Digital power factor correction

[370forlife]

Re: Digital power factor correction

Hmm. Something isn't right here. I don't believe that booster coil should get that hot. If they got that hot in power supplies you would definitely notice extreme heat coming out the back as well as discolored PCB. Often they wrap those coils with plastic tape, and I have yet to see any of that discolored or burned off.

[PCBONEZ]

Re: Digital power factor correction

Depends on insulation class.
http://www.engineeringtoolbox.com/ne...ses-d_734.html
http://www.pleo.com/ulsystem/eis_apply.htm

[everell]

Re: Digital power factor correction

Next is temperature testing using a Fluke 62 Mini IR Thermometer. I did not test the 955 uHy coil due to output voltage instability. Test results on the other two coils was pretty much the same, temperature readings being within about 3 degrees for each test.

At the minumum output setting of 274 volts, the input voltage from the bridge rectifier measured 107 volts. That means the booster coil was developing 274 volts - 107 volts equals 167 volts. After running five minutes, its temperature was 143 degrees F.

At output voltage of 330 volts, the booster coil developed 330 volts - 106.2 volts equals 224 volts. After running five minutes, its temperature was 210 degrees F.

At output voltage of 360 volts, the booster coil developed 360 volts - 106.1 volts equals 254 volts. After running five minutes, its temperature was 270 degrees F.

At maximum voltage output of 397 volts, the booster coil developed 397 volts - 106.0 volts equals 302 volts. After running five minutes, its tempereature was 317 degrees F.

As expected, the greater the output voltage, the higher the temperature. So my question is, how many years can a booster coil run at its designed voltage of about 400 volts when the temperature is running at about 317 degrees. I would think that a temperature that high would eventually cause the insulation to break down and get brittle.

[everell]

Re: Digital power factor correction

The coil being used in my first testing is a handwound coil measuring 457 uHy. Inductance measurements made with B&K 878 meter. I purchased two coils from Digikey, measuring 440 uHy and 955 uHy.

Replacing the coil with the 440 uHy coil, output voltages from 274 volts to 360 volts would stabilize in less than a second. Higher voltages up to the maximum of 408 volts would stabilize between one and two seconds. Part of this stabilizing time is caused by the time it takes the 12 volt switching power supply to stabilize. It powers the CS1500 chip. The important thing to notice about this test is that by decreasing the inductance value from 457 uHy to 440 uHy, the maximum output voltage would stabilize.

Replacing the coil with the 955 uHy coil, the output voltage would stabilize at the minimum setting of 274 volts. At the setting of 330 volts, the output voltage was fluctuating between 328 volts and 334 volts. At 360 volt setting and above the output would not stabilize at all.

I conclude that when replacing a APFC booster coil, do not increase inductance. Using the CS1500 chip it causes output voltage instability.

[everell]

Re: Digital power factor correction

I have been having a difficult time with this APFC add-on circuit. Most of the time the output voltage doesn't stabilize. It bounces back and forth between 100 volts and 400 volts. I began thinking that perhaps the Bestec ATX-250 12Z had some peculiar circuit design parameters that were causing the problem. So I took the circuitry out of the Bestec and put it in a CompUSA 250 watt ATX power supply. It also was very unstable, and after a few minutes blew the reverse voltage spike suppressor diode in the main switching circuit (1KV 2 amp diode!). So I took it out of the CompUSA power supply and built a stand-alone jig. I used the line input filter from an old gutted Antec power supply. I used a salvaged 12 volt DC power supply module to provide 12 volts to the CS1500 chip. I used a pair of 2K 50 watt power resistors in series as the load.

The basic circuit is from the data sheet. The input feedback uses three 1 meg resistors in series. The output feedback is using two 1 meg resistors and a one meg potentiometer so that I can vary the output voltage.

Powered it up. Voltage range: minimum voltage was 274 volts, maximum voltage was 397 volts. Turning it on and off many times, I determined that between 274 volts and 360 volts it stabilized very quickly. Between 375 volts and 390 volts it took 1 to 2 seconds to stabilize. Turning it on at maximum setting, it would not stabilize, but the voltage bounced between 165 volts and 370 volts.

I do not want to imply that the chip is of poor design - my circuit probably needs some tweeking. My main interest was to look at how the booster coil acts in this circuit. More later.

