Considering building a brute force, high current, adjustable linear bench supply

[AmmoJammo]

I have four 250va, 18volt (centre tapped) transformers that I bought some time ago....

Now I'm considering putting them to use...

The idea is to rectify each transformer separately, then connect them all in series, of course using the centre as ground, to give a total voltage of +-50volts or so, as well as using the +-25 (or so) "taps".

Then, of course, having +- adjustable regulators on the +-50volts, as well as separate +- regulators on the +-25volts...

Giving me the ability to get (for example) +-15volts, and +-45 volts for running/testing a preamp, and power amp at the same time...

Should I do this differently? Are there any (major) issues with this? apart from the potentially massive heat dissipation required 

Thanks!

(diagram just for reference)

[Paul Price]

Your circuit diagram is inconsistent with the single wire center tap connection of each of the two transformers, and your connection of 3-wire unknown type designated regulators is also inconsistent in your diagram.

Most 3-wire regulators I have come across have a max current output <5 amps, so  getting the brute out of the electromotive force would not be possible with your simple circuit.

Filter capacitors are needed to help rectify some problems with your circuit.

[AmmoJammo]

Your circuit diagram is inconsistent with the single wire center tap connection of each of the two transformers, and your connection of 3-wire unknown type designated regulators is also inconsistent in your diagram.

Its four transformers

Each squiggle in the drawing is two windings connected in series.

Which brings me to the question.... what happens if I connect the secondary windings "out of phase" with each other, which I believe is possible.... (same question for the primarys actually, they're a 120+120 volt input, with no markings on the leads)

[Paul Price]

If the leads of the series-connected transformers are improperly reversed there will be simply no output since the voltages would oppose each other.

[AmmoJammo]

Thats what I assumed thanks!

[AmmoJammo]

I assume that if I'm getting 19volts ac that the primaries must be wired correctly? lol and I can cut, solder, and heatshrink the fly leads, and not realise half an hour later that its wrong 

[AmmoJammo]

I found a rack mount case, and some capacitors to go along with it too... These bits were originally purchased for a multi channel amplifier.

[Eviltech]

Nice bunch of transformer you've got there! I dot really see how they will fit in that box tough... Anyway about your circuit - find the LM317 full datasheet with the transistor enhansments cirqits and that i believe will be enough to provide your brute force linear regulators. Oh, you'll need the negative version of the LM317 too . By the way i'm not shure what voltage they will witstand so you might want to use 2 of them in series.
Boy, that over compicates your design.... Oh and you can use stereo potentiometers for adjustment so your voltages are allways parralel. If you are panning to use a fan i suggest you use a separete rectifier from one of the traf's or use entirely separate power supply for it. That is because the brushess fans tend to insert a nasty interfierence on their supply line.

[AmmoJammo]

Nice bunch of transformer you've got there! I dot really see how they will fit in that box tough...

Magic! going to use the aluminium plate to bolt the transformers down, mount the bridge rectifiers to the bottom of the case, and run the rectified DC through/past the metal plate, to the capacitor banks

Plan is to find some suitable angle aluminium to make a heatsink that covers the back of the case... although, its not as high as I had originally wanted...

[AmmoJammo]

Does this work instead of the original design?

[ElectroIrradiator]

Is this supply something you intend to be reliable?

If so, then you need to rethink how you treat those transformers. First, you cannot use 'hard¨spacers to make a squeeze fit. You need the individual metal disc mounts, complete with the two rubber disc spacers. Mounting them the way you propose, will effectively squeeze the secondary windings down into the primary near the core, potentially creating an early failure.

Secondly, these things will get hot. At full oomph they will in total need to dissipate up around 80-100W of heat, and that is assuming your PSU design doesn't overload them (which many simple designs do).

[AmmoJammo]

No, I plan for it to be completely unreliable :p

I was going to use 3mm thick rubber between the base of the case/metal plate, and the transformers.

These transformers didn't come with any mounting hardware.

[ElectroIrradiator]

No, I plan for it to be completely unreliable :p

I was going to use 3mm thick rubber between the base of the case/metal plate, and the transformers.

OK, the CD had me worried...

How do you intend to cool the transformers, and how much current do you expect to be able to draw from each rail?

[AmmoJammo]

These transformers are only rated at 250va, but the cores are more like 500va transformers.

I don't think they'll have any major issues with heat, but I don't intend this supply for 100% duty cycle anyway

As for power... well, whatever it needs to do... Depending on the output voltage, the heatsink is likely to become the main issue.

I'm wanting to do regulators that will handle about 15-20amps, even though the transformers aren't rated for it.

[Jon Chandler]

I'm wanting to do regulators that will handle about 15-20amps, even though the transformers aren't rated for it.

You might start by looking for a data sheet for a 20 amp regulator...and share the result with us when you find one!

Keep in mind that for linear regulators, the power dissipated is

(voltage drop across the regulator) x (output current)

so if your drop is 5v @ 20 amps, you have 100 watts of heat that needs to go someplace.

[SeanB]

Just checking that you do not intend to place a metal mounting screw through the inner holes of any of the transformers. If so it needs to be well insulated from the metal plates at both ends.

