New PCI card on old motherboard

[keith]

Keith wrote:

The data sheet states:
Form Fator 32-bit PCI v2.2
Operation Voltage 3.3V

Would my motherboard support 3.3V or 5V?

It took a while to refresh my 34YO memory (was thinking about this one
the way to brunch ;-), but there is something here. ISTR at some point
(PCI 2.2?) the spec *required* that 3.3V be supplied to the appropriate
pins (A21, B25, A27, B31, A33, B36, A39, B41, B43, A45, A53, and B54),
even in 5V slots, to support universal cards.

According to my third edition Shanley and Anderson (covering PCI 2.1), it
is "strongly recommended" that 3.3V be supplied in a 5V system, but not
required. ISTR that PCI 2.2 made this a requirement (will check tomorrow
with the fourth edition of S&A).

So... If you're into hacking, you might wire in 3.3V to the connector and
see. If the WiFi card has the two keyways (indicating a universal card)
it should work in a 5V system if it has 3.3V on the bus. ...or replace
the board, though I don't know which socket-7 boards would be 2.2
compliant.

 

[keith]

P5 as this particular Asus or did you mean socket7 ?
I never saw socket 7 board refusing to work with USB 1.1 PCI card
(including HX).

Socket-7 in general, though it may have been the ctappy OSs of the time
from you know who. USB was all over the floor in those days. OTOH, the
USB port on my Tyan 1598-C2 (ci. '99 super-7 and K6-III) works just fine.
It even works with a USB 2.0 flash stick (Crucial Gizmo 2.0).

 

[Grumble]

The HX was 5V PCI only. That may be the issue right there. Though a
3.3V only card should not fit in a 5V PCI slot. A 3.3V card has a
keyway in positions 12 and 13 (5V has these grounded) and a 5V card
has a keyway in positions 30 and 31 (3.3V card has these grounded).
A "universal" card would have a keyway in both positions. Position 1
is at the rear of the case, so if you look at the slots in your T2P4
the key should be towards the front of the case (5V slot). Where are
the keyways on the card?

I'm not quite sure what a 'keyway' is. I assume it is the 'gap' between
the golden pins? If so, my card has two such gaps.

One gap in stead of pins 12-13 and the other in stead of pins 50-51.
(If I've counted right...)

No gap at pins 30-31.

It seems I'm trying to fit a 3.3V card into a 5V PCI slot, then?

 

[Grumble]

It took a while to refresh my 34YO memory (was thinking about this
one the way to brunch ;-), but there is something here. ISTR at some
point (PCI 2.2?) the spec *required* that 3.3V be supplied to the
appropriate pins (A21, B25, A27, B31, A33, B36, A39, B41, B43, A45,
A53, and B54), even in 5V slots, to support universal cards.

According to my third edition Shanley and Anderson (covering PCI
2.1), it is "strongly recommended" that 3.3V be supplied in a 5V
system, but not required. ISTR that PCI 2.2 made this a requirement
(will check tomorrow with the fourth edition of S&A).

The document provided by RusH states:
http://www.pcisig.com/specifications/conventional/conventional_pci/2_2sum1215.pdf

The remainder of this paper gives an overview of the differences between
revisions of 2.1 and the 2.2 draft version of the PCI Local Bus
specification. This list is not comprehensive and the Final version
should be used. [...]

Chapter 4 Electrical Specification

Section 4.3.4.1. Power Requirements, now requires the system designer to
provide 3.3 V rail in the connector. Before it was not required, but
system designer was to provide a way to add it.

So... If you're into hacking, you might wire in 3.3V to the
connector and see. If the WiFi card has the two keyways (indicating
a universal card) it should work in a 5V system if it has 3.3V on the
bus. ...or replace the board, though I don't know which socket-7
boards would be 2.2 compliant.

