PC PSU conversion

[dinges]

After the constant nagging from a certain-not-to-be-named member of this board (besides, Oztules wouldn't want me to mention his name) I figured that it would never end unless I simply converted a PSU.

So, especially for that not-to-be-named member of this board, here goes.

The victim:

A 200 W PC Power Supply Unit (PSU), old style (not an ATX), with a power on/off switch in the grid line. None of that fancy 'sleep mode' stuff here. Off is off.

The victim opened up for surgery:

Some people say it's a piece of cake to convert those (Angus, you reading this ?). Must be because they've never done the surgery. Personally I -hate- working on switchmode PSU. Guess I've had too many FETs blow up in my face. One small mistake and -poof-...

Ok, what needs to be done ?

• overvoltage/current protection needs to be disabled (the hardest part)
  
• voltage regulation modified (potmeter adjusted '12V', range 2.5-20 V)
  
• all relevant electrolytics need to be replaced by higher voltage ones (with low ESR)
  
• current regulation needs to be added (likely the simplest modification)
  
  

For starters I decided to make a schematic of the relevant parts. That always helps to understand what's going on.

Notice that each PC PSU is different; another one that I modified about 2 years ago from 12.0 to 13.8 V was slightly different. Hence, a schematic is useful mostly for the specific unit at hand.

First modification: Modify the voltage regulation to 2.5-20 V. This was relatively easy; there's already a 500 ohm pot in there that adjusts voltage from 10-12 V; removing some components and replacing by a potmeter and resistor got that part working. However, when adjusting the voltage above 9.0 V (on the 12 V line) the PSU tripped. Turned out that adding a 20n/100V capacitor over the wiper of the potmeter solved that issue.

By then I could adjust to 12.3 V (5.99 V on the 5 V line) before it tripped into overvoltage protection. After fiddling around, taking out some resistors, etc., it turned out that simply removing the SCR (part of the overvoltage protection circuit) solved that problem. By now I could adjust the voltage from 2.47 V to 18.85 V (the design goal was 20.0 V; I'll have to reduce the 7.5 kohm resistor a little to increase its upper range.

Then the first test under load: a 12V/4.5A H4 car headlight. My favourite testing dummyload. When the bulb was still cold it flashed on briefly before the PSU tripped into overcurrent protection. Quickly unplugging and plugging it in again helped it to start up (the filament of the H4 bulb was by then heated enough so its resistance wasn't so low anymore). Adjusting voltage from 2.5 to 15 V worked perfectly, but below 9.5 V the PSU starts audibly whining; the frequency of it goes down as voltage goes down.

One con of this modification: the fan takes its voltage from the 12V line; when it's adjusted to below 7 V the fan stops; above 12 V it spins too rapidly.

My current idea is to house this variable PSU, together with a 'plain' PC PSU, in one case. That would give me regulated +3.3V, +5V, +12V, -5V and -12 V. The PSU I've modified will be able to provide any variable voltage. Add a few meters for current and voltage and we're finished.

Next tasks:

• replace all elcos by higher voltage ones
  
• add current control
  
  

The current control isn't that important for me (this is my first variable voltage power supply in 25 years of tinkering with electronics; my previous supply was an old 'golden' PSU from an Apple II, which puts out +5, +12, -5 and -12 V; so far, it did everything I ever needed...

But I figure if I don't add current control Oz will likely keep pestering me. For me it's a bit of a luxury though.

Higher resolution pictures can be found in my gallery:

http://www.anotherpower.com/gallery/dinges?page=19

http://www.anotherpower.com/gallery/dinges?page=20

So far so good. Still alive, no blown up FETs or ferrites. Who knows. Maybe Oztules was right...

Peter.

PS: just for clarity, this is not a beginner's project.

[dinges]

Update:

Elcos have been replaced by 25V ones

Voltage divider has been modified; now adjusts from 2.5 - 22.4 V

Not too bad.

Also found out that, despite what it said on the PSU case that there was an electronic fan control, none was present; the part of the PSU that had been clearly reserved and marked for a fan controller board was ominously empty; the fan wires simply going to 0 and +12V.

Peter.

[Slingshot]

Peter,

Just a thought here - since you were able to successfully raise the 12V output by nearly 100%, is there any inherent blockage to doing something similar with the 5V output?

If you could reprogram the 5V output to something more like 10 (fixed only, no variation needed), then you could run the fan off this and not be captive to fan shutdown when you want to use your main output at low voltage.

Another option might be to connect the fan across the +5 and -5 outputs, if the current required doesn't exceed the -5 capability.

[oztules]

Slingshot ..."Just a thought here - since you were able to successfully raise the 12V output by nearly 100%, is there any inherent blockage to doing something similar with the 5V output?"

The problem with trying to do this is that the 12v and 5v winding are from the same secondary.... ie the 5v is a pair of taps on the 12v secondary.