[AmmoJammo]

You might start by looking for a data sheet for a 20 amp regulator...and share the result with us when you find one!

Keep in mind that for linear regulators, the power dissipated is

(voltage drop across the regulator) x (output current)

so if your drop is 5v @ 20 amps, you have 100 watts of heat that needs to go someplace.

Anything really that will handle the input voltage, with some external transistors for the extra current

I was actually working on the basis of 600watts of heat total, based on a worse case scenario of 20volts drop, at 15amps (per rail) Remember, its not intended for 100% duty cycle.... yes, it MAY have to dissipate 600watts occasionally, but at least for smaller loads I'll have a heap of headroom!

Just checking that you do not intend to place a metal mounting screw through the inner holes of any of the transformers. If so it needs to be well insulated from the metal plates at both ends.

I assume that's because it will act as half a turn? so could potentially have 0.5 - 1volt on it? I'd have to think that isolation from one end is all that would be required? not both? Thanks for the heads up

[SeanB]

I assume that's because it will act as half a turn? so could potentially have 0.5 - 1volt on it? I'd have to think that isolation from one end is all that would be required? not both? Thanks for the heads up

It will act as a shorted turn if there is any other path to connect the 2 plates. Insulating both sides means there is less chance of a single insulation fault ( like placing the bottom on a metal table) causing a shorted turn, which will burn the transformer out or heat the bolt or the plate to red heat.

[TerraHertz]

Just checking that you do not intend to place a metal mounting screw through the inner holes of any of the transformers. If so it needs to be well insulated from the metal plates at both ends.

Repeating this, since you didn't acknowledge.
If you create a 'shorted turn' conduction path via the case metalwork, in a loop around the core, you'll get a big surprise.
A lot simpler to just avoid that, fun though it might be to see the case glow red hot.

If you want to put a bolt through the middle of the cores, either insulate at least one end very well, or use a nylon threaded rod.

Btw: http://www.ko4bb.com/Manuals/HP_Agilent/HP_6269B_Technical_Manual.pdf
Details of some grunty HP supplies, such as 0-40V @50A. They are mixed mains triac phase control of a transformer input, then linear regulation of the DC output. The triac switching is arranged to maintain a fairly small voltage across the final linear regulator.
The manual scan quality is crap, but usable.

[Eviltech]

Just checking that you do not intend to place a metal mounting screw through the inner holes of any of the transformers. If so it needs to be well insulated from the metal plates at both ends.

Repeating this, since you didn't acknowledge.
If you create a 'shorted turn' conduction path via the case metalwork, in a loop around the core, you'll get a big surprise.
A lot simpler to just avoid that, fun though it might be to see the case glow red hot.

If you want to put a bolt through the middle of the cores, either insulate at least one end very well, or use a nylon threaded rod.

Btw: http://www.ko4bb.com/Manuals/HP_Agilent/HP_6269B_Technical_Manual.pdf
Details of some grunty HP supplies, such as 0-40V @50A. They are mixed mains triac phase control of a transformer input, then linear regulation of the DC output. The triac switching is arranged to maintain a fairly small voltage across the final linear regulator.
The manual scan quality is crap, but usable.

I want to kiss you right now!!! I had this HP supply for 6 years and havent used it because it was set up wrong! Thanks to you i am finaly able to use it!!!

[dannyf]

but I don't intend this supply for 100% duty cycle anyway

You may want to go back to some basics: understand how the supplies are to work and what parameters you need to design to.

[AmmoJammo]

Repeating this, since you didn't acknowledge.

yes i did

[TerraHertz]

Repeating this, since you didn't acknowledge.

yes i did

My apology. Though I'm pretty sure you hadn't when I started writing that. Then went on hunt for the HP power supply manual, got sidetracked, etc. Can't really tell the time sequence since I think the post timestamps are local times?

I had this HP supply for 6 years and havent used it because it was set up wrong! Thanks to you i am finaly able to use it!!!

Cool. If there's anything unclear in that horrible scan I found just ask. I have original physical copies of the HP manuals for that range of supplies.  Btw, you know it's not hard to find original manuals for such things?
It's annoying when people go to the trouble to scan things, but do it so badly the result is nearly unusable. Huge waste of everyone's time. To save what, a few MB?

[GamerAndds]

Yea I got some major amps when I put a metal plate at the top of my transformer to the case, dont want to do that. I got something like 10 AMPS and I think 1.12 volts  . Lets say 3 14 gauge wires were literally smoking and solder was turning molten again.....  . I am using a hammond transformer 182 series T60, 1,500 VA centertapped 60V outputs (120 total) at 25 amps!!

[AmmoJammo]

My apology. Though I'm pretty sure you hadn't when I started writing that. Then went on hunt for the HP power supply manual, got sidetracked, etc. Can't really tell the time sequence since I think the post timestamps are local times?

No I'm quite sure you replied 15 hours after I posted it....

I ran a piece of 12ga wire through the centre of the transformer, measured 0.5volts or so, and a short circuit current of about 30amps.

