VOGONS


First post, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

I've been keeping it for a long time (it was upgraded to a Zida 6DLX after the issue happened) but I'd like to try and fix it if possible.

Issues:

- will POST ONLY with any non-MMX Pentium chip, Cyrix 6x86 non-MMX and any CPU that does not require dual-voltage

- will NOT POST with any MMX Pentium chip, AMD K5 (strangely) and any K6-2 chip

- will also NOT POST with any AGP video card

Any ideas where the issue could be? I'm suspecting either the PSR16C40DT near the CPU socket or the CEB703AL MOSFET right below it. Board has already been recapped with newer OST capacitors (2007, old caps were marked En and were dated around November 1998)

Finally, I'll leave a few photos of the board. The CMOS holder is not my work and is exactly hoe I received the board.

Attachments

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 1 of 12, by Roman555

User metadata
Rank Member
Rank
Member
PcBytes wrote on 2020-06-29, 11:30:

- will also NOT POST with any AGP video card
Any ideas where the issue could be?

I suspect it might be a trouble with a 3.3v voltage regulator that is near dimm slots.
Did you measure voltage of the regulators?

Reply 2 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

You mean the big PSR16C40DT near the SDR slots?

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 3 of 12, by Roman555

User metadata
Rank Member
Rank
Member
PcBytes wrote on 2020-06-29, 20:03:

You mean the big PSR16C40DT near the SDR slots?

It's a part of the step down regulator. So measure voltage after the coil (I think it's the left pin) or on the left capacitor (see your last photo). My idea is the voltage drops when load becomes bigger. Maybe the capacitors of this VRM was soldered not properly during recap.

Reply 4 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

I have 3.27 on both pins of the PSR16C40DT near the SDR slots, and 2.8V on the cap next to it. The one at the corner of the board (to the right of the coil) reads 4.9V (so around 5v). This is without an AGP card.

With AGP card (Rage Pro Turbo AIW), I get 2.8V on both pins of the PSR16C40DT, and 2.79V on the capacitor next to it.

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 5 of 12, by Roman555

User metadata
Rank Member
Rank
Member
PcBytes wrote on 2020-06-30, 08:14:

I have 3.27 on both pins of the PSR16C40DT near the SDR slots, and 2.8V on the cap next to it. The one at the corner of the board (to the right of the coil) reads 4.9V (so around 5v). This is without an AGP card.

With AGP card (Rage Pro Turbo AIW), I get 2.8V on both pins of the PSR16C40DT, and 2.79V on the capacitor next to it.

Sorry, but I don't understand how you have measured voltages on SR16C40DT. I usually measure voltages relatively to GND rail.
But if you've measured voltages 2.79 between two pins of the capacitor than the voltage 2.79v is wrong - it should be 3.3v.

P.S. An article to understand better how the regulator works
Switching Regulator Basics: Step-Down Operation Principles

Reply 6 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

Redid the solder joints on all the caps.

Now AGP cards work (although a bit randomly), tested with both a Geforce 2 Pro and a Rage Pro Turbo AIW.

MMX chips still do not work. I don't get any beeps at all when I power the board on with a Pentium MMX 233. Same chip works on a i430TX mobo.

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 7 of 12, by Roman555

User metadata
Rank Member
Rank
Member

I don't have such board or its circuit manual to be sure in schematics.
I suppose the 3.3v dc-dc converter is a source not only for AGP but for I/O of dual-voltage CPUs. If it can't provide stabilized 3.3v voltage nothing will work.
Did you try to run a MMX CPU and a PCI (not agp) graphic card together?
I also think Geforce 2 pro is a too power hungry for the onboard 3.3v dc-dc. Try something like usual Rage or Trio3d.

P.S. As option you can try to find datasheet of the PWM controller. Description of typical usage of the chip is very useful. Compare typical and real usage and measure nominals of the components.

Another option is to unsolder the coil of the 3.3v VRM and then to get 3.3v from ATX connector. In this case the modding motherboard would stop working from AT PSU.

P.P.S. Also , imo, the reason why single voltage CPU works is the Core CPU VRM works properly (that shown on first photo). Of course it should be proved by measurement.

Reply 8 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

Putting in a MMX CPU will make the board not do anything besides spin fans. I tried a MMX CPU w/ no RAM or anything else besides PC speaker and HSF. No beeps.

I did notice the GF2 had artifacts on this board (as if it doesn't have enough power), while on a newer VIA 693 based Pentium 3 550MHz board it didn't have those artifacts at all. Triple checked for artifacts with three different boards (Zida LX98-CT, Soyo 6BA+IV, Epox EP-3VCM) just to be sure the card wasn't dying, and all boards POST'd without artefacts.. Using the Rage Pro Turbo All-in-Wonder 8MB mentioned in the OP worked perfectly on the Acorp board, no artefacts whatsoever.

I'll be looking at the coils and see what's bothering it. By the way, I tested both AT and ATX, and MMX chips didn't work on either of them.