The transformer is wound like this

first layer is half of the primary

next layer is first part of the 12 secondary (7v) in single strand, a tap off and then 3 strands to continue to the center tap of the secondary

layer three starts as three strands for 5vand tap out, then continues as 1 starnd to the 12v out.

Fourth (outer layer) layer is the second half of the primary.

So it gives us a transformer where the 5v and 12v are the same winding (but differing "thickness" part way through). The primary is wound both inside the secondaries and outside the secondaries to achieve good leakage quality... tight coupling.

The tl494 derives it's power from a rectifier directly from the 12v secondary, seperate from the main rectifier. This allows you to drive the secondary down to very low levels, but still get 12v to drive the tl494. If current overload is put in place you can drive the device at short circuit (full current out..... useful for low impedance start up of motors etc), but the ripple is sufficient to still fully power the tl494. maybe the fan as well haven't tried. But at 2v or so, the fan is probably not required anyway.

.........oztules

[oztules]

Peter,

I am pleased you have finally done part of the project, but without the current control, it is just an accident waiting to happen. I suspect that low voltage cut out worked for your test light. I am guessing your load tried to draw very high current initially and shut down the chip from undervoltage, rather than current o/load.... that is my guess anyway.

If you over current at higher voltage, you will probably end up with meltdown of the switchers.

Also it is just another prickly power device which may start the load and may not. Current control will stop complete shut down as the transformer will still work even at controlled short circuit, without current control to control the current at short circuit, this may not be the case.

So bite the bullet, lift pin 15 out of the board, pull the centertap out of the board and add a piece of wire, a trimpot, and a few resistors and a cap and make it less feral.

Once you add the current control (only 3resistors and a tiny cap anyway) you have a perfect and controllable low impedance power supply, where you can limit the current (say for led driving or sensitive circuits you wish to test etc). At the moment you will be able to get high transient currents but yet still have difficulty driving even you 4.5A start load.

...... whinge.... wine grizzle...

There are you happy now.......... oztules

[bigkahoonaa]

Hi Peter

I'm glade to see you didn't vaporise yourself (yet).  Not a long and boring story, for me anyways (don't know about that not-to-be-named member).  About how long does it take?

Mau

[dinges]

Oztules,

I figured you wouldn't be satisfied till I went all the way. That, plus the fact that your arguments make sense...

So yesterday evening I started the current control modifications. Removed the KA7900, bent pin 15 upwards, reinstalled it and added the other components. Installed a ~ 2" piece of wire in the transformer return.

At first I thought the current control didn't work; with my H4 bulb loading the PSU at 5 A I tried to reduce current, which it didn't seem to do much at all (from 5.0 to 4.7 A, maximum adjustment possible). Then, I noticed that as I turned up voltage higher, that current stayed constant. If I adjusted voltage to 18V I could control current from 5A to 6 A or so.

So, the next step is to modify your component values to increase adjustment range; my first ideas are to increase the 47 kohm resistor to, say, 220 kohm, and to increase the length of the sense line to 4" (think I'll do that first).

Bigkahoona, you've heard Oztules say it's a 45 minute operation, and most likely it is for him. He knows his PWM PSUs. For me, it took about two hours of circuit tracing and reading datasheets (TL494), then carefully modifying for limited adjustment range (2.5-12V), then for higher range (2.5-22V) which needed disabling the overvoltage protection. Now the current control is being worked on, guess I've got about one hour stuck in that so far yet.

So by the time it's electrically finished I expect 5-6 hours into it. Then 2 more hours boxing it up nicely and adding a small 12V power supply for the fan so it would run independly of the main PSU voltage.

Not rocket science, but not a 45 minute project for me either

You may want to read this link too: it's the original discussion I had with Oztules on the topic a few years ago:

http://www.fieldlines.com/story/2006/4/24/92132/1267

So far it went better than I expected. Sometimes I just need a little exhortation to get going...

Peter.

[dinges]

All original outputs I have left in place: the +5V (now 1.2-10V), +12 (now 2.5-22.5V), -5 (likely -1.2 ... -10 V now; haven't measured) and -12 V (likely -2.5...-22.5 V; not measured). All these output voltages are controlled by the one single potmeter I've added.

There's one downside: ONLY the +12V (now 2.5-22.5V) line is being regulated. This means that, as I draw power from the other (non-regulated) lines, voltage might drop. Only in the case of the +12V regulated line would any voltage drop be fully compensated for by the PSU.

Regards,

Peter.

[ghurd]

Too simple?

What about a 5V fan and 7805?

What about an AC fan, connected the the mains input?

G-

[boB]

Peter, (I won't use your name here),

Since you've gone this far, just wind another transformer or add a tertiary winding

to the existing one to run the 12V fan.

boB

[oztules]

Weee.... He's done it..... good stuff.... see it does work.

