I recently decided to upgrade my old Dell Dimension 4100, and after
changing my RAM, HDD, video card, etc. I decided to upgrade the
processor, it has the best Coppermine PIII available, but this is only
1.1 GHz, and 256k L2, so I figured I would get an upgrade adapter, the
only one I could find however was a Lin-Lin socket 370 fc-pga to fc-
pga 2 adapter, which I figured would be fine to upgrade to a Tualatin
PIII SL6BY. However, Upon using the adapter as originally configured
(original jumper settings), I turned the comp on to find that the
system wouldn't boot, and i received a diagnostic light code in which
only the "b" light was illuminated with a green LED, which according
to the Dell documentation (
http://support.dell.com/support/edocs/
systems/dzuul/codemess.htm#diagnostic_codes), means that a "PCI bus
failure has occurred". However this seems unlikely, as I had not
altered the PCI cards configuration in anyway, and even after removing
all of them from the system, I received the same error. After
returning to the previous CPU, My system booted properly, and since I
have seen reports of people being successfully able to get this to
work with the same adapter, CPU, and Dimension, I am confused as to
what I am doing wrong. I would be most appreciative of advice given
from someone who has used this adapter with the same mo-bo i am using
(basically an intel i815), or even better, someone who has used it on
a dimension 4100, or even with the same processor. Any and all help is
appreciated, so please post away!
The Lin-Lin has jumpers, some of which control Vcore VID value,
and the others control the CPU input clock (BSEL two bit code = 66,100,133MHz).
So you have to set those appropriately. The documentation on
the back of the box is misleading, and there are articles in
web forums, with the information that people have gathered
on it.
(Examples of two generations of voltage regulators. Use the first
two docs, as a reference on Vcore values and VID bits.)
(VID table page 8)
http://www.intersil.com/data/FN/FN4/FN4567/FN4567.pdf
(VID table page 7)
http://www.intersil.com/data/fn/fn4417.pdf
The SL6BY is 1.5V and the 512KB cache is intended for server systems.
http://processorfinder.intel.com/Details.aspx?sSpec=SL6BY
There are two bits of info for the Lin-Lin jumpers. You can find the
"standard table" floating around, but it doesn't tell the whole story.
Basically, what this table shows, is which signals to ground, to
set the four bit VID code and the two bit BSEL code, without referring
to what the processor is sending. So all of these codes are "override
codes", choosing to ground a signal (by using a jumper) or not ground
a signal (by not using a jumper at all with the signal being controlled).
http://discussions.hardwarecentral.com/showthread.php?t=172281
Override codes. Broken into two sets. First set controls Vcore.
Second set controls BSEL. The two sets are independent of one
another. Take the 2.05V as an example. To override the processor voltage
request, and force 0000 binary code, requires four jumpers to ground
four signals. In the case of 1.35V, the code is 1110, and only one
jumper is needed to assert the single logic 0 needed. The other three
VID signals are allowed to float to logic 1.
no jumpers a1 thru b3 1.30V
a8-a9 1.35V
a2-a3 1.40V
a2-a3 a8-a9 1.45V
a5-a6 1.50V
a5-a6 a8-a9 1.55V
a2-a3 a5-a6 1.60V
a2-a3 a5-a6 a8-a9 1.65V
b2-b3 1.70V
a8-a9 b2-b3 1.75V
a2-a3 b2-b3 1.80V
a2-a3 a8-a9 b2-b3 1.85V
a4-a5 b2-b3 1.90V
a8-a9 b2-b3 1.95V
a2-a3 a5-a6 b2-b3 2.00V
a2-a3 a5-a6 a8-a9 b2-b3 2.05V
BSEL is a two bit code, sent from the processor to the clock generator chip.
66MHz must be binary 00, requiring two grounding jumpers. 133MHz must be binary 11,
and for that, no jumpers are installed, resulting in the two signals floating to
logic 1.
b5-b6 b8-b9 66MHz
b8-b9 100MHz
no jumpers b4 thru b9 133MHz
In the post here, someone traced the pins not used. The remaining pins
are "passthru" pins, that would pass the value the processor is sending.
