From: Eric Johanson (ericj_at_cubesearch.com)
Date: 2002-03-15 09:02:06 UTC
Wow, I learned quite a bit from your post. Thanks for all the great info.
My understanding was that Cardbus was basically '32 bit' cards, regardless of voltage.
I've been bitten by the voltage issue before; DWL-650 cards don't work in most cardbus slots, due to them being 16bit, 3.3 volt. It was rumored that the dlink cards had in fact been 32bit only, but some research into the chipsets (Overheard at wireless meet, YMMV) uncovered the fact that they are only 16 bit.
I'm quite sure that the 'Ricoh Co Ltd RL5c475' is in fact a 32 bit 'cardbus' adaptor - I used a 32bit 5v scsi card (I don't recall the model) in a ricoh bridge, and it worked fine. The same card failed to function on a 16bit 5v notebook (old toshiba). I've also had Lucent Silvie (16bit, 5v) cards work under the ricoh bridges.
Anyway, great info... thanks for posting it.
-Eric
On 15 Mar 2002, Erik Walthinsen wrote:
> On Thu, 2002-03-14 at 23:32, Christopher Hart wrote:
> > (Ricoh Co Ltd RL5c475 pci cardbus bridge and a DWL-650 pcmcia card).
>
> I just want to point out something that I've been learning the hard way
> for a couple months now. Just because an adapter is a PCI card does
> *not* mean it is CardBus. I've been hunting for some kind of adapter
> that will run the Proxim RangeLAN-DS cards I have, which is difficult
> because the cards only work at 3.3v.
>
> I have the Ricoh adapter you have in my hand right now, and it's most
> definitely not CardBus. We also have a TI PCI1410-based adapter that is
> also not CardBus, despite being PCI.
>
> CardBus is a PCI-based bus, which is why the cards look different. The
> extra gold-colored grounding straps on the card are in fact almost the
> only ground pins on the whole connector, because the old PCMCIA ground
> pins have been replaced with signal pins. This is 100% proof of a
> CardBus card. Any adapter that claims to be CardBus will show a series
> of spring grounding pads, as seen in the metal shield in this picture:
>
> http://www.omegacs.net/~omega/ptp/cardbus-slot.png
>
> The specific issue I've been dealing with re: 3.3v PCMCIA adapters is
> that no one makes them. Nominally, the "PC Card 2.2" specification
> requires (afaik) that the slot be able to provide both the normal 5v and
> the lower 3.3v. However, I have yet to find a single card that is not
> actually CardBus anyway that has 3.3v, or is keyed for 3.3v cards. Even
> the PCI1410-based card does not work, even though the PCI1410 itself is
> a 3.3v chip and there's a regulator on the board. The upper-right
> corner of the Ricoh board in fact has a spot for a regulator, if it were
> actually a CardBus adapter.
>
> If you actually need 3.3v, you must make sure that the socket is
> actually keyed for it. The left edge of a PCMCIA/CardBus card (looking
> into the slot) has a key for either 5v or 3.3v. If the protrusion is
> about 1/3 the thickness, it's 5v. If it's 2/3, it's a 3.3v-*only*
> card. Inconveniently enough, the key in the slot is buried deep where
> you can't see it sometimes even with enough light and the right angle.
>
> I even spent a significant amount of time trying to coax any kind of
> real information out of SCM, who makes all sorts of PCMCIA/CardBus
> adapters. They couldn't figure out whether their own product would
> handle 3.3v cards. Even the engineer really had no clue.
>
> The moral of all this is that you cannot trust *anything* any marketing
> materials or vendor tells you about card, even if they claim it's a
> CardBus adapter. Look very carefully at the adapter, see if there's a
> regulator, see if the slots are CardBus or not, and lastly actually try
> it out and see if it will work.
>
> I'm very tempted to create a wiki page listing all known details about
> various adapters...
>
> Erik Walthinsen <omega_at_temple-baptist.com> - System Administrator
> __
> / \ GStreamer - The only way to stream!
> | | M E G A ***** http://gstreamer.net/ *****
> _\ /_
>
>
