I see... well I'm not really that interested in 5x86/POD at this point, if I am to upgrade this further I might as well make a p1 build 😁
I think the 100mhz 486 is really fine, plays DOOM stuff decently and with turbo=OFF you get 386DX performance

If you have luck, the board will do 486 DX4 120mhz if an new bios is installed.. Most boards do end up here.. Mostly because of lack from the manufacturer(in mind you could better buy a new board)

keropi wrote:

http://i39.tinypic.com/2usyl9j.jpg […]
Show full quote

I'd attach a heatsink to the back of that LT1085 before it fries.

^ that's just for testing, I have a small fan cooling it atm... the plan is to bend it properly and attach it to it's place on the mobo's ground plane to act as a heatsink, if you see the pics on the previous page you'll see the empty space behind it 😀

Don't bolt the regulator to the motherboard without making sure you can do it!

The regulator tab is same metal as the middle pin, don't connect it to ground.
Make sure if the metal plane is connected to middle pin, or completely disconnected. Then you can bolt it directly.
If the metal plane is ground or something else, you need some isolation material between regulator and PCB, and you also need to isolate the screw.

^ ah, thanks for that , I'll check what the space behind it is and proceed accordingly 😀

I tried it also to stick the LT1085 in that plastic 3 pins thing, but i dont have JP6 headers soldered in for now.. Its just hardwired on the right side on mine board.. And removed the 3 bridge jumpers for 3.3 volt setting.. Actually it should do now 4 volt with that setting.. Also mine gave me the same read-outs>

Measured on the regulator:
V-in = 4.98 volt.
V-out = 3.36 volt. same is the outer tab.. (very stable voltage)
Adj = 2.10 volts

JP6 = 3.36 volts (hardwired)

JP8 = 4.98 volts (hardwired)

On the three bridge jumpers:

The upperside measured 3.36volt same as the output voltage.
The lower measured 4.98 volts.

Should it make any difference when i going try out some other regulators, are shoud i get the same measurements?

I'd leave it standing, with a DIY tailored vertical heatsink attached. Any flat piece of copper or aluminium cut and bent to shape will do, something of this sort:

This would unload heat from the mobo. Decent case ventilation would help also.

What could be wrong why we have the not correct output voltages?? Should we maybe need a 4V one or so?

So the input would still 4,98 volt.. But in the first phase it need to go 4v, and the second phase its need to do 3.3 volt.

@TELVM

hm, interesting but kind of dangerous because of the position the LT1085 is... it's between the vesa slots so it's easy to touch a card or something...

@Robin4

no idea about the lack of 4v , TBH I don't really care about it... it's only for some Cyrix cpus apparently

allright, back at the office for some more testing 😀

First of all, I found this pinout here

So I measured a VCC pin on the socket (K2), on the default 5v setting I get 5v and on the new 3.3v setting I get 3.3v 😀 did this so I can be 100% sure the mod works and the cpu really gets 3.3v .

Next is a good pic of the area:

I have tried all possible combos with JP24 and the JP1/2/4 block, and I found that it outputs 4.19v when there is no jumper installed in any pin. Measured the socket pin and it was 4.19v as well. So for my 486SLE mobo the jumper settings become:

JP1/2/4 ON , JP24 1-2 = 5v
JP1/2/4 OFF , JP24 1-2 = 3.3v
JP1/2/4 OFF , JP24 OFF = 4.19v

Next is the TC29 and TC30 capacitor pads. Their (+) point measures the LT output, either 5v or 3.3v . Do you think I should add them? 2 more 106F tantalum caps? Maybe they help in stability when 3.3v is used..?

And I had a look on the pad behind the LT , it's not conencted to ground or anything, infact it seems it's just a tinned pad on the pcb , clearly made to be used as a heatsink... I'll just bolt the LT there 😀

^ Good detective work Keropi!

thanks!!!
any comment though on the TC29/TC30 caps? 🤣

I bet the 4.2 volts you see is a result of no proper voltage setting (feedback) resistors connected at all. So as VADJ floats freely, the VOUT also is pretty much floating around. Most likely the voltage would change depending on what load you put there, but I won't guess what the voltage would be if you had a CPU connected. I would advise against this setting, although it should not break or anything.

If it would be meant for 4.0 V CPU, the voltage setting would be much more near to 4.0, although 4.2V is just at the +5% limit. Maybe the motherboard is so old that it just has either 5V or 3.3V support?

Regarding the capacitors, if the only bulk capacitance there is only a 10uF tantalum, it might be a good idea to add some more.
The datasheet says the minimum capacitance on output is 10uF (if there is no capacitor on VADJ pin, and I think there is not), and regulator is said to be stable in all conditions with 22uF tantalum output capacitor. So having three 10uF tantalums should work without worrying much. There is no maximum limit for capacitance, but it's usually in the 100uF range for improved transient response with heavy load current changes. But hey, I am just reading the datasheet aloud here 😀 Best thing would be to spy on a motherboard with the regulator already installed, and see what capacitors it has.

ah, thanks for the insight Jepael... well, 4v setting you are going away! 🤣 not needed anyways... I'll just stick to the 5/3.3v ones that are silkscreened... as a matter of fact, I'm gonna remove the jumper block on JP24 and hardwire it again, it does not affect the 5v or 3.3v setting at all, the jumper needs to bridge 1+2 all the time.
And I'm gonna add the extra caps , the LT is already screwed in place 😀

Did you have seen these voltage settings?

ftp://ftp.mrynet.com/Hardware/x/th99full/m/C-D/33003.htm

Seems on TH99 the setting for 4V is different.. Any test possible?

I'm afraid these jumpers are completely different... JP6 on my mobo is lust a jumper near the cpu, not a block to close pins 2&3
I'm gonna measure the cpu vcc during a Quake session now 🤣

edit: using an AMD486/100 and during quake timedemo I measured 3.17v cpu vcc ... I've also added the extra tantalum caps

edit2: about an hour later with quake palying it's demos, the cpu is not even hot, the LT is cool and the cpu vcc is 3.19 ... I'll call it a day, the mod seems a success

edit3: finalized the whole thing:

restored JP24 hard connection, added the 2 extra caps, secured the LT1085... also did the CR2032 mod on the battery area

thank you all for the info/help on this!!!! 😊 😊 😊

Nice battery mod. What diode did you use there?

BTW, Did you notice the three diodes next to the battery? I think you could have just removed the one that charges the battery, and plug in your battery holder. This way you get more volts at the real-time clock from the battery, as there is no drop in the extra diode.

thanks, it's a 1N48 IIRC , nothing special...
I did notice the 3 diodes there, but I don't know which one is responsible for charging... so I just went the easy way (my knowledge is more on the technical side of things, soldering/cutting/assembling stuff...)

Did you also tested to highest processor possible on you motherboard?
Iam still missing around.. I hope mine board would do 4 volt to?? Its handy if it could work in further future... When needed, its always available..

And yes this board needs also work with 4 volt processors, because Cyrix M7 needs that.

Did you buy the tantalum capacitors on ebay?