VOGONS


First post, by Eleanor1967

User metadata
Rank Member
Rank
Member

hey folks

I've owned this nice LS-486e rev D motherboard (picture from the internet) for quite some time now, and I always remember it being a bit unreliable, at least when using a Cyrix 5x86. A month or two ago I was playing around with it, testing different CPU's and stuff, trying to get it stable. In the heat of battle, swapping out different CPU's left and right, I inserted a CPU 90 degrees turned in the socket. Luckily, the CPU and the board survived and at least the CPU is still working perfectly, but the board is performing quite bad ever since. It is very slow, speedsys detects the CPU at 53 MHz even though it is jumpered to 100. Memory throughput (with a single 16MB of EDO) is only 16,6 MB/s. VGA memory throughput is shit too, you get the idea. Interestingly, the performance stays exactly the same when I put a jumper on the turbo switch.

So question is: Does anybody have an idea what I could have possible grilled that it would behave this way? Is their a likelihood that I can fix it, just by replacing a component/chip or two?

In other, slightly unrelated news the Cyrix 5x86 QFP chip I posted about in the bought this hardware thread is working perfectly (at 53 MHz) in this board, so that was a nice surprise this evening. 😀

Cheers, Eleanor1967

Reply 1 of 50, by Auzner

User metadata
Rank Member
Rank
Member

Wow, neat. Some how you messed up all of your bus speeds. Try replacing the crystal oscillator. Wild guess but maybe it was damaged and that changed its resonance freq which references everything on the board.

Reply 2 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member
Auzner wrote:

Wow, neat. Some how you messed up all of your bus speeds. Try replacing the crystal oscillator. Wild guess but maybe it was damaged and that changed its resonance freq which references everything on the board.

This was my first idea as well, so I guess I will start searching for that part. It does not however explain (to my understanding at least) why the board acts the same with the turbo switch jumper on. Maybe it was happening before though, I never tested it.

EDIT: One of these should do it, right? http://www.ebay.de/itm/1x-TIC-Quarzoszillator … TUAAOSwCypWoe0R

Reply 3 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie
Auzner wrote:

Try replacing the crystal oscillator.

Don't, it has nothing to do with it. It's more likely that you have damaged "turbo" circuitry, which now permanently halves your FSB. Measure CPU frequency or benchmark some very well known CPU for comparison to be sure that it's the case.

Reply 4 of 50, by Auzner

User metadata
Rank Member
Rank
Member

It's the clock synthesizer IC which is damaged? I'm guessing it's that "MX" IC below the crystal.

Cool, there's a thread on here about clock gens: Clock generators on Socket 7 and older motherboards

Datasheet: http://pdf.dzsc.com/88889/9931.pdf

Reply 5 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie

It may or it may not, looking at PPLs block diagram, two clock outputs can be seen going from PLL0, one is used for default FSB setting, another - halves it. At current state motherboard gets only 1/2 of the supposed clock, regardless of "turbo" settings. OP has to trace those pins to turbo circuitry 1st.

Reply 6 of 50, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Is it the keyboard controller which adjusts timing for turbo, the BIOS itself, or the chipset (via command of the BIOS, via command of the KBC?)?

Plan your life wisely, you'll be dead before you know it.

Reply 7 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member

So I soldered the crystal back in..

SSTV2 wrote:

Measure CPU frequency or benchmark some very well known CPU for comparison to be sure that it's the case.

As I said, speedsys is telling me that the Cyrix 5x86 which is jumpered to 100 MHz and is also deteced at POST as 100MHz is running at 53 MHz. I get a CPU score of 26,47 which I would say is realistic for that CPU at this clock and unoptimized BIOS settings if you compare to what feipoa is getting in his 486 benchmark comparison. I got similar bad performance with another two 5x86 and an AMD DX4 when I tested a month or so ago.

SSTV2 wrote:

It may or it may not, looking at PPLs block diagram, two clock outputs can be seen going from PLL0, one is used for default FSB setting, another - halves it. At current state motherboard gets only 1/2 of the supposed clock, regardless of "turbo" settings. OP has to trace those pins to turbo circuitry 1st.

