Re: [802.3EEESG] 10BASE-T question
Hi,
10BaseT not only have voltage problem (5V) but also have power consumption
issue. Based on existing measurement, its operating power consumption is
higher than 100BaseT even though it is only sending link pulses. So, 10BaseT
is not suitable to become a "speed downgrade" role in order to save power.
IMHO.
Please bear in mind that the power delivered in the TP line is very small
(5dBm or so) compared with the power consumed by transmitter driver and
receiver (hundreds of mw).
I will second the concept to develop a new electrical idle mode (standby
mode) if it is not toward a NEW phy mode (I don't know how yet) which system
vendors may consider as a major disadvantage and against it. Under this new
idle mode, any speed of PHY (100, 1000, 10G) can be downgraded directly to
this mode to save maximum power.
Entering this idle mode is straightforward and can be conducted by control
packet (frame based). During the idle state no signal or physical link pulse
is established. (The link state is still viewed as ON.) The exit of this
mode (or wake-up mode) is therefore somewhat tricky. A low power beacon (or
chirp signal as mentioned by Mike) can be issued to signal the request of
restoration of speed. A signal based handshake needs to be completed to
allow the link up. Whether or not this beacon needs to encode information
exchanged by both parties are left for further study. However, it is also
possible the physical link has been broken and reconnected during idle mode.
In this case a normal auto-negotiation will be carried on since link pulses
will be sent instead.
Mike's idea is a little bit different. In his text, the sending and
receiving of low power chirp signal will be maintained through the entire
idle state. The advantage of this idea is that the phy will be aware of any
physical disconnection of link immediately at the expense of some sustaining
power consumption.
My conclusions so far are:
1. 10BaseT is really not useful and unimportant in EEE application.
2. A new electrical idle mode (0 BaseT) can save maximum power at a risk to
develop a new phy mode or interface specification. The existing state
machine for each phy may need to modify to adopt this mode.
3. Without introduction of the idle mode, phy can still freely switch
between any two modes of three major modes (100, 1000, and 10G) based on
traffic measurement.
4. The save of DSP coefficients will help the startup time. However, the
timing or phase information may be less helpful after a speed turnaround.
Those information are implementation dependent and should be not exchanged
and be excluded from the standard.
Best Regards,
-Joseph Chou
-----Original Message-----
From: mike mcconnell [mailto:mike.mcconnell@KEYEYE.NET]
Sent: Wednesday, March 28, 2007 3:55 PM
To: STDS-802-3-EEE@listserv.ieee.org
Subject: Re: [802.3EEESG] 10BASE-T question
I believe the conversation about changing voltages was stimulated when
10BASE-T was mentioned as candidate for the minimum operating speed.
Legacy 10BASE-T voltages are indeed a problem for current silicon
processes. I think the real question is not the level of difficulty to
modify the standard but rather what is the motivation to do that and are
there alternatives.
The advantage of 10BASE-T was that during idle periods it only sends
link pulses. One suggestion might be to exclude 10BASE-T due to voltages
and adopt a 0M mode which exchanges a modified link pulse. Indeed I
would suggest a "chirp pulse." The idea is to send a very low energy
signal that would not be confused with link pulse but would permit the
collection of enough information about the channel so that the startup
time for 100M or even potentially a 1G PHY could be reduced. Of course
timing and phase information is necessary too but maybe that acquisition
time could be reduced too with a cleverly design chirp. I'm thinking
radar link pulses - low energy, broad spectrum, spread over time.
I don't know if this is technically feasible or what kind of power it
would take to implement it but it seems to be worth a look and
completely side steps the questions about modifying any legacy modes
beyond the creation of alternative entry points in the startup state
machines as already discussed.
Mike
-----Original Message-----
From: Mike Bennett [mailto:mjbennett@LBL.GOV]
Sent: Wednesday, March 28, 2007 1:46 PM
To: STDS-802-3-EEE@listserv.ieee.org
Subject: [802.3EEESG] 10BASE-T question
Folks,
For those of you who were able to attend the March meeting, you may
recall we had a discussion on 10BASE-T (in the context of having a low
energy state mode) and what we might change to specify this, which
included possibly changing the output voltage. Concern was raised that
the work required to specify a new output voltage for 10BASE-T would far
outweigh the benefit. Additionally, there was a question regarding the
use of 100BASE-TX instead of doing anything with 10BASE-T. Would
someone please explain just how much work it would be to change 10BASE-T
and what the benefit would be compared to using 10BASE-T with the
originally specified voltage or 100BASE-TX for a low energy (aka
"0BASE-T" or "sleep") state?
Thanks,
Mike