"So far it went better than I expected. Sometimes I just need a little exhortation to get going... ".. or a year or two and a red hot poker up the nether-regions.

Perhaps just decrease the length of the sense wire so that the curent return voltage is less, that will give the present setup a smaller voltage to work with.

I usually leave the scr in circuit to catch any evil undervoltage on the tl494. (I don't recall it triggering in normal operation) I normally get rid of all the non 12v stuff on the board (including the extra coils on the filter... although they may help in transients as they can return some voltage into the 12v filter wire (act as tranny) although this effect may be more imaginary than real.

Bigkahoona, ..Peter is a perfectionist, so it took him longer (ie he removed the 494 to bend the pin up, i just heat and yank it out of the board, he draws the circuit, i just modify the input resistor to the voltage error amp. He would carefully take out the extra components, i just cut em off with the knippex cutters..... etc. I only replace the 12v electrolytic and cut the rest off, he replaces them all (and probably calculates the correct value)..... one of us is an engineer, one of us is an animal.

An interesting distraction... and it's only Peters diary I'm messing up..

I fixed up the automatic door opener and some helmet intercoms for an elderly gentleman over here yesterday, and as I wouldn't accept payment for it, he went on to tell me he had an Ultralight. He has just come back from flying over most of Australia with it. (Ayers rock ... Darwin, back through the outback picking up fuel from the cattle stations etc........ He is over 80 years old. Apparently it was news worthy, and it was covered on the networks (I didn't see it sadly)

His father was the first person to build an airplane in this country..... in 1910. A very interesting person to engage in conversation actually.

Anyway, having fixed a few things for other people with ultralights, I had been offered (almost begged in fact) to be taken up for a fly around the island, and always the powers that be (the boss) has refused to let me go. This time.. with over 100 years of flying experience in his family, I'm allowed to go. So as I type this, if the weather holds up, he may fly over and land at the front door and I'll go for a fly with him. Not too many folks can claim the aerial history this fellow has.

If you never hear from me again, then you can assume that he didn't make it to 85years old.

All the best

Now I think about it, She will only let me go with the oldest bloke flying an ultralight that we know.... wonder what that really means.......

........oztules

[dinges]

Update:

I think I've let the magic smoke out...

Whilst playing around with the current control part, it seemed to work (sort of); the current adjustment range was still very limited with just 1" of added wire and 4k7 (instead of 47 k) resistor to pin 14 (reference, 5.0 V). Only a very small part of the 2k potmeter was used (lower 100 ohm).

Whilst deliberately shorting the outputs it played possum. Unplugging and plugging it in again didn't help anymore.

But I'm wondering: as the output of opamp 2 (current control opamp, pin 15(-) and 16(+) goes UP, the TL494 starts to fold back. So, in order for current control to work, shouldn't pin 15(-) be tied to ground and pin 15 (+) be driven by the positive current feedback voltage (i.e., more current through the transformer, more drop over the line, sample point becomes more positive w.r.t. gnd, and pin 16 should go more positive (as opposed to pin 15) to reduce current again and close the loop ? In that case, couldn't the sample resistor be placed not at the transformer, but in the outgoing ground line (where pos. and gnd exit the PSU for use) ?

Either that or I'm missing how this circuit really works.

In this link (40 kB drawing) is the circuit as I've built it, but again, the way I understand it, pin 15 and 16 should be reversed for a closed loop with negative feedback.

http://www.anotherpower.com/gallery/dinges/PC_PSU_schematic?full=1

Oztules (or anyone else that knows how to deal with the TL494), I'd be interested in your opinion on this.

Pity though; in the past few days I already got used to having a variable voltage power supply on the bench (even though current control wasn't working yet).

Ah well. There's already another victim lying on the workbench (and if that one gives up the ghost too, there are a few more waiting for their butchery, ehm, I mean, surgery).

Peter.

[dinges]

correction:

"But I'm wondering: as the output of opamp 2 (current control opamp, pin 15(-) and 16(+) goes UP, the TL494 starts to fold back. So, in order for current control to work, shouldn't pin 15(-) be tied to ground and pin 16 (+) be driven by the positive current

[oztules]

Peter, your reasoning for the operation of the tl494 is fair enough....... however, ground in this case is more positive than the feedback voltage ( look at how we have achieved the voltage..... we take the voltage from below ground ie more negative than ground. Ground in this case is 2" above the sense voltage. The wire is a voltage divider, the ground point is at a slightly more positive condition than the connection between the sense wire and the lifted tap output. As the current rises ground becomes more positive in relation to the tap/connection interface.

So if we tie pin 16+ to ground, pin 15- becomes more negative with respect to ground and overcomes the positive bias we put on it via the 47k resistor to vref.

