you did your testing, how long did you run the supply at each setting?

[momaka]

Re: Super 5vsb mod for Bestec ATX-250 12E

Looks like it can do just about 5A before the 5VSB voltage goes out of spec on the lower limit (4.75V). Very impressive mod!

I think you just may have reached the limits of the 5VSB transformer. 5A at 4.74V is nearly 24W! I'm sure Th3_uN1Qu3 has a better idea than me of how much that transformer can handle, though.

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

After doing some high tech troubleshooting (probing with the DVM ohmeter) I determined that the diode in the 5vsb output circuit was shorted. I removed it for identification. It is a SB340. That means it is a shottkey barrier diode rated for 3 amps at 40 volts. Considering it reached about 6 amps before shorting, I would say it was a pretty good diode! I ordered some SB540 from Digikey, and the power supply is now back in operation with the same output voltage measurements. I ordered SB540 made by Vishay because it had the least amount of forward voltage in their specification. The lower the forward voltage the less the heat.

Lesson learned: If you are redesigning the 5vsb to do 5 amps or so, need to change the diode to one rated for that much current.

Stay tuned for the next test.

[Khron]

Re: Super 5vsb mod for Bestec ATX-250 12E

Live 'n learn, eh?

[ben7]

Re: Super 5vsb mod for Bestec ATX-250 12E

That might be why the output voltage sagged under high load. The diode just couldn't take it!

[PeteS in CA]

Re: Super 5vsb mod for Bestec ATX-250 12E

everell, an 8A or 10A rectifier would be a better choice, unless this is more an experiment than something you want to use for an extended time. Using parts at their full rated current is not a good idea unless you have really great heatsinking and a decent amount of cooling air. An 8A or 10A schottky will also have a lower V(f), resulting in better efficiency.

Another thing to look at, if you have an oscilloscope available to you, is the gate drive to your MOSFET. While it should have a nice, low, R(D-S On), you need to be sure it is turning on and off quickly. High current MOSFETs often have very high gate capacitance, resulting in either relatively high drive currents when being switched or relatively slow switching if the available gate drive current is too low. I'm not suggesting you have a problem, only saying it's something to look at as a sanity check.

[LDSisHere]

Re: Super 5vsb mod for Bestec ATX-250 12E

@PeteS in CA

Is there a non-oscilloscope way to determine if a transformer or inductor has reached saturation? In some of my experiments I think this may be the bottle neck in the output but this is only a guess.

[ben7]

Re: Super 5vsb mod for Bestec ATX-250 12E

@PeteS in CA

Is there a non-oscilloscope way to determine if a transformer or inductor has reached saturation? In some of my experiments I think this may be the bottle neck in the output but this is only a guess.

Things blow up or catch fire xD

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

The next issue to be addressed is the power (Vdd) to the NCP1216 chip. Taking measurements, the voltage increased from 11.38 volts to 15.60 volts. According to the data sheet, Vdd max is 12 volts. The other thing I noticed is that the 5vsb would start up when no load resistors were present. But after turning power off, then adding any load resistors, then turning power back on, the 5vsb would not come on. So.....

Back to reading the data sheet. This chip uses Dynamic Self Supply.

"Dynamic Self−Supply (DSS): Due to its Very High
Voltage Integrated Circuit (VHVIC) technology,
ON Semiconductor's NCP1216 allows for a direct pin
connection to the high−voltage DC rail. A dynamic current
source charges up a capacitor and thus provides a fully
independent VCC level to the NCP1216. As a result, there is
no need for an auxiliary winding whose management is
always a problem in variable output voltage designs"

So.....I should not be using the auxiliary feedback winding to provide Vdd power to the chip. So I removed the 110 ohm resistor between the diode rectifier and the Vdd pin on the NCP1216 chip. I also removed the zener between the Vdd pin and ground. Now all that remains is a 4.7 uF capacitor to ground. I added a 1K load resistance to provide some loading on the auxiliary feedback.

