Re: Voting for classification method
I would like to cast a vote for #1 with #2 as a second choice.
Selection 1 sets a CLASSIFICATION current at the PD. If this is done
in a small simple integrated solution that also contains the other
misc. circuits (like the isolation switch control) the current
accuracy could be quite good (~2%).
This allows additional classification levels without overlap. Also as was
already mentioned by Brian Lynch, the PD quiescent current could be part
of the classification current thus eliminating current offsets.
The same resistor that sets the CLASSIFICATION current could be used to
set the max PD current so that the PD would not draw more than advertised.
At the PSE, the CLASSIFICATION currents would be much higher than the
DISCOVERY level. The higher signal to noise level would improve
measurement accuracy removing the need for a two point measurement and
reducing the noise filter time constants. Also the PSE would just have to
set a single terminal voltage during CLASSIFICATION.
_________________________________________________________________
Barry Male
Senior Principle IC Designer Internet: barry_male@xxxxxx
Power Management Products NH Phone: 603.429.8705
Texas Instruments CT Phone: 860.844.8183
7 Continental Blvd. NH FAX : 603.424.3460
Merrimack, NH 03054-0399 USA CT FAX : 860.844.8806
On Mon, 2 Apr 2001, Schwartz, Peter wrote:
>
> I'll support #2. Thanks for your work on this, Don.
>
> Peter Schwartz
> Applications Engineer
> Micrel Semiconductor
> Phone: 408.435.2460
> FAX: 408.456.0490
> peter.schwartz@xxxxxxxxxx
>
>
> > From: Stewart, Donald S (Don) [SMTP:dsstewart@xxxxxxxxx]
> <mailto:[SMTP:dsstewart@xxxxxxxxx]>
> > Sent: March 29, 2001 7:19 AM
> > To: IEEE Detection Ad Hoc (E-mail)
> > Cc: Stewart, Donald S (Don)
> > Subject: Beginning to summarize the viable candidates for
> detection
> > with classification
> >
> > Detection Ad Hoc,
> >
> > Before we take any multi-votes or votes on possibilities for
> detection
> > with classification, I want to give people a chance to tune-up the
> list we
> > use. Please see the list below. Any comments are welcomed, but
> > particularly if a category is missing or if I am seeing this
> differently
> > than many of you are. I realize I have included some candidates
> that have
> > been unpopular. But I see no harm in including them in voting.
> >
> > Thanks
> > Don
> > ____________________
> >
> > Candidate approaches to detection with classification
> >
> > 1) PD: exhibits 25 K ohm slope for 2.8 to 10 volts and exhibits
> one of a
> > few constant current levels for ~12 to 30v
> > PSE: detects slope with equivalent of two measurements and
> constant
> > current with at least one measurement
> >
> > 2) PD: exhibits 25 K ohm slope for 2.8 to 10 volts and exhibits
> one of a
> > few different resistance slopes for ~12 to 30v
> >
> > PSE: detects first slope with equivalent of two mesurements and
> second
> > slope with equivalent of two measurements
> >
> > (3) Have PD exhibit maximum current draw shortly after detection
> (infinite
> > power classes)
> >
> > (4) Combination of 1, 2. [Similar to one we voted on in Hilton
> Head. I
> > propose we not vote on something as vague as this. Could someone
> propose a
> > specific behavior they champion if they really want this voted?]
> >
> > (5) Do it with the PHY
> >
> > (6) Any others?
> >
> >
> >
> > _________________________________________________
> > Donald (Don) S. Stewart Phone: 732-817-5495
> > Avaya Inc. FAX: 732-817-4666
>
> > Cross_Product Architecture
> > 101 Crawfords Corner Road
> > Holmdel, NJ
>