Originally pin 15 was tied to vref. This held the op amp in off condition as the 15- was above 16+ If you look at how we generate a pulse width from a triangular wave beaten against a fixed voltage through a comparator, we find that the pulse width is related to width of the triangle at the voltage point of the fixed voltage..... graphically.... draw a triangle, and move a horizontal line up and down it. The max width is when the horiz line is at the bottom of the triangle... not the top. So with no control voltage (original condition before butchery 16+ gnd and 15- at vref+, the output from the pwm would be at max. (ignoring the voltage control for the moment).

In the new case we positivly bias the 15- pin via 47k from the vref. This will mimic the original wiring.... now we hook up the below ground voltage from the transformer tap. As voltage through the sense wire raises, pin15- becomes more negative (proportional to the current through the sense wire) with respect to pin 16+.

This turns on the op amp,and the horizontal line we drew rises and so the pulse width decreases...... clear as mud really

If I have made this difficult, let me know and I will try again.

You are reading this drivel because the ultralight didn't fall from the sky........ it was a buzz.

.........oztules

[oztules]

I think your 4k7 pulled you undone. use 47k and a 1.5"wire and try that .. only the switchers should be blown, just replace and re-run. If you lenghten the wire, you will lower the output. if you use 4k7, you may overpower the correction voltage from the sense wire..... just a thought.

.........oztules

[dinges]

Thanks for your thoughts, Oz.

I measured both the diodes and the FETs, they didn't measure shorted. I never liked this PSU in the first place; just the act of putting the DMM probes on it caused it to go into shutdown mode often. The first one I converted (from 12 to 13.8V; no current mods) was much easier to work with.

I'll simply try on another PSU. Have about 2 dozen of the things lying around.

That's the part I don't get: that the voltage on the sample point is more negative than ground. I would think it to be more positive (should've measured it with the DMM though; neglected to do that). And the fact that the design more or less worked (though not with the current range I wanted) seemed to prove your design was ok.

Let's for a moment assume that ground (on the PCB) stays at ground. We lift the return wire of the transformer and put in 2" of our own wire in between that and ground. Now, where our extra wire is soldered to the PCB-ground is actually ground. The other end of the wire (attached to the transformer return) becomes more positive as current increases (I would think), but apparently, you say it becomes more negative ?

Another question: how much current adjustment range do you get on your PSUs ? Can you adjust from, say, 100 mA to 10+ A (it's mainly the lower current range I'm interested in; does it work down to a few 100 mA or so ?)

I think I understand how it works, except from the actually sampled voltage; positive or negative... (yes, I believe you when you say it's negative, I just don't understand why)

Good to see the flight was a succes. That's the first step. Now, it'll be much easier to convince the wife to let you fly a 2nd time. And a 3rd time. And from then, it should be easier again to ask for permission to get flying lessons. Tell her about all the benefits an UL would have for her too (grocery shopping ?). And those UL aircraft really aren't -that- expensive...

Peter.

[oztules]

Sorry about all these added posts, but on thinking (difficult for me), if you were only using the first 100R of your pot, you were definately living on the edge. If you had kept the original wire length, and just increased the trim value to say 5k you may have had what you wanted. If you only had control with 5% of the trim pot with your 4k7, you had virtually no control. Over current was just a whisker past 100R.

You may be getting misinformation from your current readings because of load change.. ie you think you have current control under control, only to find out that the test load was not capable of loading the current up due to voltage considerations. ie if you used a 4.5A globe, you may have felt that you were limiting at 5A only to find that with a lower impedance load (read short), you were really not limiting at  that point at all....ie you had the load limiting the current, not just your current control.... probably what happened here.

........oztules

[oztules]

Hi Peter,

Yes, the voltage differential is difficult to see clearly... until you lose the firm grip you have on gnd=0 .... the end of the emf generating device is zero to full potential... anything across it is in circuit from it's full potential down to zero at it's terminals like this:

I suspect by now your kicking yourself...... I know you won't be surprised to know that somehow at some lower level.....it  makes me feel better.

I will go out and test the lower regions of the psu I have in the shed. Will report back to you re: min current and any changes to get that definition..... prob longer wire and bigger trim pot.

Upside down current is easier seen below the equator I guess.

Here's a pic from yesterdays flight of one of the offshore islands (just off my beach)

.......oztules

[oztules]

Hi Peter,

Yes, the voltage differential is difficult to see clearly... until you lose the firm grip you have on gnd=0 .... the end of the emf generating device is zero to full potential... anything across it is in circuit from it's full potential down to zero at it's terminals like this:

I suspect by now your kicking yourself...... I know you won't be surprised to know that somehow at some lower level.....it  makes me feel better.

I will go out and test the lower regions of the psu I have in the shed. Will report back to you re: min current and any changes to get that definition..... prob longer wire and bigger trim pot.

Upside down current is easier seen below the equator I guess.