.

I'm not 100% certain that I know what you want me to do, I attached an image where I checked connections with the multimeter between the turbo pins and the PLL. I did not find any connections for the upper turbo pin.

feipoa wrote:

Is it the keyboard controller which adjusts timing for turbo, the BIOS itself, or the chipset (via command of the BIOS, via command of the KBC?)?

I do not know that. I also did not find a datasheet for the keyboard controller.

Btw, the datasheet stats that the PLL can also do 75 MHz, I do wonder if you can set it to that on the board.

Attachments

  • pinsls486e.jpg
    Filename
    pinsls486e.jpg
    File size
    4.71 MiB
    Views
    1473 views
    File license
    Fair use/fair dealing exception

Reply 8 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie
Eleanor1967 wrote:

I'm not 100% certain that I know what you want me to do, I attached an image where I checked connections with the multimeter between the turbo pins and the PLL. I did not find any connections for the upper turbo pin.

You need to figure out whole clock switching circuit, that goes in between PLL chip, turbo header and CPU socket with northbridge, there lies your problem. Start tracing PLLs pin 9 and 24 to turbo header, there might be some logic gate circuits that does clock signal switching in logic level.

Reply 9 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member

Just a quick update. I also ran cachechk to confirm the CPU speed and indeed it is reported as 49,5 MHz.

I've previously found out that PIN 26 from the PLL (S0) is connected to the lower turbo jumper pin.
The upper turbo jumper pin goes to the SIS497 chip, but since I did not find a datasheet I have no idea where it is going from there.
I've also found out that PIN 20 from the PLL (BOUT10) is connected to the CPU socket CLK pin.

Attachments

  • morels486.jpg
    Filename
    morels486.jpg
    File size
    2.72 MiB
    Views
    1399 views
    File license
    Fair use/fair dealing exception

Reply 11 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie

Turbo header connects to SIS 497 pin 39 which is "DETURBO" and from there turbo function is controlled by adjusting certain registers bits (165 pg. from chipset datasheet) so the only thing that controls turbo mode is that chip. I think that PLLs CPUCLK mode is adjusted by MS0 pin, when it's high (1), CPUCLK (pin 24) is selected for FSB freq., when it's low (0), then 1/2 CPUCLK (pin 9) should be selected. Perhaps one of the SMOUT pins on SIS497 chip connects to the PLLs MS0? By default MS0 should be pulled high (3.3 or 5v).

Pin 26 (S0) is not connected to the lower turbo jumper pin, it's simply grounded because you've set it so with that red jumper above PLL. If BOUT10 connects to CPU clock pin, then both CPUCLK and 1/2 CPUCLK signals should be buffered through PLLs internal buffers (BIN0,1,2). Figure out to which buffers pin 9 and 24 connects to and where buffer outputs meet.

Also, jumper that JP16 and if turbo mode is controlled in register level, make sure that BIOS are set to default settings, CPU is compatible with motherboard and all jumpers are set correctly. I've also got one motherboard that uses same SIS 496/497 chipset and I can't see any difference between turbo being toggled on or off as long as I load some more demanding software, because in my case, it doesn't adjust FSB.

Reply 12 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member

It works!!! I just got 45 points in speedsys!

The unsatisfying thing is that I don't know what I did to fix it. I was just going through SSTV2's last post (No I did not find any connection between the SiS497 and MS0; and yes I measured 4,7 volts on MS0 so its pulled high by default.) I did a BIOS reset but since I wasn't using a battery to begin with the board had reseted itself days ago. I also didn't change a jumper, JP16 was jumpered all the time.

I want to thank you guys for helping me out here, I at least learned something along the way 😀

Just one quick question left. I just measured a VCC pin in the socket (that's the pin where current goes in the CPU right?), because the Cyrix was running quite hot even after a short time, and got a reading of solid 4,9x volt. That's not normal for the board being jumpered for 3,3 volt??