I guess I'll have to look for a PCI 2.1 compliant 802.11g adapter. The
other solution is a so-called "game adapter" which provides an Ethernet
to Wi-Fi (802.3 to 802.11) bridge. Although they are still somewhat
pricey, in my opinion. (I have an ageless ISA Ethernet card handy.)

 

[daytripper]

Hmm, after a quick web search, it appears there is a problem with some
WiFi (Netgear in particular) and other than PCI 2.2. Indeed Netgear
specifies a PCI 2.2 compliant system for the WG311.

Now I'm curious. What is the issue (I've never seen a real difference
before)? I'll have to read up on the differences between PCI 2.1 and 2.2
tomorrow.

Delayed Transactions, probably...

Roll-outs are often painful...

/daytripper

 

[Grumble]

Yes.. the problem is PCI version incompability...

PCI wifi cards need PCI 2.2 or newer.

And those older boards are only PCI 2.1 compliant.

How do I know? I had the same problem with my BX chipset motherboard
with a PCI wireless card.

Have you ever found a PCI 2.1 compliant 802.11g card?

Have you tried the U.S. Robotics USR5416?
http://www.usr.com/products/networking/wireless-product.asp?sku=USR5416

The specs explicitly mention "PCI 2.1/2.2 compliant".
http://www.usr.com/products/networking/wireless-product.asp?sku=USR5416&type=specs

Have you tried a game adapter (Ethernet to WiFi bridge)?

 

[keith]

I'm not quite sure what a 'keyway' is. I assume it is the 'gap' between
the golden pins? If so, my card has two such gaps.

Yes, a "keyway" is the slot through witch a key goes. The PCI connnector
is "keyed", thus the "keyway" is the slot through which the key passes.
Your card is thus a "universal" card and as such will work in a 5V or 3.3V
PCI bus. *However*, it may require 3.3V on the connector which your board
obviously doesn't supply. Thus it's not recognized because it has no
power supplied.

One gap in stead of pins 12-13 and the other in stead of pins 50-51. (If
I've counted right...)

No gap at pins 30-31.

Sorry, the light isn't great in my "office". You are correct, it's 50-51.

It seems I'm trying to fit a 3.3V card into a 5V PCI slot, then?

It's more like you're trying to fit a universal card into a slot that has
no 3.3V. My bet is that if you even hack in one 3.3V wire from elsewhere
this card will work. ...perhaps not reliably (may need more and lower
inductance), but I think this is the problem.

Yes, I was half-wrong (PCI version didn't matter). It appears that PCI
2.2 is the problem, but only because your board doesn't follow the PCI 2.1
recommendations. ;-) or is that :-(

I did run into this four years or so ago. Fortunately the IBM servers I
was designing for had plenty of 3.3V available at the PCI connectors.

 

[keith]

It took a while to refresh my 34YO memory (was thinking about this
one the way to brunch ;-), but there is something here. ISTR at some
point (PCI 2.2?) the spec *required* that 3.3V be supplied to the
appropriate pins (A21, B25, A27, B31, A33, B36, A39, B41, B43, A45,
A53, and B54), even in 5V slots, to support universal cards.

According to my third edition Shanley and Anderson (covering PCI
2.1), it is "strongly recommended" that 3.3V be supplied in a 5V
system, but not required. ISTR that PCI 2.2 made this a requirement
(will check tomorrow with the fourth edition of S&A).

The document provided by RusH states:
http://www.pcisig.com/specifications/conventional/conventional_pci/2_2sum1215.pdf

The remainder of this paper gives an overview of the differences between
revisions of 2.1 and the 2.2 draft version of the PCI Local Bus
specification. This list is not comprehensive and the Final version
should be used. [...]

Chapter 4 Electrical Specification

Section 4.3.4.1. Power Requirements, now requires the system designer to
provide 3.3 V rail in the connector. Before it was not required, but
system designer was to provide a way to add it.

