VOGONS

Socket 7 EPoX EP-MVP3C Voltage Regulator issues

Topic actions

First post, by smesgr

Posted on 
User metadata
Rank Newbie
Rank
Newbie

Hi,

I (again) have an issue with my EPoX EP-MVP3C after replacing the buck converter the board was working several month and does not boot anymore.

The issue:
- 3.3V & 5V is there
- POST code shows "--"
- Voltage regulator for the CPU (Q10 on image) does not provide selected voltage
- LM2636 has VCC and Boost Voltage, VIDs are set correctly to (in this case 2.8V)
- Output Enable of LM2636 floats around 1.5V to 2V. According to Specification this is near the disable trigger point

If you power on/off several times - there is a chance the system boots, Voltage regulator provides the requested voltage and the system is running some seconds before voltage is dropped and the system hangs. Thus I think the output enable circuit is to blame. But I can't figure out where the traces are going. Has anybody an idea?

The OE pin 19 on LM2636(U15) is connected to 10k resistor R176 and Q11 (likely a NPN - top pin - likely collector pin). One of the Q11 pins goes to R175 and R174 and than to the socket. But I couldn't find any connected pin. The other Q11 goes to GND. R174 and R176 are connected to 5V. Between R174 and U15 is also a Via which i couldn't figure out.

In the picture R174 is missing because i cracked it while probing something

Reply 1 of 4, by smesgr

Posted on 
User metadata
Rank Newbie
Rank
Newbie

I have now desolder the Q11 chip. Which turned out to be an NPN transistor with hFE ~200 and Ube ~700mA. If it is a 3904 it is in specification.
Without the NPN the system boots fine. I'm certain this NPN is used as an inverter. The input of the "inverter" at the base is very low. I'm so confused what is going on.

Reply 2 of 4, by rasz_pl

Posted on 
User metadata
Rank l33t
Rank
l33t

Regulator has separate powergood input. Why is this outputEnable going from the CPU socket mechanism there in the first place? I would just leave it as is with no transistor.
edit: Im looking thru my collection of schematics/boardviews. Not many SS7 boards, but the ones I have dont do any socket magic. For example https://theretroweb.com/motherboards/s/gigaby … -ga-5aa-rev-3.2 RC5057 has PWROK linked to chipset generated powergood and EN pin strapped to VCC thru 47K and 0.1uF to ground = starts disabled until cap charges.

No idea what the designers thought with that "goes to Pentium 2 socket" in LM2636 datasheet.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 3 of 4, by smesgr

Posted on 
User metadata
Rank Newbie
Rank
Newbie

thanks for your input. Because of you I started probing again - this time with an CPU inside the socket AND oh surprise - Now it is shorted to ground. Without the CPU it isn't. I still could not find out where the trace is going. But seems like it is just a protection to not power on the circuit without load.

As described here: Buck Regulator Controller LM2636 - Alternatives? I got this replacement LM2636 from china and they still give me a look if they aren't genuine chips. Maybe they are chips which failed the QS or reproduction chips. I try to rebuild the circuit. The seen behaviour would make sense if the OUTEN doesn't have a pull up resistor as described in the specification but is pulled down instead. But I think i will do as you describe and just leave the circuit as it is - as it seems just to be an protection to the power regulation circuit.

Reply 4 of 4, by rasz_pl

Posted on 
User metadata
Rank l33t
Rank
l33t

What I dont get is protection from what? from letting the user measure output voltage with no CPU inserted? 😀

It could feasibly make some limited sense on a SLOT1 board because there CPU is controlling VID pins = Idiot pushing CPU into powered board can generate random VID state causing unpredictable voltage going into CPU, and as LM2636 Rated Output goes all the way to 3.5V this could be catastrophic. But on S5/7 VIDs are controlled by onboard jumpers so no risk of unpredictable state

Someone was overthinking this when designing this EPoX board.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Go to top of page Go to top of page
Back to General Old Hardware
Close window
Close window