I also have another board to diagnose in parallel - PCChips M577.
That one does NOT start at all with a ATX PSU, and will only start with an AT PSU, but same thing happens with all CPUs - no POST, no beeps, nothing else than fans spinning.
It had quite a lot of missing components:

-two coils - I'm not sure if both were supposed to have 11 turns, and the ones I used both have 10 turns each.
-almost all the bigger caps (470uF 10v and higher)
-AT connector (added it back from a dead UMC491F 486 board)
-missing LPT and floppy connectors (though I can live without these)
-missing RTC holder
-missing jumpers (though this wouldn't be really surprising)

CPUs tested were the same P233MMX, as well as a non-MMX Pentium 150 and a Cyrix 6x86-P133+GP. None yielded any beeps.

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 9 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

Finally, it works!

Apparently it needed both JV_L1 and JV_L2 jumpers to be shorted. I have no idea what do they do (Roman555 if you have any idea what are those used for, please let me know), but once I shorted JV_L1, both AGP and MMX chips started working.

2020-07-06 13.22.07.jpg
Filename
2020-07-06 13.22.07.jpg
File size
1.88 MiB
Views
97 views
File comment
Mobo w/ P233MMX and Rage Pro Turbo 8MB
File license
Fair use/fair dealing exception
2020-07-06 13.23.31.jpg
Filename
2020-07-06 13.23.31.jpg
File size
1.34 MiB
Views
97 views
File comment
POST screen showing Pentium MMX.
File license
Fair use/fair dealing exception

Disregard the 166MHz speed. That's probably my mistake on the jumpers (I set 66x2.5) and will be corrected somewhat (I plan on running it at 100x2.5) but for now it's great.

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 10 of 12, by Roman555

User metadata
Rank Member
Rank
Member
PcBytes wrote on 2020-07-06, 10:27:
Finally, it works! […]
Show full quote

Finally, it works!

Apparently it needed both JV_L1 and JV_L2 jumpers to be shorted. I have no idea what do they do (Roman555 if you have any idea what are those used for, please let me know), but once I shorted JV_L1, both AGP and MMX chips started working.

2020-07-06 13.22.07.jpg
2020-07-06 13.23.31.jpg

Disregard the 166MHz speed. That's probably my mistake on the jumpers (I set 66x2.5) and will be corrected somewhat (I plan on running it at 100x2.5) but for now it's great.

https://www.elhvb.com/mobokive/Archive/Acorp/ … /via/5via77.htm
Your board is A4 version and JV_L1 and JV_L2 isn't mentioned for this version.
I don't have good look of the board to read jumpers designations so I really don't know where they are situated on the board.
I think one of them near the socket and another - near the 3.3v VRM.
IMO, they are needed to switch power properly for single-voltage and dual-voltage CPUs.
When the Core VRM settings is 3.5v or 3.3v some jumpers ( supposely JV_L1 and JV_L2) should connect IO voltage pins and Core voltage pins of single-voltage CPU. And be disconnected for dual-voltage CPUs. But It's difficult to say how exactly the should be set without a good manual or measurements.

P.S. I suppose:
JV_L1 - near the socket 7 - if it is ON it connects I/O voltage pins and Core voltage pins
JV_L2 - near the 3.3V VRM - if it is ON it connects the output of the 3.3v VRM to I/O voltage pins of dual-core CPU.
So they shouldn't be ON simultanously.

P.P.S. IMO, when you shorted both jumpers in the same moment the AGP device, RAM, both power rails of CPU were powered from one working VRM and that is NOT good and even dangerous.

Reply 11 of 12, by PcBytes

User metadata
Rank Oldbie
Rank
Oldbie

Yeah, I'll check if it still works with JV_L2 removed.

And you're right - JV_L2 is near the SDRAM slots, and L1 is near CPU.

EDIT: Tried removing JV_L2 jumper and got met with a surprise - MMX chip now registers as PENTIUM-S CPU at 250MHz (250 is because that's how I set it, 100x2.5). Placing the jumper back makes it correctly identify as PENTIUM-MMX at 250MHz.

Main: i5-3470, 16GB RAM DDR3, Gigabyte B75M-D3H
G.T.2: P2 266MHz, 256MB SDR, Zida/Tomato 6DLX "LX98-AT"
Shaman: P3 650MHz, 384MB RAM, Soyo SY-6BA+ IV

Reply 12 of 12, by Roman555

User metadata
Rank Member
Rank
Member
PcBytes wrote on 2020-07-06, 14:14:

Yeah, I'll check if it still works with JV_L2 removed.

And you're right - JV_L2 is near the SDRAM slots, and L1 is near CPU.

EDIT: Tried removing JV_L2 jumper and got met with a surprise - MMX chip now registers as PENTIUM-S CPU at 250MHz (250 is because that's how I set it, 100x2.5). Placing the jumper back makes it correctly identify as PENTIUM-MMX at 250MHz.

if JV_L1 is shorted then the IO and the Core rails are connected and BIOS recognizes CPU as single voltage P54C one.