Sure, you can hack it in! ;-)

I guess I'll have to look for a PCI 2.1 compliant 802.11g adapter. The
other solution is a so-called "game adapter" which provides an Ethernet
to Wi-Fi (802.3 to 802.11) bridge. Although they are still somewhat
pricey, in my opinion. (I have an ageless ISA Ethernet card handy.)

ISA? You're nutz! ;-) PCI cards are a dime-a-dozen. ...and they work.

The other alternative is a USB to WiFi. I think I saw several of these
while browsing about the web on the PCI 2.2 issue. I wasn't paying much
attention to the trivia (like whether they actually work) though.

 

[keith]

Delayed Transactions, probably...

At configuration time? I think the real problem here is the lack of 3.3V
at the PCI connector.

Roll-outs are often painful...

Here M$ is king.

 

[RusH]

Have you tried the U.S. Robotics USR5416?
http://www.usr.com/products/networking/wireless-product.asp?sku=USR
5416

The specs explicitly mention "PCI 2.1/2.2 compliant".
http://www.usr.com/products/networking/wireless-product.asp?sku=USR
5416&type=specs

like I said earlier, its only buzzword to make feature list bigger.
This one doesnt like PCI 2.1 either.

Have you tried a game adapter (Ethernet to WiFi bridge)?

This is his only reliable option.

Pozdrawiam.

 

[Mike Smith]

The P/I-P55T2P4 came with an Intel 430HX chipset, aka Triton 2:

Woo, hoo... famous ol' motherboard, back in the day - one of the first
to do 83 MHz FSB, IIRC. (I had its archrival, the Abit IT5H... ;-)

 

[keith]

Woo, hoo... famous ol' motherboard, back in the day - one of the first
to do 83 MHz FSB, IIRC. (I had its archrival, the Abit IT5H... ;-)

Oh, please! It wasn't the first and when it did it overclocked the piss
out of everything on the board. The HX chipset was *never* rated for
anything above 66MHz.

 

[Mike Smith]

Oh, please! It wasn't the first and when it did it overclocked the piss
out of everything on the board. The HX chipset was *never* rated for
anything above 66MHz.

What's with the attitude? For one thing, I said "*one* of the first".
For another, I never said it wasn't overclocking, nor did I imply that
it was a no-brainer; i.e. that you didn't have to go looking for PCI
cards that could handle the higher clock rate. But I guess some people
just *live* to be nasty to others on Usenet, huh?

 

[keith]

What's with the attitude? For one thing, I said "*one* of the first".

The "attitude" is because I spent a couple of years working to get the S7
bus to 75, 83MHz, then 100MHz, which nothing from Intel ever did (indeed
they made it impossible with their components). Specifications do matter,
though lusers like you don't care about such trivia.

For another, I never said it wasn't overclocking, nor did Iimply that
it was a no-brainer; i.e. that you didn't have to go looking for PCI
cards that could handle the higher clock rate.

Many didn't. If you want to play Russian roulette, please don't let me
stop you.

But I guess some people just *live* to be nasty to others on Usenet, huh?

I guess some people just don't care about reliability or data integrity.
Over-clocking rulz! <gag-cough>

 

[Mike Smith]

The "attitude" is because I spent a couple of years working to get the S7
bus to 75, 83MHz, then 100MHz, which nothing from Intel ever did (indeed
they made it impossible with their components). Specifications do matter,
though lusers like you don't care about such trivia.

Well, if you're so anal about specifications, then why were you trying
to overclock? Can't have your cake and eat it, too. Me, I just did it
'cause it was fun.

Many didn't. If you want to play Russian roulette, please don't let me
stop you.


I guess some people just don't care about reliability or data integrity.

Not when it's my own computer, I'm doing it for fun, and the machine's
got nothing on it but games.

Over-clocking rulz! <gag-cough>

Again, seriously - what's with the arrogant, uptight, holier-than-thou
attitude? Who the hell are you, anyway?

 

[Keith R. Williams]

Well, if you're so anal about specifications, then why were you trying
to overclock? Can't have your cake and eat it, too. Me, I just did it
'cause it was fun.

