P2B and Wireless G

0 views
Skip to first unread message

Rogers

unread,
Sep 15, 2003, 2:51:08 PM9/15/03
to
Hi All,

Had an old machine lying around and decided to connect it to my wireless
router.

When I install the SMC 2802W card, the system won't post.

All is well without that card installed, and it won't boot even with all
other cards (except the NVidia video card).

Any thoughts?

Thanks.


Paul

unread,
Sep 15, 2003, 4:56:47 PM9/15/03
to
In article <wSn9b.1191$Ch2...@news02.bloor.is.net.cable.rogers.com>,
"Rogers" <news...@sympatico.ca> wrote:

There are a number of card types that don't seem to work in the P2B
family. My guess is it is related to supply voltage. PCI can run
from +3.3 or +5V, and some cards now have both slots cut in the edge
of the board, meaning they are universal. The +3.3V slot is nearer
the faceplate of the card, while the +5V slot is at the other end
of the PCI connector.

The spec for the card is here:
http://www.smc.com/drivers_downloads/library/SMC2802W_DS.pdf

It is powered by +3.3V, which means it must be 5V tolerant, in order
to have the universal slot pattern on the PCI connector.

PCI cards are powered by +3.3V, +5V, and VIO pins. VIO would depend on
the motherboard, and the P2B is probably 5V I/O, so that is the
rail used by the I/O supply on the PCI chips. But, your symptoms suggest
that either the +3.3V or the VIO don't have power on them - I'm guessing
the +3.3V is not powered, and maybe at the time the P2B was designed,
it was acceptable not to power both the +3.3V and +5V supply pins on the
PCI bus.

At the risk of annoying you, here is the pinout of the PCI connector
from PCI version 2.2 spec, page 167. A 32 bit card stops at pin 62, so
the table does have the pinout for 64 bit cards as well. What I suspect
is the Asus motherboard isn't putting +3.3V power on pin 21a, 27a, and
so on. With the card in place, you'll probably find a leakage voltage
sitting on these pins, of maybe 1.3 volts or so. With no cards in
place, this pin could be closer to zero volts. To access these pins
safely, I would solder a wire to the surface of one of the PCI cards
plugged into the motherboard, on the pin in question. This is safer
than reaching inside the connector (or alternately, you could pull the
motherboard from the chassis, flip it up on edge, and measure the
voltage on the pins as they protrude through the motherboard).

PCI 2.2 Page 167 PCI Pinout...

