PC PSU conversion

[oztules]

Think I have recovered now.

If you build another one you will have three. If you put them on your three phase motor conversion, you may give a well matched load to the conversion and get your efficiency up and heat down. For 48v batts , perhaps series the outputs (float the earth away from chassis where the board screws to the case (small "O" ring's will be sufficient) and set each psu for 19v. or 9.5v for 24v etc.

I have found that voltages above 27v are not really attainable without a rewind. For low input voltages, a rewind and higher current switchers will get a reasonable unit, and make use of the inbuilt voltage doubler in the front end. As standard, it makes a reasonable fist of it at voltages as low as 50vac. This characteristic may make it plumb for booster without sensing, as at low rpm/voltage the conversion will be poor, as voltage raises, so too will conversion output.... this is conjecture on my part though. I will test as soon as a mill eventuates.

The total ease that you can control the voltage and current makes these things very handy, and a bank of say 10 of these putting out 2000w total costs nothing..... don't get better than that really.

.........oztules

[dinges]

"The next step is setting up the control loop for mppt [...]"

Funny that; I've been thinking about exactly the same thing lately. Looking at this schematic, I get the feel that such a switch-mode PSU would take much of the hard part of building a MPPT  out of it. I assume you are familiar with this discussion ? :

http://www.thebackshed.com/windmill/FORUM1/forum_posts.asp?TID=753&PN=1

Learnt a thing or two about how critical those inductors really are. But, using an old PC PSU and a bit of extra control. Current sensor, a few integrators. Or a PIC or some similar microcontroller. Hm...

Perhaps there is an EE student somewhere looking for a graduation project ?  

Peter.

[dinges]

I'll hook up one of them PSUs to a variac and do a quick experiment. Checking output voltage and current at various input voltages. Quick test that should provide quite a bit of useful information.

Peter.

[commanda]

If you connect a resistor into the voltage reference circuit, so an external voltage (say 0-10 volts) can control the output voltage, you can use my analog mppt circuit.

http://www.fieldlines.com/story/2006/3/12/14840/1315

If you want the mppt circuit, just drop me an email.

I'm currently setting up a test jig so I can test it with the latest iteration of my universal dc-dc converter.

Amanda

[oztules]

Yes Peter,

have browsed the forum topic at backshed, and lost a bit of interest as they seem keen to explore the micro route. I am of the mind that an analogue version will do the job sufficiently well if designed well.

Enter Commanda.

At this point I think you would do well to email her as she suggested, and take a look at her mppt circuit. This will simply integrate into the psu. I will be interested in your variac tests. One suspects that with some higher voltage switchers, and caps, (for hv machines) a simple and scalable mppt of the kilowatt and above range will not be terribly difficult. If Amanda's foray into mppt via the analogue route is successful, this may well prove to be the easiest and most stable (we know it doesn't blow up once the current system is perfected), way of controlling the mill's output.

In fact I have been unable to blow one of these up once the current regulation system is implemented, they seem pretty bullet proof. So all the hard work of the switcher has been done for us, which obviates most of the discussion at backshed, and if Amanda is successful with her tests, I think mppt for the masses is close to a reality. All the high tech stuff (power switching) is done.

The only other consideration will be the overvoltage crowbar if required, but this is simple to implement. The AWP HV mill here tops out at about 550v unloaded. We have overvoltage set for 500v on it. (it then dumps to a 3000w radiator array of standard bar radiator elements) Loaded it goes to about 280vac.

While your in the mood to blow these things up, it may also be worth trying much lower value caps on the input (10uf instead of 200uf etc.). This will force the pwm to track the incoming sine waves better, and improve the power factor. Whether this works on this topology or not I have not tested,(I have used it on push pull topologies like the one in my gallery) but if it does, it will improve the efficiency as well.

Plenty to play with.

..........oztules

[dinges]

Oztules,

I would like to nominate you for the Fieldlines Innovation Award 2007 for the PC PSU conversion idea

Just finished up the 2nd PC PSU. This one was a breeze compared to the first one. Voltage modification was easy, no snags. No overvoltage protection that tripped either (strange...).