Here's a pic from yesterdays flight of one of the offshore islands (just off my beach)

.......oztules

[dinges]

Yes, I figured you got a wicked sense of pleasure out of that explanation.

What's worse, it even makes sense. Thanks.

I'm pretty sure the current control worked at higher load though, and that it wasn't limited by a too-high impedance load.

Not sure if a 5k potmeter would have made a difference, as I only got to use the lower 100 ohm (0-100 ohm) part of the 2k2; with a 5k resistor it would still be that small part I'd be using ? In effect, it would be even a smaller adjustment angle on the 5k pot (about half), I think.

After rummaging through the stack of PC PSUs I found only three other XT PSUs. One I want to keep in original condition, the other two I opened up. Turned out both of them already use the 2nd opamp (pin 15/16). Tried modifying one of them just for adjustable voltage but that wasn't trivial either (i.e. only limited succes - 6..15 V). I didn't bother any more since I couldn't add current control to it anyway.

So, I'll be on the lookout for some more, plain simple XT PSUs in the near future. I did have a look at a (generic) schematic of an ATX power supply and noticed it wouldn't be any harder to modify one of those. We'll see. This story hasn't ended just yet.

Your story of the 85 year old pilot reminds me of one of my old (now deceased) glider instructors. He was well into his '80s back then, a former physics professor. A great guy to just hang around and talk with. But, as an instructor, he earned the nickname 'the autopilot' by us youngsters: whenever he told someone to make a left turn, he would, subconciously, make the left turn himself (the student could just let go of the stick and the glider would 'magically' fly itself ) and then compliment -you- on your nice, coordinated turn. The same would also happen during starts and landings Incidentally, now that I think of it: he was also the one that let me go solo to the surprize of just about anyone in the club.

I know you're very proud of your island but really, there's no need to post the same picture twice.

Nearly forgot: after I showed Lenneke your place on the map, she said she'd only let me visit if she could come too  

Stand-by for updates on the PC PSU story.

Peter.

[dinges]

...And another one bites the dust...

Big blue flame, dramatically accentuated by the sudden darkness of the lights going out.

Peter mutters bad words w.r.t. Oztules' intelligence as he walks to the meter closet to reset the GFCI. '45 minute job', he said. 'Easy as pie'. Yeah, sure. More bad words leave Peter's mouth. This time remarks about how one should never trust an Aussie. After all, the population of Australia was hand-selected by the finest judges in Old England.

Peter walks up to the attic again, looking for another victim. 'Hm, that stack of power supplies is rapidly getting smaller', he remarks, as he lays his hands on another victim. Was it his imagination or could he hear that PSU begging for mercy: 'Please, don't take me' ? Nah, that was probably just his imagination, Peter thought. As he walked down the flight of stairs again he caressed the PSU, saying 'tomorrow it'll be your turn to be converted into a powerful laboratory bench supply; aren't you thrilled?'. Of course, anyone knows electronics don't understand Dutch. His talking was in vain, but at least it made Peter feel better. 'Tomorrow', comrade, 'tomorrow...'

Peter.

PS: Strangely though, of the 4 PSUs I've opened, only one (the first one that I got pretty far with) had an unused opamp 2. All the rest use that opamp. Strange.

[oztules]

Peter,

"Yes, I figured you got a wicked sense of pleasure out of that explanation.

What's worse, it even makes sense. Thanks."

Yes.... insert evil grim here  

"I'm pretty sure the current control worked at higher load though, and that it wasn't limited by a too-high impedance load."

If you think about it, with no feedback, the pulse width will only be limited by the voltage comparator. If you only saw 5A or so, I can only assume that the load was not capable of drawing any more current. If you had disconnected the sense, pin 15- would be high, (via 47k) 16+ low and comparator would be turned off for all intensive purposes. (no current limit)

If you only got currrent control in the first 100R, then you would have to think that the neg feedback ran out of use at 100r..... or the load was not heavy enough to generate any higher feedback voltage.... my choice. Remember, low feedback voltage.... high current out. As you go past the controlled 100r and increase, you should have gone to maxcurrent that the load can draw. Thats why I think that the load was not big enough to draw more current... as the comparator was in a condition of 15- higher than 16+. Current is only adversely affected when 15- goes neg  compared to 16+. Thats why 5k would have made the difference to higher current settings. It would decrease the chance of 15 going neg wrt 16 and so higer current.

"Not sure if a 5k potmeter would have made a difference, as I only got to use the lower 100 ohm (0-100 ohm) part of the 2k2; with a 5k resistor it would still be that small part I'd be using ? In effect, it would be even a smaller adjustment angle on the 5k pot (about half), I think."

