RE: Auxiliary power support for PCI

[Tony Goodfellow <tonygd@earthlink.net>]

Thanks for the ATX web site.

Re 100mA. Try keeping a 100Mb/s Ethernet Adapter alive and leave enough power for a modem adapter!!!!

-----Original Message-----
From:	B. P. Lame' [SMTP:blame@prolog.com]
Sent:	Sunday, May 18, 1997 9:53 PM
To:	'Tony Goodfellow'; 'PCISIGList'
Subject:	RE: Auxiliary power support for PCI

Uh, isn't this what I said the first time?  I wouldn't expect widespread use (as compared to other form-factors) of  ATX for at least another year, but do what you want.  Regarding the 5Vsb 100mA limit on your supply, if you design with power conservation in mind, you should have no trouble powering all your system wake-up logic with much, much less than 100mA (on a modem, you need only to power the ring detector ckt).  Unfortunately, unless you are doing an integrated motherboard design, ATX will not solve your problem very well.  In my opinion, you might as well forget about an ECN to change a PCI reserved pin to 5Vsb because of the logistics of it's use and availability - not just for PCI in the consumer market, but PCI in its several other environments and form-factors as well.

The ATX spec is not directly a part of the MS On-Now initiative, but is an attempt (by the Intel-MS alliance?) to create a more integrated, generic PC form-factor, that is slightly better suited for mass commercial consumption.  It actually retains many of the integrated qualities that Apple designed into the Mac series, but creates an "open standard" available for use by any third party supplier of any component (thus helping to ensure it's real life as a defacto standard nurtured by Intel).

See:  http://www.teleport.com/~atx/
 
--
Opinions are mine and not necessarily those of the company I keep.
Brooks Lame'
Engineer, Pro-Log Corp.

On Sunday, 18 May, 1997 14:17, Tony Goodfellow[SMTP:tonygd@earthlink.net] wrote:
> There is a solution that can and does work. The solution is the use of ATX power supplies. The ones we have seen are form/fit the same as the standard "PS-2" style silver boxes. They have an additional connector that provides 5Vsb (5V standby). In addition they have a PWRON line. The PWRON line is used to power-on (and off) the main supply. This means that the whole system is powered down except those items powered by 5Vsb. A device such as a modem can be powered by 5Vsb, activate PWRON and bring the whole system to life. 
> 
> There are several PSU suppliers offering these supplies but most are at the "sampling" stage. I believe ASTEC was the first but they only provide 100mA of power. This is just not enough for many applications. Other supplies are now specifying 700-800 mA. 
> 
> There is a connector specified but not much has been said about add-in adapters. It seems that most of the work has been done by m-board oriented folks. 
> 
> The ATX psu, is I believe, part of the MS On-Now initiative, as is the provision of "ring-indicator" as a means of waking up a "snoozing" PC. I have not seen any specs on that connector etc..
> 
> If anyone has more info, or where to find it please post it.
> 
> Thanks
> 
> -----Original Message-----
> From:	B. P. Lame' [SMTP:blame@prolog.com]
> Sent:	Thursday, May 01, 1997 11:47 AM
> To:	Mailing List Recipients
> Cc:	'PCISIGList'
> Subject:	RE: Auxiliary power support
> 
> On Tuesday, 29 April, 1997 18:29, Eric Rehm[SMTP:eric@equator.com] wrote:
> > Hi.  Since I'm interested in this and no one has responded, I'll
> > post this again.  Any comments on how a modem or multifunction
> > add-in card w/modem is supposed to handle this?
> > 
> > /eric rehm
> > eric@equator.com
> > 206.812.1290 x213
> > 
> > ----------
> > From: 	Tad Ball
> > Sent: 	Monday, April 14, 1997 11:19 AM
> > To: 	Mailing List Recipients
> > Subject: 	Auxiliary power support
> > 
> > Hi,
> > 
> > We're interested in developing a power-managed PCI device that is
> > capable of generating wake events from the D3cold state, and I'm
> > concerned about the support for auxiliary power in the 1.0 rev of the
> > power management spec.  Specifically, I'm concerned that there is no
> > standard way for add-in cards to get trickle power from the system. 
> > Near as I can tell, if you are not a motherboard device you have to have
> > some sort of external power source.  This seems bulky, cumbersone, and
> > rather user-unfriendly to me.
> > 
> > True power-managed buses like PCMCIA and CardBus have had support for
> > these sorts of devices for a long time, and I'm disappointed that the
> > PCI Power Management spec doesn't have similar capabilities.  What's the
> > rationale for omitting standardized trickle power?  It certainly seems
> > like any add-in that wants to generate PME to wake a sleeping system is
> > going to need it (functions like Modems, Network adapters, etc).
> > 
> > What do other people think of this?
> 
> The power management spec does suggest some ways to deal with this.  An on-card rechargeable supply would probably be the cleanest.  But sure, a PCI connector RESERVED pin could be defined for 5v Soft-Power to provide some limited amount of power while the main supply is off.  However, as of yet, the ATX power supply is the only common spec supply that supports 5v Soft-Power and I'd guess > 95% of all PCI mother boards/backplanes sold today still use PS/2 type supplys, which have no support for Soft-Power.  Since it's not very practical to _require_ Soft-Power support in all PCI systems, your card could not rely on its availability and you'd probably need some config jumper for that.  And lastly, how should 5v Soft-Power be handled in hot-swapping?
> 
> -- 
> Brooks Lame'
> Engineer
> Pro-Log Corporation
> blame@prolog.com
> 408.646.3631

���