Current modification was easy too this time. I tried something else, btw. Instead of taking the centertap of the transformer out of the PCB, I simply added a wire to it halfway up. See picture below:

Mechanically much more solid and, in this particular PSU, I got enough voltage feedback to make it work. The current potmeter is 100 ohm (the only suitable 10 turn potmeter I had around), so I had to modify the 47k resistor to 15k. Current control worked perfectly from the start, giving a range of 250 mA to 11 A. No strange effects as in the first PSU I converted. In short, this PSU was a breeze.

You can short it and nothing happens; as the short is removed it simply starts itself up again. Image below shows it shorted at 11 A (shorting from an open-voltage situation of 24Vdc)

Once I accidently shorted the output to the PSU case (instead of the ground return lead). The PSU turned off, but reset itself on removing power. Impressive.

I've installed two multiturn potmeters, one for voltage control, the other for current.

The sky's the limit...

Also:

I tested another (unmodified) PC PSU with the variac. This one was for 220V only (no way to switch it over to 110V). I loaded it with a 12V/10A 120W load. When turning up voltage, it starts working at 110Vac. At 140Vac it puts out the full power (12V@10A). It continues to work normally till 260Vac (the upper limit of the variac).

Notice that it starts working at 110V, however, when turning voltage down (from 260 to 0Vac) it continues to work till 140Vac at full power and continues to work with reduced power till 50Vac. Then it turns fully off.

So in effect, we have a PSU that works normally from at least 140-260Vac, and, depending on the situation, works down to 50Vac (but can't start up at 50Vac; needs at least 110Vac to start and 140Vac for full power)

Since this PSU had no external switch for the doubler I couldn't test that, but extrapolating, I'd expect it to work at 50/2 = 25Vac, 110/2 = 55Vac (starting up) and 140/2 = 70Vac (full power).

Quite a dynamic range, but not sure if it's enough. Especially the switchover with/without doubler may be a critical thing; not sure how the PSU reacts to it being switched over whilst powered. I expect it to be designed to be switched over unpowered.

Peter.

[dinges]

For another project (battery tab welder, capacitor discharge welder) I needed another big (~10A) variable voltage power supply. What better thing to use than an old PC PSU...

Thus, I present youse with PC PSU nr. 5 (nr. 4 was started earlier in the day but, ehm, let the magic smoke out; it's now sitting on the workbench to be fixed)

As usual, the voltage modification for this project was pretty easy. Initially it was only adjustable from 5-24 V, but some messing with the reference voltage divider solved that problem. Replacing one resistor yielded a variable voltage from 2.5-24V. Replaced the usual capacitors, removed unnecessary components (amazing, I have an ash tray filled with removed components and the PSU still works )

Current modification was harder (and the reason nr. 4 blew up). I seem to only have PC PSUs that use the 2nd opamp of the TL494. Decided to disconnect it and free the opamp for my own use (current control). This worked without problem. I grounded pin 16 to the nearest ground, added a 47k resistor, 330 ohm and 10 n network to pin 3. Added a pickup-wire to the transformer. But, for some reason current control didn't react properly to the feedback voltage.

Took me a while (and one temporarily deceased PSU) to find out what was the problem. Apparently, my ground wasn't right. Connecting pin 16 to another ground point (at the black wires, where power is taken out) solved the problem.

Added a 180 ohm resistor (connected to the white wire in the photo, with the black heatshrink tube over it) to limit current to a safe value of 8.5 A. It works perfectly now, short-proof variable voltage power supply to quickly charge the capacitor bank of the welder.

So, for anyone else trying to modify PC PSUs and having trouble getting the current control to work; try another ground point for pin 16. I bet that would have solved the problem in the very first PC PSU I had modified too (see beginning of this diary). It can be seen in the picture below that the black wire (connected to now-removed R51, which connects to pin 16) goes to the ground near the output wires.

During the process I managed to let some more magic smoke out (the 2 switching transistors (2SC4242) blew and were replaced, as was the input rectifier (4*1N4007). Replaced the soldered fuse with a fuse holder for easy exchange of future fuses (I know myself too well...) and added a connector for high-voltage input, as the wires were originally soldered to the board.

That's it. PC PSU nr. 5, ready for use. With this PSU it shouldn't take more than a few seconds to charge the capacitor bank.

Peter.