<---5V Board---> Universal Board <--3.3V Board-->
Pin Side B Side A Side B Side A Side B Side A Comments
1 -12V TRST# -12V TRST# -12V TRST# 32bit start
2 TCK +12V TCK +12V TCK +12V
3 Ground TMS Ground TMS Ground TMS
4 TDO TDI TDO TDI TDO TDI
5 +5V +5V +5V +5V +5V +5V
6 +5V INTA# +5V INTA# +5V INTA#
7 INTB# INTC# INTB# INTC# INTB# INTC#
8 INTD# +5V INTD# +5V INTD# +5V
9 PRSNT1# Reserved PRSNT1# Reserved PRSNT1# Reserved
10 Reserved +5V Reserved +Vio Reserved +3.3V
11 PRSNT2# Reserved PRSNT2# Reserved PRSNT2# Reserved
12 Ground Ground KEYWAY KEYWAY 3.3V key
13 Ground Ground KEYWAY KEYWAY 3.3V key
14 Reserved 3.3Vaux Reserved 3.3Vaux Reserved 3.3Vaux
15 Ground RST# Ground RST# Ground RST#
16 CLK +5V CLK +Vio CLK +3.3V
17 Ground GNT# Ground GNT# Ground GNT#
18 REQ# Ground REQ# Ground REQ# Ground
19 +5V PME# +Vio PME# +3.3V PME#
20 AD[31] AD[30] AD[31] AD[30] AD[31] AD[30]
21 AD[29] +3.3V AD[29] +3.3V AD[29] +3.3V<---- Suspect
22 Ground AD[28] Ground AD[28] Ground AD[28]
23 AD[27] AD[26] AD[27] AD[26] AD[27] AD[26]
24 AD[25] Ground AD[25] Ground AD[25] Ground
25 +3.3V AD[24] +3.3V AD[24] +3.3V AD[24]
26 C/BE[3]# IDSEL C/BE[3]# IDSEL C/BE[3]# IDSEL
27 AD[23] +3.3V AD[23] +3.3V AD[23] +3.3V<---- Suspect
28 Ground AD[22] Ground AD[22] Ground AD[22]
29 AD[21] AD[20] AD[21] AD[20] AD[21] AD[20]
30 AD[19] Ground AD[19] Ground AD[19] Ground
31 +3.3V AD[18] +3.3V AD[18] +3.3V AD[18]
32 AD[17] AD[16] AD[17] AD[16] AD[17] AD[16]
33 C/BE[2]# +3.3V C/BE[2]# +3.3V C/BE[2]# +3.3V<---- Suspect
34 Ground FRAME# Ground FRAME# Ground FRAME#
35 IRDY# Ground IRDY# Ground IRDY# Ground
36 +3.3V TRDY# +3.3V TRDY# +3.3V TRDY#
37 DEVSEL# Ground DEVSEL# Ground DEVSEL# Ground
38 Ground STOP# Ground STOP# Ground STOP#
39 LOCK# +3.3V LOCK# +3.3V LOCK# +3.3V<---- Suspect
40 PERR# Reserved PERR# Reserved PERR# Reserved
41 +3.3V Reserved +3.3V Reserved +3.3V Reserved
42 SERR# Ground SERR# Ground SERR# Ground
43 +3.3V PAR +3.3V PAR +3.3V PAR
44 C/BE[1]# AD[15] C/BE[1]# AD[15] C/BE[1]# AD[15]
45 AD[14] +3.3V AD[14] +3.3V AD[14] +3.3V<---- Suspect
46 Ground AD[13] Ground AD[13] Ground AD[13]
47 AD[12] AD[11] AD[12] AD[11] AD[12] AD[11]
48 AD[10] Ground AD[10] Ground AD[10] Ground
49 Ground AD[09] M66EN AD[09] M66EN AD[09] 66MHz /gnd
50 KEYWAY KEYWAY Ground Ground 5V key
51 KEYWAY KEYWAY Ground Ground 5V key
52 AD[08] C/BE[0]# AD[08] C/BE[0]# AD[08] C/BE[0]#
53 AD[07] +3.3V AD[07] +3.3V AD[07] +3.3V<---- Suspect
54 +3.3V AD[06] +3.3V AD[06] +3.3V AD[06]
55 AD[05] AD[04] AD[05] AD[04] AD[05] AD[04]
56 AD[03] Ground AD[03] Ground AD[03] Ground
57 Ground AD[02] Ground AD[02] Ground AD[02]
58 AD[01] AD[00] AD[01] AD[00] AD[01] AD[00]
59 +5V +5V +Vio +Vio +3.3V +3.3V
60 ACK64# REQ64# ACK64# REQ64# ACK64# REQ64#
61 +5V +5V +5V +5V +5V +5V
62 +5V +5V +5V +5V +5V +5V 32bit end
KEYWAY KEYWAY KEYWAY 64bit spacer
KEYWAY KEYWAY KEYWAY 64bit spacer
63 Reserved Ground Reserved Ground Reserved Ground 64bit start
64 Ground C/BE[7]# Ground C/BE[7]# Ground C/BE[7]#
65 C/BE[6]# C/BE[5]# C/BE[6]# C/BE[5]# C/BE[6]# C/BE[5]#
66 C/BE[4]# +5V C/BE[4]# +Vio C/BE[4]# +3.3V
67 Ground PAR64 Ground PAR64 Ground PAR64
68 AD[63] AD[62] AD[63] AD[62] AD[63] AD[62]
69 AD[61] Ground AD[61] Ground AD[61] Ground
70 +5V AD[60] +Vio AD[60] +3.3V AD[60]
71 AD[59] AD[58] AD[59] AD[58] AD[59] AD[58]
72 AD[57] Ground AD[57] Ground AD[57] Ground
73 Ground AD[56] Ground AD[56] Ground AD[56]
74 AD[55] AD[54] AD[55] AD[54] AD[55] AD[54]
75 AD[53] +5V AD[53] +Vio AD[53] +3.3V
76 Ground AD[52] Ground AD[52] Ground AD[52]
77 AD[51] AD[50] AD[51] AD[50] AD[51] AD[50]
78 AD[49] Ground AD[49] Ground AD[49] Ground
79 +5V AD[48] +Vio AD[48] +3.3V AD[48]
80 AD[47] AD[46] AD[47] AD[46] AD[47] AD[46]
81 AD[45] Ground AD[45] Ground AD[45] Ground
82 Ground AD[44] Ground AD[44] Ground AD[44]
83 AD[43] AD[42] AD[43] AD[42] AD[43] AD[42]
84 AD[41] +5V AD[41] +Vio AD[41] +3.3V
85 Ground AD[40] Ground AD[40] Ground AD[40]
86 AD[39] AD[38] AD[39] AD[38] AD[39] AD[38]
87 AD[37] Ground AD[37] Ground AD[37] Ground
88 +5V AD[36] +Vio AD[36] +3.3V AD[36]
89 AD[35] AD[34] AD[35] AD[34] AD[35] AD[34]
90 AD[33] Ground AD[33] Ground AD[33] Ground
91 Ground AD[32] Ground AD[32] Ground AD[32]
92 Reserved Reserved Reserved Reserved Reserved Reserved
93 Reserved Ground Reserved Ground Reserved Ground
94 Ground Reserved Ground Reserved Ground Reserved 64bit end