Ya' twit! I wasn't overclocking. There were *designs* that ran at
75MHz, 83MHz, and (horror-of-horrors) 100MHz. In case you hadn't
figured it out, I don't work for Intel. Sheesh!

"Russian Roulette". LOL <sigh> Have you ever thought of trying decaf?

<back to the coffee thread?> Tried it, don't like it. I go for the
real stuff (Green Mountain "Harvard Blend", thanks); one *large* cup
before starting in to work each day - *HOT*.

It's a *PC*, not a nuclear bomb. Get over yourself.

When one is paid to design reliable hardware, one designs reliable
hardware. To do that one follows specifications, or writes them. One
doesn't just throw a pile of parts on the floor and hope it works.
It's really as simple as that.

Not when it's my own computer, I'm doing it for fun, and the machine's
got nothing on it but games.

Some people do a little more than play Free-Cell with their toys. To
some it's a serious business. If you're happy with crap...
Unfortunately, you're not the only one who thinks this way, which is
why we have so much crap.

Again, seriously - what's with the arrogant, uptight, holier-than-thou
attitude? Who the hell are you, anyway?

Keith

 

[mikele]

I've have the usr 5416 in my PCI 2.1 and it and it works perfectly. I
tried several other G (netgear, linksys and d-link), but none worked.
If you can use a B also the d-link DWL 520 works in a PCI 2.1 slot.
my configuration: Mobo Asus p2b-f, p3@650 Mhz, 256Mb pc100, win98se

 

[keith]

I've have the usr 5416 in my PCI 2.1 and it and it works perfectly. I
tried several other G (netgear, linksys and d-link), but none worked.
If you can use a B also the d-link DWL 520 works in a PCI 2.1 slot.
my configuration: Mobo Asus p2b-f, p3@650 Mhz, 256Mb pc100, win98se

I think it was decided (two months ago, BTW) that PCI 2.1 allowed 3.3V on
the PCI slot, but didn't require it. PCI 2.2 does, this the card
specifies PCI 2.2. It's an iffy proposition assuming it will run on in a
PCI 2.1 slot. Though it may be fine, there is no guarantee.

I had a similar quandry looking for a sound card for my "antique" Tyan
1598C2. The card I ended up with only claimed PCI 2.2 compatability, but
it worked anyway. Claiminng spec conformance doesn't tell all.

 

[tony22r]

Doh! I wish I woulda read this thread Yesterday

I bought a Netgear WG311 Wireless PCI card (PCI2.2) for Asus P2B-
mobo (PCI2.1).. and it wouldn't even Boot
..tried flashing mobo Bios

I hope BestBuy takes it back, I opened it and everything

So any folks here with Older motherboards... what solution worked fo
you

I'm thinking of exchanging for a USB 802.11g Wireless adapter
My mobo is only USB1.1 so transfer rate will be ~12Mbps
I think that should work.. since i only have the $14.95 Verizo
768/128 DSL

 

[Tony Hill]

Doh! I wish I woulda read this thread Yesterday!

I bought a Netgear WG311 Wireless PCI card (PCI2.2) for Asus P2B-F
mobo (PCI2.1).. and it wouldn't even Boot!
.tried flashing mobo Bios.

I hope BestBuy takes it back, I opened it and everything.

So any folks here with Older motherboards... what solution worked for
you?

I'm thinking of exchanging for a USB 802.11g Wireless adapter.
My mobo is only USB1.1 so transfer rate will be ~12Mbps ?
I think that should work.. since i only have the $14.95 Verizon
768/128 DSL.

You've already got your answer, USB. Just make sure that you get a
USB adapter that will work with a USB 1.1 motherboard! (most/all
should).

Ohh, and 12Mbps should be plenty of bandwidth for your DSL connection
which is topping out at less than 1Mbps. Even with a bit of overhead
you've got WAY more bandwidth than you need.