Following the my above argument, the higher value the pot, the less likelyhood of decreasing current out from the unlimited original condition . So remember, less feedback voltage, higher current out. I think your thinking about it backwards (north of the equator thing) So like I postulated originally, if you ran out of "more current after 100R" it must be because your load couldn't draw any more at the voltage you had set.... other wise it would have after 100r as it would be harder to drive the pin 15- out of it's full bore positive bias from the 47k.

I think you may be looking at it all back to front.

I have a sneaking suspicion that another evil grim could be lurking around the corner....

Yes the double post... I thought I had hit preview, but apparently not.

"Nearly forgot: after I showed Lenneke your place on the map, she said she'd only let me visit if she could come too  "....... she is welcome to leave you behind you know and come on her own

Yes, John Duigan ( i think is is proper name) is a true pioneer type character, It wasn't till we landed that he told me he had force landed over ten times previously... three involved seized engines, and he has layed claim to several fences and trees about the place.

He is a natural thrill seeker, when he took off, he lined the fence up from only 40 or so yards, taxied straight at it and cleared it by about 2 feet..... just to show me that his plane could take off in a short space.... he could have used 2km in the other direction.

Here he is lined up with gate in the fence.... he did clear it .... just

[oztules]

Dear Peter who used to own lots of psu's

Looks like a 50 watt incandescent light in series with your  240v line might help mitigate your disasters.

When your ready to cry uncle, I will send you a board that works....evil grim is starting to gather up in facial muscles about now.

Try to keep the scr in circuit, I found that helps slow up the  stack of burnouts. Get a good clear picture in your head about how the comparators work.... ie up is down left is right bad is good etc.

In their rest condition, they are full bore. ie the voltage comparator will tend to max held down by the feedback voltage, lose the feedback and instant full pulse width, same with the current comparator, loose feedback voltage and instant full pulse width. (look at your argument aginst 5k pot?? it don't add up)

The first of these two comparators to get into a limit condition, will take control, the other will be in unlimited mode until circumstances change and it is in a limiting situation. ie set for 10v and control current, or set for 10A and control the voltage. depending on the load. I have had mine as low as 160ma. but ripple to drive the 494 caused instability at these current levels at too low a voltage. Conversely, at high current, could lower voltage to 1v, as the current ripple was just enough to keep the 494 alive. Seperate tiny 12v tranny to provide for the 494, and your limits are 0-whatever you can stand. For 12v battery charging (what I developed it for), you can use any current/ voltage setting you can use, as the 494 is kept alive by the battery.

Have fun

.........oztules

[oztules]

Oh and here is the proof he got away... look at the distance he had from the last pic to the gate.

[dinges]

"Nearly forgot: after I showed Lenneke your place on the map, she said she'd only let me visit if she could come too  " ....... she is welcome to leave you behind you know and come on her own

Well, in that case I'll have to disappoint you, Oztules, as there is no Lenneke in my life. If I visit it'll just be me.

I took the liberty of abusing this diary entry a little. You know how in electronics we can often test a system by doing a stimulus-response test ? You put in a signal, a pulse for example, on one end and measure the response of the system at the other end. The response of the system tells one a lot about the workings of it. Actually, I've built a little gadget to do exactly that in electronics: an audio-signal injector and a small amplifier to make any response audible again. But I digress.

In this particular case my 'Lenneke' was the stimulus. I've observed the response somewhere else on the web; it told me exactly what I wanted to know and what I feared.

I doubt this makes any sense to you, Oztules, my apologies for that, so I'll direct the rest to the person it's really meant for.

Hi Abigail!

Strange to see you're still reading up on me. I have not much of an idea as to why but obviously you know best yourself. I doubt it is because you are interested in renewable energy though. I can only guess to your motives but no scenario I can come up with is an attractive one. I must admit it feels a bit like being stalked for me. Definitely not a pleasant experience, I'd say.

So, I'd really appreciate it if you behave like a grown-up, get yourself over any feelings you may have (just like I did in your case) and get on with your own life. It would be the sensible thing to do.

Anyway, just to prevent further escalation I've taken some precautionary steps to prevent future interaction between the two of us. It wouldn't be in either your or my interest.

Whatever you do, don't make the mistake of considering me to be either a brother or a friend. And please don't ever make the mistake of contacting me again in any way, shape or form. Should it make you feel any better, I've forgiven you for all the things you've hurled at me or implied about me, so consider yourself to be 'free' in that respect.

Abigail, I wish you a nice life with much of whatever it is that you are looking for.

May the Spirit of Peace be with you!

And thank you for respecting my above expressed wishes.

Peter.

(my apologies to all the other readers of this forum. It has nothing to do with RE but since this is my diary and this particular thing is important to me at the moment, I will put it here anyway. Besides, I have nothing to hide (anymore) in this matter. If there's one thing I've come to utterly dislike over the past two years it's vague, indirect communications. It has the potential for enormous damage, does generally more harm than good, causes a lot of anxiety and is a bit cowardly too, I've come to realize. That includes anonymous postcards. If someone can't say to me what they want to say straight in my face I'd rather they shut up alltogether. 'Let them speak now or be silent for ever'. But I guess I'm a bit funny in that way)

Please abstain from replies.