|+5V supply
|
/ \
/ \
+3.3 +5 +Vio
| | |
x | | <--- Motherboard PCI connector here
| |
| |

| x |
| | | <--- SMC tries to use +3.3V and +VIO
-------------
| SMC2802W |
--------------
|
GND

In the figure above, the P2B is probably a +5V PCI motherboard.
The +5V pins and the +Vio pins are probably powered by the +5V
supply. But, maybe the +3.3V didn't get any power.

On the SMC side, the SMC could have had a power circuit that
selects and converts power from either +3.3V or +5V. This would
solve the problem, but increase product cost. Instead, the chip
on the SMC probably connects straight to +3.3, which in this
case may not be powered. As a result, the core of the SMC chip
isn't powered.

That is my theory at least, or where I'd start looking. And,
I don't know if it is safe to connect +3.3V from your PS to
those pins - I don't know what other shortcuts may have been
taken in the name of economy...

Paul

P2B

unread,
Sep 15, 2003, 9:59:06 PM9/15/03
to

The +3.3V LED on my PCI POST diagnostic card lights up on a P2B. Not
definitive, but a pretty good indication +3.3 is supplied IMHO.

Newsgroups eMail

unread,
Sep 16, 2003, 12:42:52 AM9/16/03
to
Put in a Linksys card, and all is well.

Must be specific to the SMC type.

Thanks for your time.

"Paul" <nos...@needed.com> wrote in message
news:nospam-1509...@192.168.1.177...

Paul

unread,
Sep 16, 2003, 1:40:09 AM9/16/03
to
In article <w7u9b.3387$hF3.4...@news20.bellglobal.com>, P2B
<p...@sympatico.ca> wrote:

> Paul wrote:
> > In article <wSn9b.1191$Ch2...@news02.bloor.is.net.cable.rogers.com>,
> > "Rogers" <news...@sympatico.ca> wrote:
> >
> >
> >>Hi All,
> >>
> >>Had an old machine lying around and decided to connect it to my wireless
> >>router.
> >>
> >>When I install the SMC 2802W card, the system won't post.
> >>
> >>All is well without that card installed, and it won't boot even with all
> >>other cards (except the NVidia video card).
> >>
> >>Any thoughts?
> >>
> >>Thanks.
> >
> >
> > There are a number of card types that don't seem to work in the P2B
> > family. My guess is it is related to supply voltage. PCI can run
> > from +3.3 or +5V, and some cards now have both slots cut in the edge
> > of the board, meaning they are universal. The +3.3V slot is nearer
> > the faceplate of the card, while the +5V slot is at the other end
> > of the PCI connector.
>

> The +3.3V LED on my PCI POST diagnostic card lights up on a P2B. Not
> definitive, but a pretty good indication +3.3 is supplied IMHO.
>

Have you ever debugged a non-working PCI card on one of the P2B
family ? I see reports of various cards that work or don't work -
for example, I think I saw a post by a person who tried three
USB cards, of which one worked and the other two didn't. Same
symptoms as the OP. Either the BIOS is becoming very confused
by the declaration info on the card, or there is some kind of
voltage problem.

I read another post in Google, where a person measured the +3.3V
and found it to be 1.3V. Now, either this is current flowing
backwards through CMOS I/O and charging a non-connected supply rail
to that voltage, or there is an overload, and where ever that 3.3V
comes from is dropping out.

I've never had any problems here with my old PCI cards, so haven't
had a chance to debug a problem like this.

Paul

Paul

unread,
Sep 16, 2003, 2:07:57 AM9/16/03
to
In article <gxw9b.35687$DZ....@news04.bloor.is.net.cable.rogers.com>,
"Newsgroups eMail" <newsmailR...@sympatico.ca> wrote:

> Put in a Linksys card, and all is well.
>
> Must be specific to the SMC type.
>
> Thanks for your time.
>

This is a real puzzler... Same thing as the USB cards that a poster
in Google tried, one brand worked and two others didn't. Maybe some
day, we'll find someone with a P2B-xx, a new PCI card that doesn't
work, and a PCI POST card, so we can find out how far through the
BIOS code it gets before dying.