Reply 14 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie

Does CPU work at full frequency (100MHz) now? If you get a 5V reading at CPU VCC pins, then voltage is unregulated or in other words, PSUs 5V rail connects directly to CPU VCC. I guess LDO regulator on that motherboard has died which keeps power transistor closed at all times (input is shorted with output).

Reply 15 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member
SSTV2 wrote:

Does CPU work at full frequency (100MHz) now? If you get a 5V reading at CPU VCC pins, then voltage is unregulated or in other words, PSUs 5V rail connects directly to CPU VCC. I guess LDO regulator on that motherboard has died which keeps power transistor closed at all times (input is shorted with output).

Yes it does run at 100 MHz now. If stopped using the 5x86 since 5V is obviously not healthy for that chip. I did jumper it for a 486DX and my DX-50 runs fine at 40 MHz, I get 15 in speedsys which I would say is about right for unoptimized BIOS settings.

These should do the job, right?

https://www.ebay.de/itm/5-Stuck-5-pieces-TIP1 … g4AAOSwAC1Z51KW

Reply 16 of 50, by SSTV2

User metadata
Rank Oldbie
Rank
Oldbie

Don't jump to conclusions so early, it's more likely that LDO reg. is bad. Remove all voltage selection jumpers and check for continuity between "outboard" transistors (the one with a heatsink) collector/emitter or source/drain, if it shorts - it needs replacing. Study some application circuits in which a 2951 LDO reg. is used together with outboard transistor, you'll then understand how it works and how can it be tested.

Reply 17 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member
SSTV2 wrote:

Don't jump to conclusions so early, it's more likely that LDO reg. is bad. Remove all voltage selection jumpers and check for continuity between "outboard" transistors (the one with a heatsink) collector/emitter or source/drain, if it shorts - it needs replacing. Study some application circuits in which a 2951 LDO reg. is used together with outboard transistor, you'll then understand how it works and how can it be tested.

I'm afraid you lost me here. I do not understand enough about electronics to learn how to test that 2951 circuit.

I checked connections with the multimeter and made another artful picture. That's all the connections I could find. I would assume that I should find a connections coming from the 2951 pin 1 (out) to somewhere, could this mean its at fault?

Anyway I have to kindly ask you to help me out here once again since I lack the skill (and time necessary to acquire those right now anyway).

Attachments

  • 20180123_colormess.jpg
    Filename
    20180123_colormess.jpg
    File size
    1.03 MiB
    Views
    1214 views
    File license
    Fair use/fair dealing exception

Reply 18 of 50, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

The pass transistor is the device in the TO-220 package (standing up with the metal tab). In the simplest sense, it provides (in the context of this circuit) a controllable resistance based on the input from the 2951 regulator, like an electronically controlled rheostat. Based on your drawing, I'm assuming we're dealing with the base or gate (i.e. the control input) on the right most pin (generally referred to as pin 1) which is typical for TO-220 transistors. If the impedance across the other two pins in either direction is less than 10s or 100s of kOhms, the device has likely failed short, should be removed from circuit (and tested to confirm another part in the circuit hasn't failed), and replaced if confirmed to be bad.

All hail the Great Capacitor Brand Finder

Reply 19 of 50, by Eleanor1967

User metadata
Rank Member
Rank
Member
gdjacobs wrote:

The pass transistor is the device in the TO-220 package (standing up with the metal tab). In the simplest sense, it provides (in the context of this circuit) a controllable resistance based on the input from the 2951 regulator, like an electronically controlled rheostat. Based on your drawing, I'm assuming we're dealing with the base or gate (i.e. the control input) on the right most pin (generally referred to as pin 1) which is typical for TO-220 transistors. If the impedance across the other two pins in either direction is less than 10s or 100s of kOhms, the device has likely failed short, should be removed from circuit (and tested to confirm another part in the circuit hasn't failed), and replaced if confirmed to be bad.

Thanks for the detailed explanation. Just to make sure “the other two pins” is referring to the TO-220, right? Not a pin on the LP2951? I just measured the inexpedience and its always jumping between 0 and 0.1 Ohm. If I understand you correct that means its bad, can you/someone please confirm me that the part in post 16 is the correct one to use?