[dinges]

Last bit of advice:

If you continue to fsck around with people like you do you -will- receive your dues sooner or later. Either in this world by the hand of man or in another way.

Your meat*** boy is the perfect illustration of the issue; I strongly doubt seducing/tempting a man, so she can then have the opportunity to dump him, is the way your mother raised you.

Consider yourself to be sternly warned.

Peter.

'Gentle doctors cause stinking wounds'.

[dinges]

...And we're back in business...

The original PC PSU has been repaired. It turned out (as Oztules said) that one of the switcher transistors (MJE13007) was a goner. Strangely enough, with my DMM on diode-test it did NOT measure shorted. With a semiconductor component tester it did. Interesting information...

After replacing (originally contemplated inserting a MJE13007A replacement which I had, but this had to be isolated; lots of extra work, etc, so tried something else; removing the 2 switcher transistors from another PSU. Of unknown type (type nr. wasn't familiar to me) but at least they were also NPN, had roughly the same gain (Hfe=24) and same pin-out (BCE). Installed these transistors on the original cooling fin and it worked again.

Now, before I give the current modification another try I'd like to know whether I understand the working correctly now. See scanned drawing.

A higher resolution version (80 kB .jpg) can be found here:

http://www.anotherpower.com/gallery/dinges/PC_PSU_schematic_2_resized_IRC?full=1

The way I see it; the higher the output current, the more negative point C becomes. As C becomes more negative it pulls the voltage on point A down; when the voltage on A becomes negative (less than on the + pin 16) the output goes high and current limiting kicks in.

Now, at a certain voltage on point C, it turns out that the LOWER the potmeter is set (lower resistance), the sooner the voltage on A goes down and hence current limit kicks in.

So, in order to have a current setting range of, say, 100 mA to 10A, it is (in my case) necessary to LOWER the potmeter value of 2k2 to, say, 500 ohm or 1k. That way at a given negative voltage on C the voltage on point A becomes negative sooner.

Another option might be to increase the 47k resistor to 220k or 470k. But at least it should NOT be lower (as it was before; I also tried it with 4k7 after trying 47k first and with 4k7 it blew a switcher transistor)

Am I correct so far, Oztules ?

So here's what I plan to do:

1. try a 4" wire; a longer wire should give me more negative voltage on point C for a given load.
   
2. try a lower potmeter (500 ohm or 1k); that way at a certain negative voltage at C point A becomes more negative; that should give me a lower current range too.
   
3. try a 220k or 470k resistor instead of the 47k one. This should have the same effect as lowering the potmeter.
   
   

I plan to carry out these steps in this order; perhaps switch step 2 and 3.

As a load I'll use 2 H4 car bulbs this time instead of the single one I used so far; that should give me a load of 10A, as opposed to the previous 5A.

If the PSU works properly then I'll add another small 12V power supply for the fan and the TL494; that way the TL494 would be powered independly of its own system which should improve functioning (reduce instability) at certain settings.

Oztules, sorry if this all was a bit longwinded but I'd be interested in your opinion before I smoke more silicon.

Regards,

Peter.

[dinges]

Update:

With a 4" long sense line I get -30mV voltage drop at 10 A.

Tried a potmeter of both 100 ohm and one of 25 kohm; the 25 kohm one seemed to give the most adjustment range, but I could only regulate in the upper part: 9-10 A. It was definitely the current control that was working as I could turn the voltage potmeter way up without voltage rising; then, when increasing current control, voltage would (rapidly) rise. The same happened when disconnecting the potmeter (thus disabling current control): voltage rapidly rose to its volt-potmeter setting.

So, -30 mV at 10A; that means about -3 mV at 1 A.

Guess I'll have to figure out such a combination of resistors (47k resistor and potmeter) that the current setting is adjustable from ~1A to ~10A.

Peter.

[oztules]

Good to see you back at it

 "  1.  try a 4" wire; a longer wire should give me more negative voltage on point C for a given load.

   2. try a lower potmeter (500 ohm or 1k); that way at a certain negative voltage at C point A becomes more negative; that should give me a lower current range too.

   3. try a 220k or 470k resistor instead of the 47k one. This should have the same effect as lowering the potmeter. "

to point 1..............yes true.

to point 2..............Yes, on reflection, keep the pot at a fairly low value ... less than 1k. On thinking about this, the pot value will dynamically change the relationship of the 47k voltage from the vref. ie as we go to a larger pot, you decrease the impedance load on the 47k, (and + voltage presented to pin 15- will increase) and so may get into runaway when you adjust for A max in experimental mode (twiddling). If your load cannot accept A max, you may be lulled into a false sense of security, and a new load may exceed your upper pot limit. When you establish it for sure, you can increase to further the range upward.

to point 3........ yes and see point 2 and the ramifications. this should work to lower the A out.