Paul

P2B

unread,
Sep 17, 2003, 1:24:03 AM9/17/03
to

Paul wrote:
> In article <w7u9b.3387$hF3.4...@news20.bellglobal.com>, P2B
> <p...@sympatico.ca> wrote:
>
>
>>Paul wrote:
>>
>>>In article <wSn9b.1191$Ch2...@news02.bloor.is.net.cable.rogers.com>,
>>>"Rogers" <news...@sympatico.ca> wrote:
>>>
>>>
>>>
>>>>Hi All,
>>>>
>>>>Had an old machine lying around and decided to connect it to my wireless
>>>>router.
>>>>
>>>>When I install the SMC 2802W card, the system won't post.
>>>>
>>>>All is well without that card installed, and it won't boot even with all
>>>>other cards (except the NVidia video card).
>>>>
>>>>Any thoughts?
>>>>
>>>>Thanks.
>>>
>>>
>>>There are a number of card types that don't seem to work in the P2B
>>>family. My guess is it is related to supply voltage. PCI can run
>>>from +3.3 or +5V, and some cards now have both slots cut in the edge
>>>of the board, meaning they are universal. The +3.3V slot is nearer
>>>the faceplate of the card, while the +5V slot is at the other end
>>>of the PCI connector.
>>
>>The +3.3V LED on my PCI POST diagnostic card lights up on a P2B. Not
>>definitive, but a pretty good indication +3.3 is supplied IMHO.
>>
>
> Have you ever debugged a non-working PCI card on one of the P2B
> family ?

No, because I haven't met a non-working PCI card on one of the P2B family.

>I see reports of various cards that work or don't work -
> for example, I think I saw a post by a person who tried three
> USB cards, of which one worked and the other two didn't. Same
> symptoms as the OP. Either the BIOS is becoming very confused
> by the declaration info on the card, or there is some kind of
> voltage problem.
>
> I read another post in Google, where a person measured the +3.3V
> and found it to be 1.3V. Now, either this is current flowing
> backwards through CMOS I/O and charging a non-connected supply rail
> to that voltage, or there is an overload, and where ever that 3.3V
> comes from is dropping out.
>
> I've never had any problems here with my old PCI cards, so haven't
> had a chance to debug a problem like this.
>
> Paul

Same here - never had a problem with old PCI cards (except on unmodified
boards @133Mhz FSB - which seriously overclocks the PCI bus) and never
been forced to buy a new one that someone somewhere hadn't said worked
on a P2B :-)

P2B

unread,
Sep 17, 2003, 1:35:49 AM9/17/03
to

Paul wrote:

> In article <gxw9b.35687$DZ....@news04.bloor.is.net.cable.rogers.com>,
> "Newsgroups eMail" <newsmailR...@sympatico.ca> wrote:
>
>
>>Put in a Linksys card, and all is well.
>>
>>Must be specific to the SMC type.
>>
>>Thanks for your time.
>>
>
> This is a real puzzler...

Indeed.

> Same thing as the USB cards that a poster
> in Google tried, one brand worked and two others didn't. Maybe some
> day, we'll find someone with a P2B-xx, a new PCI card that doesn't
> work, and a PCI POST card, so we can find out how far through the
> BIOS code it gets before dying.
>
> Paul

I have all of the above except for the new PCI card that doesn't work.

Maybe some day I'll get one and either make it work or figure out why
not :-)

Paul

unread,
Sep 17, 2003, 5:19:52 AM9/17/03
to
In article <GoS9b.6267$hF3.7...@news20.bellglobal.com>, P2B
<p...@sympatico.ca> wrote:

Another data point - the problem of cards working or not seems to be
happening on some Abit boards based on the 440BX. So, it could be
a 440BX problem, or a problem related to the reference design provided
by Intel back in the 440BX era. Another phenomenon seen, is a card
can be "invisible" to the system, in that it doesn't even respond
to probes of the config space (so, a system posts but Windows doesn't
see the hardware).

In the datasheets for the 82443 and 82371, these chips are
"PCI Rev. 2.1, 3.3V and 5V, 33MHz Interface Compliant", so it isn't
a spec version problem.

I also read some comments about the number of bus masters allowed in
a 440BX system. From the datasheet, it says:

"...five external PCI bus masters in addition to the I/O
bridge (PIIX4/PIIX4E)."

There is at least one 440BX design that has no bus mastership
capability on one of the PCI slots. Maybe a P2B-LS could suffer
from this, as LAN+SCSI+4*PCI > 5 bus request signals ?

Paul

Reply all
Reply to author
Forward
0 new messages