Sounds like the regulator isn't working properly, or you have a short somewhere. The LT1085CT should easily provide 3A at 3.3V, and the 486 isn't supposed to use more than 10W. Either there is a short somewhere demanding a much higher current that the regulator can't supply, or the regulator itself is malfunctioning. Where did you get it from, and are you sure it isn't fake?

Are you sure no other components are missing besides the regulator? Check the board for unpopulated components such as capacitors and resistors.

Also, you mention that you bought the fixed 3.3V version of the regulator, but maybe the board was designed to use the adjustable version (without the "-3.3"). I checked a couple of my boards and they use the adjustable version.

My guess is you got a used one and not a new one. I'm thinking about adding one to my Packard Bell Pack-Mate 28 Plus, but, I need to look and see if I need more components, such as capacitors in that vicinity.

When I restored 3.3V support on one of my 486 mobos I had to remove some hardwired jumpers, change some components on the board and add a bunch more missing ones (mostly capacitors).
http://www.tmeeco.eu/BitShit/PCschit/AbitAN4R2.jpg

Well it looks new to me :

and here's how it was before

The only thing I could think of that might have broken it (even though I kinda doubt it) is that the middle pin was harder to solder because there's much more copper to heat up in that area, but I already installed the heatsink with some solder paste, so the heat was already being dissipated, and I didn't leave the solder tip for more than 10-15s.

There's not much more missing components except "Q4" and "Q5" that are useless if "Q6" (the present regulator) is installed, and some CMS nearby the VLB slots and in the socket that seem to have nothing to do with the CPU voltage (I checked for continuity with my multimeter which gave me hundreds of ohms with anything related to the regulator, and voltages once the mobo was being powered on and that gave me 5V and 12V)

My only bets so far are either my regulator is fake, broken, or is the wrong one

(let me some more time to retrieve older pictures of the board, before and during the other mods, etc)

Edit : well I don't see any hardwired jumpers. There was some before, but these were used to control the cache as far as I know

Can you check with the multimeter the trace that connects to pin 1 of the regulator (the one close to the edge of the board)? In case of the fixed version (the one you have), it should connect to ground. If it goes to some resistor it means you probably need the adjustable version of the component as that pin is ADJ instead of GND in that case.

I get 5V on that trace. I've read the datasheet as well, it's kinda weird indeed, I'm not sure if the documentation is flipped or if there's something else wrong there (Wrong regulator recommended on elhvb ?)

Edit : well, looking at that blurry picture Bought these (retro) hardware today, the regulator seems to be in the other orientation ... That's odd, but that would somewhat explain why ?

Though this is another revision of the board so I have no idea ...

Edit 2: After checking for continuity, the 1st pin seems to be connected to +5V indeed, but the 3rd one don't seem to be connected to the main GND.

I might try flipping that regulator, hoping it's the right thing to do and if so, that it didn't break ! But if so, why did the manufacturer did put "1" on the pin the nearest to the edge if that's the other way around :'(

Yes, the pin locations seem to be flipped, very weird indeed.

I can see that the pin labeled "3" connects to some resistors, which would make sense if it corresponds to the ADJ (pin 1) in the non-fixed version of the regulator. I don't know if a fixed regulator would work in that case, you may probably get a wrong output voltage.

Well I'll flip it and check which voltage I'm getting in the socket. If it seems allright, I'll try a CPU there

(And if it doesn't, what if I unplug the pin connected to ADJ and get some ground from somewhere else 😁. Probably a bad idea ...)

Edit : flip done, the regulator now sends 4.1V to the CPU core. I've tested with one CPU and it works indeed !

... but yeah that's still too much x)

Actually grounding the "adj" pin might be the (temporary) solution as the voltage on the middle pin is exactly the same as the one in the socket so I guess there's not much done to the output voltage afterall, what do you guys think ?

At least part of the problem is fixed.

To anyone having that board that wants to add 3.3V support : DON'T FOLLOW THE MARKING ON THE BOARD, IT'S PROBABLY WRONG ! Check the voltage on the pins and see which position match the datasheet instead !