Power up, and adding the first two 5vsb load resistors gave readings of 10.95 volts (no resistor), 8.36 volts (one ten ohm resistor), 8.08 volts (two ten ohm resistors). Adding any more resistors gave reading of 6.55 volts.

Whereas the data sheet says that 22 uF is the typical value of capacitor used, I removed the 4.7 uF capacitor and installed a 22 uF. Now readings were 11.25 volts, 8.46 volts, 8.08 volts, and 6.53 for any additional resistors. So there was very little difference between using a 4.7 uF and a 22 uF capacitor.

So why were the first three readings higher before the Vdd stabilized at about 6.5 volts? This chip uses a "skip cycle" technique when the load is small. The amount of skip cycle is determined by the F/B voltage. Any time the F/B voltage goes below 1.4 volts "skip cycle" mode kicks in. When looking at the first three readings, the F/B voltage was below 1.4 volts. F/B voltage readings on readings afterward continuously increased from 1.67 volts to 2.61 volts.

My conclusion: not using the auxiliary feedback winding from the 5vsb transformer to provide power to the NCP1216 pin works correctly. Only thing needed is a 4.7 uF or 22 uF capacitor from Vdd pin to ground on the NCP1216 chip. I also noticed that I could turn off power to the circuit, add any number of resistors, then turn power back on, and the 5vsb would start up. The increased voltage from the auxiliary winding was therefore causing the NCP1216 chip to not start when under load.

Stay tuned for more exciting experiments..............

[PeteS in CA]

Re: Super 5vsb mod for Bestec ATX-250 12E

Things blow up or catch fire xD

That's about it, specifically and usually, the switch device. And measuring current with an oscilloscope isn't pleasant, either. Current probes cost $$. Hall effect type probes are under $1K, but their BW, that I remember, is about 100KHz. There are probes that are under $2K with BW >10MHz, but those are usually designed to connect to a certain type of scope. And there are others that are independent of the scope that cost $3K-$4K.

The other method with which I'm familiar is to put a low value resistor from source to return and use a voltage probe. You have to select the resistor to be low enough not to disturb the circuit too much, but high enough to get a useful signal (and noise will always be a pain).

If you get a usable signal, look at the ramp during the on time. If it's nice and linear, you aren't going into saturation. If it starts to curve upward more sharply toward the end of the on time, your circuit is starting to saturate. If the curve goes way high at the very end, prepare to order a new MOSFET from Digi-Key.

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

Another thing to look at, if you have an oscilloscope available to you, is the gate drive to your MOSFET. While it should have a nice, low, R(D-S On), you need to be sure it is turning on and off quickly. High current MOSFETs often have very high gate capacitance, resulting in either relatively high drive currents when being switched or relatively slow switching if the available gate drive current is too low. I'm not suggesting you have a problem, only saying it's something to look at as a sanity check.

I was reading about that very thing in the data sheet. So I removed my junk box 20N60 and replaced it with a brand new Digikey Fairchild 5N60 FET. Fired it up, and the Vdd on the NCP1216 was very stable across all load resistors at 11.12 volts. That's up a little from the previous 20N60 which had NCP1216 Vdd at 6.5 volts. Data sheet says it normally runs at about 10 volts. So I will leave the new 5N60 FET in the circuit.

Here is an Excel spreadsheet which shows the data in a more readable form.

[everell]

Re: Super 5vsb mod for Bestec ATX-250 12E

everell, an 8A or 10A rectifier would be a better choice, unless this is more an experiment than something you want to use for an extended time. Using parts at their full rated current is not a good idea unless you have really great heatsinking and a decent amount of cooling air. An 8A or 10A schottky will also have a lower V(f), resulting in better efficiency.

With my purchase of the SB540 diodes, I also purchased a couple of ten amp SB diodes in TO-220 case. As a further experiment down the road I plan to mount one to the output heat sink. That would give added current ability and probably be a lot cooler.