[everell]

Re: Digital power factor correction

I think its time to review a previous repair job which "smoked"

https://www.badcaps.net/forum/showthread.php?t=9660

At first I thought that the root cause was shorted MOSFET transistor in the APFC circuit. But the two booster coils (in series) were badly burned. If burn damage caused short or arc, then one of the MOSFET transistors would short.

I rewound replacement coils on undamaged cores. Power supply works fine, but those coils do get super hot.

[Th3_uN1Qu3]

Re: Digital power factor correction

Hey shovenose - I just have to ask........just what do you consider to be "normal"

Good one. Also, IMO power factor correction isn't really needed below 1kW or so. You don't pay for it after all.

[everell]

Re: Digital power factor correction

.....it wont reallly be cool unless it looks normal.....

Hey shovenose - I just have to ask........just what do you consider to be "normal"

In my opinion, APFC is "abnormal" progress. It doesn't really matter how pretty the case looks. If the circuit fails and smokes.....calling the fire department is NOT normal.

[Th3_uN1Qu3]

Re: Digital power factor correction

How hot is "hot" ? If it's under 100C, it'll be fine. Micrometals has a nifty calculator on their site, if you decide to bother winding your own coil it'll tell you everything, from wire gauge and length to temperature rise and expected lifetime.

It can also analyze existing designs and tell you just how hot that coil gets and why (core or wire).

[everell]

Re: Digital power factor correction

Here is the computer it will be powering. As you can see, there is no case for it! So why worry if part of the power supply is hanging out.

Besides, this was just experimental stuff. Not intended for long term usage. I am still worried about how hot that booster coil gets. The power supply may not last that long!

[shovenose]

Re: Digital power factor correction

wow!
but how is that going to fit in the psu case? it wont reallly be cool unless it looks normal or otherwise is the same from the outside!

[everell]

Re: Digital power factor correction

This evening I put it all together and fired it up. The output voltage using the suggested feedback resistor values came to 425 volts. A little higher than expected. I will do some further experimentation to see if I can reduce that voltage to something more like about 385 volts.

The dark picture shows the psu tester LEDs lit while operating. The three green ones are the three rails, 3.3 volts, 5 volts, and 12 volts. The amber one is the 5vsb. the red one is the power good indicator.

The new chip may function great for better power factor correction, but the BIG problem with this circuit....and ALL APFC circuits I have worked on so far......the booster coil gets super hot. That means either a burn up over time, or a shorted coil at some point in the future.

Since so few technicians are willing to take a power supply apart and determine WHY it failed, I predict that we are going to be seeing lots of power supplies fail because of the heat generated in the booster coil of the APFC circuit.

[everell]

Re: Digital power factor correction

There simply was not enough room on the main pc board to add a heat sink, two power MOSFET transistors, and a switching diode. So I built an outboard circuit with these parts.

[everell]

Re: Digital power factor correction

After a week and a half vacation in Tucson, AZ to observe the annual Feast of Tabernacles, I am back on the project. Did a lot of "cut and paste" on the circuit traces on the main pc board. Then built the add on board with the new CS1500 chip.

[shovenose]

Re: Digital power factor correction

Any updates to the project?

[PCBONEZ]

Re: Digital power factor correction

Congrats!

[Toasty]

Re: Digital power factor correction

Impressive. Most impressive.

[everell]

Re: Digital power factor correction

I was a bit concerned as to how well the A6351 chip would perform the 5vsb function when the startup voltage across the main capacitor is only 160 volts rather than the 320 volts used in the usual doubler circuit. So today I did the basic wiring to hook up bridge rectifier, main capacitor, and 5vsb circuit. On power up, the main capacitor charged to 162 volts. The A6351 came up just fine giving 4.98 volts on the 5vsb line, and powering the supervisory chip such that the "Power Good" LED came on as expected. The feedback circuit voltage measured 14 volts, plenty to power the main pwm chip, and hopefully the new APFC chip also. So this first test before wiring the APFC components was a success.

[shovenose]

Re: Digital power factor correction

dood how cool

[everell]

Re: Digital power factor correction

Toasty is right......another everell project has been launched. I am using a discarded Bestec ATX-250 12Z to convert from what it is to one having APFC using this new CS1500 chip.

So far I have removed the two main capacitors and replaced with the one main capacitor. I have wound a booster coil and mounted it. And I have added the big .47 Mfd filter capacitor across the bridge rectifier. Picture attached.

Next is some "cut and paste" of circuit traces on the bottom of the board. Then I will prototype the breadboard using the new CS1500 chip.