Yea I got some major amps when I put a metal plate at the top of my transformer to the case, dont want to do that. I got something like 10 AMPS and I think 1.12 volts  . Lets say 3 14 gauge wires were literally smoking and solder was turning molten again.....  . I am using a hammond transformer 182 series T60, 1,500 VA centertapped 60V outputs (120 total) at 25 amps!!

I don't understand what you're saying? or what you did.. Its the bolt, or wire, through the middle of the transformer that creates a voltage across it, 10amps across three 14ga wires (not sure where these wires even were?) shouldn't have gotten them hot...

[GamerAndds]

I have an aluminum chassis and I put an aluminum plate ontop of the transformer, when I tried to ground that plate to the chassis through 3 14 guage woes they started to smoke. I don't know if it applies to you because ur using smaller ones, but for me that was a real scare!!! I'm pretty sure it's the plate, since that happened I switched out the metal plate for a plexiglass one and everything was fine, perhaps it was the bolt causing the major amps. The 3 14 guage wires were grounding the plate to the chassis and that's why they were hot, it was the weak link I'm the chain, they were no where bear the transformer

[johansen]

I have an aluminum chassis and I put an aluminum plate ontop of the transformer, when I tried to ground that plate to the chassis through 3 14 guage woes they started to smoke. I don't know if it applies to you because ur using smaller ones, but for me that was a real scare!!! I'm pretty sure it's the plate, since that happened I switched out the metal plate for a plexiglass one and everything was fine, perhaps it was the bolt causing the major amps. The 3 14 guage wires were grounding the plate to the chassis and that's why they were hot, it was the weak link I'm the chain, they were no where bear the transformer

looking at your signature you have a rather large transformer, probably on the order of 1 volt per turn.
so when you short out the "aluminum plate" you are forming a shorted turn and you've got 1 volt dropped across whatever.
so of course it glows red.

OP: i have the pass transistors from two 0-36v 0-30 amp keptco power supplies if you want them. shipping from seattle area, they probably weigh 2 kilograms for a set, i can throw in the fans if you want, they will be sufficient for a 0-50 volt 0-25 amp supply.

[AmmoJammo]

Thanks for the offer, I've pretty much figured out what I'm going to do (I think).

[sprocket]

Hmm..

Have you considered paralleling LDO's? Or perhaps incorporte som power transistors to take some of the load of the LDO's. You could for an exsample use a 2N4399, I think it can handle 30amps or so.

You could potentially use LT1083/LT1084 or LT1085.. I have not reallly read the spreadsheets carefully. But they can handle some thing like 30 volts input, dish out about 7amps per LDO.

Best of all you can request free samples from liniar technology and have a few of them to play around with for free. I think it take about a week or so for you to get the samples. I have some in transist now I want to play around with for a similar project like yours.

Though my project is going to be more modest and less lethal. 15volt 30amps CC and CV ajusted by an arduino and a OLED display of some sorts.

Oh and you might seriously think about some sort of short circuit protection. I'm always super careful when it comes to high amp equibment. I have been zapped more times by high voltage circuits then I care to admit, but as unpleasant as that is, it's always been a circuit with barely any amps in it. 

[AmmoJammo]

Plan is four separate, "floating" adjustable supplies, one per transformer.

LM317 with five external MJE3055 transistors each.

Four separate switches too, one per transformer

By having four supplies, I can connect them all in series, or parallel, or parallel series, or tie the positive output to ground for negative rails.



Transformers "mounted" it all has to come apart again, as I'm not sure in what order I need to assemble it...

[calexanian]

Does this work instead of the original design?

Yes. That is the preferred method.

[AmmoJammo]

Plan is four separate, "floating" adjustable supplies, one per transformer.

LM317 with five external MJE3055 transistors each.

Plans changed again!

LM723 regulator! same external transistors, but this regulator can easily do adjustable current limiting...

[AmmoJammo]

Switched! and, of course, I snapped a drill bit, and smacked the drill into the front panel.... ah well! not overly worried about it.

[AmmoJammo]

Any potential issues with this layout?

Bottom row of binding posts will be grounded.

[AmmoJammo]

Well, that's the layout I went with!

Seeing as nothing has arrived yet (binding posts, voltmeters, transistors, resistors, shunt resistors, knobs, pots, etc?) I'm using ebay specs for the cutouts... and hoping they're right! 



The bottom row of binding posts line up with the lip on the aluminium plate, so they will be grounded straight to that... no stuffing around with terminals and links
(hence the drilling holes with the case half assembled, so they line up properly)

Oh, and I bought a crusty, cheap(ish) second hand drill press.

[AmmoJammo]

Two runs of 7.5amp cable from the back of the socket... one side of the socket says 10amp, the other says 15amp 







Really can't do anything else with it till all the other bits arrive!
I've got a heatsink being bent up at the moment, hopefully it works out.

All up, its going to weigh about 20kgs...

[AmmoJammo]

[AmmoJammo]

I've decided to go with mains powered fans, operated by thermal switches mounted to the heatsink.

Of course, this means I need to change the mains wiring again... planning ahead? no, not me!

Decided to use three 120mm fans, centre will come on at a low 40 degree Celsius, and the other two will be 60 degrees... No point in running things hotter than they need to be!