It will be patently obvious by now that an EE I aint. I just try to understand the workings, and develop a work around. There are lots of folks on this board who would be better able to give you a far more decent current control circuit, but this one works very well for me and has proved to be very very robust. I use the yellow wire from the power supplies red black yellow output strings for my current sense wire. Maybe you used heavier stuff, and this is what is giving you limited readings for your lower currents.

Remember, as you twiddle the pot settings to lower feedback from the tap, you are also changing the voltage drop over the 47k resistor (by resistance change from the pot), and so the vref + voltage given to pin 15 will be changing it's positive bias as well as the feedback quotient from the neg tap.... makes calculating what is happening quite dynamic..

..........oztules

[dinges]

Hey Oztules,

Just got back from the shack and read your reply. Good news; it now works mostly.

Current is now adjustable over a range from 2.5-12 A; voltage over 2.5-22.7V.

I want to increase the current range from 1-12 A and that'll work; just that it's getting too late now.

Your remark on the dynamics of it is correct and it is what caused me to blow the circuit in the first place; today I noticed that, as you turn the potmeter from extreme left to extreme right, that current goes from max (12A) to minimum (2.5A) (over about 90% of the wiper range) and then, over the last ~10% of rotation, goes back up to 12A!

I have solved that by adding a small resistor in series with the current potmeter; the current potmeter is now 25k; in series is a 1k5 resistor. The '47k' resistor is now 1M5.

By increasing 1M5 to, say, 4M7, and increasing the potmeter to ~50k, I should get the (lower) current range I desire, from ~1A to 12A. Have been playing with various resistances in the circuit and am finally getting the feel for how it reacts.

Also tried short-circuiting it (with current limit set) and it survives the shorts; current stays exactly put to the value I set it too.

At 12A the cooling fin of the diodes gets slightly warm (~40 deg. C). This is without a cooling fan blowing on it. Excellent.

Must say I'm very impressed with the performance. This won't be the last PSU I've converted. Way cool!

The test load I've used was a 12V/10A (120W) car heater.

Thanks mate. Who knows, maybe them Aussies can be trusted after all...

Peter.

[oztules]

Hmmm Now I'll have to find something else to piss you off about.... can't have a happy Peter.

Well done, you have explored it's character and tamed it appropriately. I never bothered to make it more refined than I had use for. It does seem to make a remarkably useful test supply for projects you don't wish to vaporise on first test.

You will never look at PSU supplies with the same distain that you once had (storing 20 of em in the darkness of the attic).

"today I noticed that, as you turn the potmeter from extreme left to extreme right, that current goes from max (12A) to minimum (2.5A) (over about 90% of the wiper range) and then, over the last ~10% of rotation, goes back up to 12A!"

...... yes I suspect that someone who actually knows what is going on will be more useful at explaining this behaviour as opposed to a bushranger like me. I haven't experienced this behaviour, but my crummy trim pots seem to have no start/finish  point that I can feel, and just go open circuit over the last/first part as the wiper jumps over the start/finish of the carbon tracks. This may explain your 12A upper limit... same as adjusting for your upper limit. If the comparator were to truly loose the feedback, pin15- would go vref and it would go to max,and blow up ( if the load is high enough) rather than the trim pot max value.... seems an odd co-incidence.

If that really happens and pot is 0R, I can't explain it..... which you can add to the myriad of other things I can't explain.

Note, out of the 20,000,000 people down here, they can't all be as dispicable and disgusting as me.

just been told to go and cut some wood for the fire..... later

............oztules

[dinges]

So, I came across another PC the other day. The computer was a worthless 486SX, but its power supply certainly was yummy...

Started converting it an hour ago. Wanted it to go up to 30 V but alas, it tops out at 24.5 V unloaded; loaded with 3A it tops out at 23.3 V. Still pretty good results.

The voltage conversion was amazingly simple this time. Also, there seems to be no overvoltage protection circuitry. Or at least, it doesn't kick in. Makes one wonder... Anyhow, it makes my job easier.

Next task: implement the current control modification.

The 2nd opamp is unused, so that's good too.

Darn, Oztules. Now look at what you made me do... I doubt I can ever come across one of those PSUs again without immediately feeling the urge to convert it. Shame on you!

Peter.

[oztules]

Yes Peter, it's a sickness..... I think I may have mentioned that you wouldn't look at psu's the same way again.

The next step is setting up the control loop for mppt, and using it as a 200w high impedance load for hv windmills.... and then paralleling them for kilowatt then multi kilowatt... and then multi multi kilowatt and then ... and then...

Think I need to take my tablets now.

..........oztules