So I did the GND mod ... and it looks fine so far. Now the CPU gets 3.3V as it should. But I'll buy the proper regulator anyways : that mod is ugly and should never be done x)

A temporary solution is better than no solution. I don't see any adjustable regulators online (fixed only) sadly.

bjwil1991 wrote:

A temporary solution is better than no solution. I don't see any adjustable regulators online (fixed only) sadly.

LM350 http://www.ti.com/lit/ds/symlink/lm350-n-mil.pdf adjustable 3A 1.2-33V regulator pin-compatible with the LT1085 http://www.analog.com/media/en/technical-docu … ts/108345fh.pdf

For even higher current capability you can use LM338 http://www.ti.com/lit/ds/symlink/lm338.pdf goes to 5A.

Both can be had cheaply on ebay etc.

Huh, too late, I've bought this one instead https://www.ebay.fr/itm/LT1085CT-IC-ADJUSTABL … XEAAOSwA3dYUlk5 ...

But that's still a neat information ^^

stamasd wrote:

LM350 http://www.ti.com/lit/ds/symlink/lm350-n-mil.pdf adjustable 3A 1.2-33V regulator pin-compatible with the LT1085 http://www […]
Show full quote

bjwil1991 wrote:

A temporary solution is better than no solution. I don't see any adjustable regulators online (fixed only) sadly.

LM350 http://www.ti.com/lit/ds/symlink/lm350-n-mil.pdf adjustable 3A 1.2-33V regulator pin-compatible with the LT1085 http://www.analog.com/media/en/technical-docu … ts/108345fh.pdf

For even higher current capability you can use LM338 http://www.ti.com/lit/ds/symlink/lm338.pdf goes to 5A.

Both can be had cheaply on ebay etc.

Thanks for the info. I have a few blank spots on my Packard Bell (2 resistors, 2 SMD caps, and 1 other thing that goes behind the CPU socket) that might be for the voltage regulator (will look at the traces and check for continuity to see which ones go where exactly).

Wow that's expensive. The ones above can be had for under $1/pc and they're actually from a newer generation of regulators.

Well it's paid, so there's no turning back x)

I'll give some news when I'll receive that.

Now even though it's not really needed, as promised, here are shoots of the board during the mods ^^

Pimpin' a cheap yet promizing 486 mobo :

Here's how it was before any operation.

As you can see the cache is soldered ...

... as well as the ROM chip

To me this screamed for fake cache (and obviously even though the bios claimed there was 256KB of cache, no benchmark could find it), and indeed, here's what's inside these "chips"

How many bytes do you think a metal part can store ? 🤣

Now there's also this light battery leak to neutralize

After quite some time using solder flux and tin here's how it looks

(the large copper trace still needed to be treated when I shot this, but now it's done, I don't have any picture of that yet. I just scrubbed the solder mask and tined the copper underneath)

Meanwhile ...

Oops, I was too quick at removing that one "chip" and the trace came with the chip ... Fortunately the via was mostly fine and the trace was still long enough to be soldered back where it was.
With a socket above it nobody will ever spot that 😀

Photos of the finished product will come later

Most components were salvaged from a PCChips m912 (including the regulator heatsink you saw earlier).

Why ? Because first of all, despite looking ok, the PCChips mobo it totally dead, and secondly, this :

The PCB of that "new" mobo is much thicker than the PCChips' and the overall quality seems much better (even though that pcchips had a regulator and real cache ... which is now on the "new" board 🤣)

So here's the current state of the board :

Not too bad, except for that ugly tweak 🤣

Nicely done. You inspired me.

Here's a close up of my shuttle Hot 419 without 3.3v:

Here's a closeup of a board with 3.3v:

I see a transistor, 2 added capacitors, 2 upgraded capacitors, 2 regulators, a bunch of jumpers, 8 resistors and a ferrite.

There's another pair of upgraded capacitors on the other side of the socket.

Seems daunting.

I can take a pretty good guess at the resistors from the stripes, but I'm not sure what to do with the capacitors or transistor without finding someone that has a board with the parts.

Thanks ! I'm glad I could inspire you ^^

This photo is really low res/old but I think that on this board specifically there's only one regulator https://www.elhvb.com/supportbios.info/archiv … -419htm/419.htm

Edit : another one here https://www.elhvb.com/webhq/models/486vlb3/hot419p.jpg