Re: [ptp] Re: Corrections re: PCMCIA/PC Card/CardBus/PCI/etc.


From: Erik Walthinsen (omega_at_temple-baptist.com)
Date: 2002-03-20 04:54:08 UTC



On Tue, 2002-03-19 at 20:13, Wim Lewis wrote:

> - Synchrotech PCI -> PCMCIA/CardBus adaptor: [1]. Specifically claims
> that it supports 3.3v operation. Web site imples, but does not state,
> that it's based on the Vadem VG-469, [2] which is 82365-compatible and
> therefore (I hope?) ought to work with anyone's favorite open source OS.
According to that page, the board is based on the TI PCI1410 or PCI1420, depending on slot count. It says it will handle both 3.3v and 5v cards, which I can't really verify visually, mostly because it's not possible to from an image. If the page is fully correct, that board would do the trick.

The VG-469 is an ISA-based 365-compatible controller, which is definitely not CardBus capable, and can't be put on the PCI bus. Not sure where you found that mentioned.

> - Amtron "PCMCIA & CardBus PC Card Drive" [3]. Again, specifically
> claims 3.3v operation. Based on the Ricoh R5C476 II, [4] also
> 82365-compatible, and mentions that it works with Linux (and gives
> specific pcmcia-cs versions to use!). $75/$99 from amtron.com (1/2
> slots).

The R5c475 is a PCI chip, and so it will do CardBus, assuming the board is wired for it. You can wire up chips like that (the Ricoh, the TI PCI1410, etc.) for either CardBus or PC Card. Again, the only way to be sure is if the slot actually shows evidence of being an actual CardBus slot. One hopes they wouldn't spent the extra few cents on a CardBus slot if the adapter didn't support it.

The picture on the website doesn't show any CardBus-isms, but it entirely possible they're hidden under the ejector mechanism. Problem is that it looks very much like the R5c475 board that D-Link ships with the DWL-650, which is not CardBus or 3.3v capable. There are significant differences though, such as the piezo element where the voltage regulator would be. On visual inspection alone it's a toss-up.

More critically, I've learned not to trust *anything* the vendor or reseller's marketing materials say, because they turn out to be wrong or just hopelessly confused about 2/3 of the time, from my experience. I am happy to say that this seems to be primarily a phenomenon of the PCMCIA/CardBus adapter industry. Not that that helps.

> - trendnet TEW-PCIA: [5] $44 at needpcparts.com[9] (1 slot); $46 at
> neticamall.com [10]
> (this one doesn't say up front it does CardBus, but buried in the
> mfr's pages was something that made me think that it does. I'm curious
> about this one.)

Definitely not a CardBus or even 3.3v-capable adapter. The slot lacks any kind of CardBus-isms, and note the big silver square in the upper-right corner of the card where the 3.3v voltage regulator is supposed to go. This is in fact exactly the same board I mentioned above as shipping with the DWL-650, part-for-part, though it may be labeled differently for TrendNet. Standard Ricoh reference design, apparently.

> - actiontec PC750: [6]
> (I don't know why this is on my list --- it doesn't specify 3.3v
> or CardBus in its specs --- but it is the only one that puts the card
> slots in a drive bay instead of a rear panel, which could be handy.
> Based on the TI PCI-1225 [7], which *can* do 3.3v and CardBus...)
Very definitely not CardBus, because CardBus is PCI, and PCI can't go more than about 6" before the signal is totally destroyed. The 18" cable means that it is PCMCIA without any doubt.

The PCI1225 actually has nothing to do with whether a card supplies 3.3v to the socket or not, just as with the Ricoh or PCI1410 chips. Look at page 25 of the datasheet (http://www-s.ti.com/sc/ds/pci1225.pdf), where it shows the TPS2206. That's the chip that handles switching of power between the cards. There are also chips that can only handle 5v and maybe 12v.

The main feature of the PCI1225 as relevant to 3.3v is that the I/O ports are 3.3v-tolerant, else it wouldn't be able to communicate with the cards. The card and controller-chip supply voltages are totally separate issues. The Ricoh chip operates on 5v, but the TI chips use 3.3v. For that reason, the presence of a 3.3v regulator on the board doesn't necessarily mean the cards supports 3.3v either...

Sorry to have to shoot down a lot of your suggestions, but I've been at this for a while and have learned (as you can see) how many pitfalls there are in finding these stupid things.

We did manage to find some cheap ($35) supposed CardBus adapters (and the pictures indicate they really are), but there are only 10 and they're already all spoken for.... ;-| We expect to get the first one in a few days, so we can test it out before committing to getting the other 9.

      Erik Walthinsen <omega_at_temple-baptist.com> - System Administrator
        __
       /  \                GStreamer - The only way to stream!
      |    | M E G A        ***** http://gstreamer.net/ *****
      _\  /_


This archive was generated by hypermail 2.1.4.