Forum Home
PC World Chat
Thread ID: 56063	2005-03-26 06:19:00	13.8v regulated power supply from AT PSU?	george12 (7)	PC World Chat

Post ID	Timestamp	Content	User
338202	2005-03-26 23:38:00	I've got a non-switchmode one that will do I suppose. It's rated 5A, so 7A is pushing it, but it should be OK. Not unless you provide some forced and very effective cooling. The rectifier chip will soon go into meltdown at that level of continuous overload. Cheers Billy 8-{)	Billy T (70)
338203	2005-03-27 01:32:00	A 7805 + 3 diodes will give 6.8 volts Try a 7812.... And remember the maximum current for a 78xx is 1.5 amps, so you need a series pass 2N3055 or two. Yeah sorry, I thought I had written 7812. And that's the transistor I mentioned, one should be enough as I usually use 8A rated ones which makes it pretty much perfect. Not unless you provide some forced and very effective cooling. The rectifier chip will soon go into meltdown at that level of continuous overload. It's a bridge rectifier thermal-pasted and screwed to the metal frame. But I wish there was a better way to up the voltage of a switchmode, because this box is big! :(. If I could use the switchmode, the board would easily fit inside the case, and it's really light. If only the PC power supply was linearly regulated - I could just stick some diodes in. George	george12 (7)
338204	2005-03-27 04:57:00	It's a bridge rectifier thermal-pasted and screwed to the metal frame. But I wish there was a better way to up the voltage of a switchmode, because this box is big! :(.George Yeah, well, I actually meant the rectifier and the linear regulator, but not enough sleep for several days turns my brain to mush. Back to the SMPS, all regulated PSUs have a linear voltage regulator at their heart, be it a zener diode, integrated regulator chip or a regulator built into the main IC. If you can find it you can adjust it, change it or jack it up with the three-diode trick. Upping 12 volts to 13.8 is not a big shift so I'm not sure why anybody would want to go into a full redesign unless the PSU was to be used in a mission-critical development. Similarly, the requirement for a full 13.8 volts may not be that critical and 13 volts may well do the trick. There are many ways to skin a cat (or so the saying goes) some messier than others. Cheers Billy 8-{)	Billy T (70)
338205	2005-03-27 05:13:00	I think the regulation in a PSU applies to all the rails, so taking the 12v to 13.8 would mean that the 5v would be close to 6v as an example. If you were not using the PSU for anything else, this may suffice of course. Then there is the usual PSU need for both the 5v and 12v rails to have a minimum load of around 2 amps each, for many supplies to actually fire up.	godfather (25)
338206	2005-03-27 05:44:00	Well I don't care about the 5v rail, it can be 300v for all I care. Or not even exist. I don't think it's going to be possible so I have another plan, which is kind of more complex but will work. I'll have the usual 12v (11v under load) output of this PSU I'm trying to convert run the PC usually, and my old 5 amp battery charger (well, the transformer is actually out of an inverter) hooked to a 13.8v linear regulator which I already have made and working. Normally the 12v from the switching PSU will run the PC, when there's a power cut it will (by relay) switch to the battery. It's more complicated but it will have to do. --------- Fake edit: New plan: Actually I'm gonna just put the 5v rail of one PSU in series with the 12v of another and regualate it to 13.8v. May need a little modification to run in series without shorting to common ground, but I can handle that. I'll give this plan a go now :)	george12 (7)
338207	2005-03-27 12:49:00	George If you really want to get your 13 . 8 volts you could try this: The PSU will have separate windings for 3 . 3, 5, and 12 volts . If you are able to identify these windings on the transformer, placing the 12 volt and 5 volt windings in phase and in series will give an output at 17 volts once rectified . If you then series the 3 . 3 volt winding in anti-phase, you will get a winding capable of providing 13 . 7 volts . This technique is called voltage bucking and used to appear quite often in old electronic equipment . The same antiphase injection of ripple voltage was used in early radios to reduce or "buck" the normal hum levels created by the use of an electromagnet for the speaker . Back to reality, the current capabilities of the ATX 3 . 3, 5 and 12 volt windings are quite compatible so you could get the voltage and current out that you require . Success will depend on where the reference output reference for the voltage regulation comes from but if you are lucky it will be from one of the lower current rails for stability reasons . You would be stacking ac windings only, not DC output voltages and would have to ensure that the net effect was one single winding with a ground at one end and an AC feed to the 12 volt rectifier at the other . Work with care, computer power supplies can bite . Cheers Billy 8-{) Another tip, you won't be able to meaure the AC voltages very easily because they will not be a sinewave . If you have a true-rms meter you might get a close figure but I'd rely on tracing back from the output to identify the windings .	Billy T (70)
338208	2005-03-27 14:41:00	Cool, thanks. Might as well try it! I'll just go with the 17v and then linearly regulate down to 13.8, because the 12v winding is unregulated. But I didn't think of that, thanks very much.	george12 (7)
338209	2005-03-27 21:07:00	Cool, thanks . Might as well try it! I'll just go with the 17v and then linearly regulate down to 13 . 8, because the 12v winding is unregulated . All the windings are regulated George, they only need to monitor one because the secondary output level is controlled from the primary winding, thus any correction to one secondary output corrects them all . That is why I said that you needed to identify which output provides the feedback to the primary . They often work through an opto-coupler to maintain isolation between primary and secondary though older technology used a separate reference winding . You can't disable the feedback or the power supply will self destruct, but if you go for a linear regulator (I wouldn't, it is too inefficient) if it will work on a 1 . 5 volt differential I'd use 12V + 3 . 3V to minimise the dissipation in the regulator . If you are not sure that you have kept the feedback loop intact, to avoid a chernobyl at switch on, borrow a variac (variable output autotransformer) and run the supply up on that . If the output on one of the untouched rails keeps rising past its nominal output voltage then you can stop right there without damage . As an alternative, wire a 60 watt light bulb in series with the phase supply . If everything is OK it will not light up very bright, but if it comes on full you have a problem . To get full output you might have to go up to 100-150watts but start small . I have lost count of how many switchmode power supplies I have repaired, but it would be well in excess of 2000 and I never lost a unit while using variacs and lightbulbs on the input . Cheers Billy 8-{)	Billy T (70)
338210	2005-03-28 01:51:00	Cool, will do. Good old current limiting :). The feedback appears to be a small audio-type transformer. I need to experiment with loading up the 5v rail and see what happens to the 12v's regulation. Billy, it doesn't have a 3.3v rail - it's AT. My theory was it'd be simpler due to not having +3.3, +VSB, and PS_ON cluttering things up. Here's an idea, probably wouldn't work: If I load the 5v rail to the brim (well say 15a) then will it increase the pulse-width to compensate, upping the 12v rail? Better still (saving me 75 watts of power) I just have to make it think the 5v rail is really low. I'll have a bit of an experiment with the lightbulb in series. Cheers	george12 (7)
338211	2005-03-28 01:57:00	If you can identify the 12v winding, just add a few more turns to get the desired voltage.	godfather (25)
1 2 3