The processor is coded for the Vcore VID it needs to run, and also
controls the BSEL pins, to set the CPU input clock.
http://www.bp6.com/board/viewtopic.php?t=1134&postdays=0&postorder=asc&start=75
*******
"bardos, when you jumper a1-2, a4-5 and a7-8 you are allowing the VIDx signals to
pass from the CPU for autodetect....reason you saw a 1.30 when you boot is that
you didn't pass VID3 for autodetect, you floated it.
I probed my lin lin today... A1, A4, A7, B1, B4, B7 allow the passthru of the
adjacent signals (when jumpered) to the CPU to allow auto detection of voltages
and fsb. The lin lin box instructions then interpret :
For FCPGA1 (cel2 or p3 cu), on PPGA board, use auto detect.
For FCPGA2 on FCPGA boards , examples are for setting 1.5v (intel board)and 1.45v
(VIA/SIS) respectively with fsb=100
A3, A6, A9, B3, B6, B9 tie adjacent signal to Vss
i.e. a logical 0 for the intel VIDx table in the intel datasheets
B2 = VID3
A5 = VID2
A2 = VID1
A8 = VID0
B5= BSEL0
B8 = BSEL1
*******
So, for the six named signals (four VID logic signals, selecting 16
different voltage values, and two BSEL signals selecting 66,100,133),
each three pin section looks like this. The following circuit is repeated
six times, and uses a total of 18 jumper pegs on the Lin-Lin.
+
|
Pullup_resistor (makes logic 1,
| if no jumper is used)
Passthru_from_processor ----------X |
X----+----> To motherboard (VID to Vcore,
Logic_0_grounding_the_signal -----X BSEL to clockgen)
options - no jumper (force logic 1)
- upper X to center X (passthru)
- lower X to center X (force logic 0)
The voltage regulator on the motherboard, must be designed for
the task. Voltage regulators of that era came in two types.
Ones that delivered voltages as low as 1.8V (not suited to
running a Tualatin), and ones that offered more of the
Vcore VID table values all the way to 1.5V (suitable for
Tualatin). I expect your Vcore regulator is OK for this.
(The two Intersil documents above, are examples of the two possible
types of Vcore regulators used back then.)
(On my motherboard, I changed to a pin compatible regulator
chip, so I could get 1.5V, but it was a pretty dangerous
soldering mod. I accidentally removed an unused copper pad
while doing the mod, but the board still worked with the
Tualatin when I was finished.)
So one symptom that an upgrader can see, is a "black screen
and no beeps". When that happens, it means the Vcore regulator
is the 1.8V type, and when the VID bit pattern for a voltage
less than 1.8V is fed to it, the output is set to zero volts
for safety. In that case, there is no damage to the processor.
Another issue, is BIOS recognition of the processor. In that
era, there were issues with the multiplier value locked inside
the processor, and certain higher multiplier values could cause
a problem when the BIOS parses the CPU type during POST.
One suggested test for a BIOS/multiplier problem, is to force
the CPU input clock to 66MHz. Apparently, that will allow
some of the BIOS multiplier problems to be bypassed, so you can at
least prove that the voltage issue is OK. (You're testing at 66MHz,
to prove that the processor works.) So, on your Lin-Lin,
you're going to have to dig up the proper documentation for the
Ax-Bx jumpers, so that you can set the thing to 66MHz.
On my 440BX motherboard, I installed the latest BIOS available, and
that had a fix for the multiplier problem. But the BIOS did not
have the microcode patch for the Tualatin (not really a big deal),
and I fixed that with the CTMC package (heise.de). It means taking
a 2KB microcode file from another BIOS, and using a built-in feature
of the BIOS, to store the new microcode in a "volatile" location
in the BIOS EEPROM chip. As long as you didn't screw around with
changing the processor to different models, the microcode would
stay put. The microcode patch would remove any warning message
seen during the BIOS POST sequence, about no microcode being
present for the processor. WinXP has a microcode loader, so
at least for that OS, a microcode is loaded once WinXP is booted.
(I haven't checked to see if it is still here.)
ftp://ftp.heise.de/pub/ct/ctsi/ctmc10.zip
Other than that, I couldn't find an article that addressed your
exact configuration. On my computer, I used an Upgradeware Slot-T
adapter, as my computer is a slot 1 and needs a different adapter.
Hope that gives you a few ideas,
Paul
