| Going into all that (rather high-level) detail allows us to figure out the pitfalls beforehand (making it easier to set goals in retrospect).
I don't remember who said it, but predictions about the past are rather easy, making predictions about the future is what is hard...
Edgar
On Jul 9, 2010, at 12:30 AM, LEI / Rene Koch wrote:
Dear Bob,
I guess we are all very excited and
enthusiastic about this thing, making us probably going into too much detail
already.
Cheers,
Rene
Gentlemen,
The first set of choices
will need to be the overall goals for the system to meet a 10+ year life
expectancy in anticipation of changing conditions. After the overall
goals are elaborated, argued and voted on, the detailed design proposals
can be examined against that criteria.
Bob
From: upamd@xxxxxxxx [mailto:upamd@xxxxxxxx]
On Behalf Of LEI / Rene Koch
Sent: Thursday, July 08, 2010 7:17 PM
To: Edgar Brown; Per Hassel Sørensen
Cc: upamd@xxxxxxxx
Subject: Re: SV: Potential Goals of the group for discussion
Dear Edgar and Per,
Very good, I think we're getting somewhere.
I see a problem though with kelvin sensing, when it will be based on the
load's feedback setting. It will be very hard to get a stable feedback
loop if the sensing is in the active feedback loop.
Therefore I think we should have an active feedback loop (1st loop) that
satisfies stability needs and a second loop that could be driven by the
load.
The first loop should be based on the contacts of the power supply, which
creates a stable output and then the second loop, by the kelvin sensing
or communication as proposed. This way we establish a very simple high/low
signal on that sensing that just requests more or less voltage/power from
the power supply, regardless how high that is; the response to that signal
however will have to be slow (<1kHz) in order not to interfere with
the 1st feedback loop.
Edgar raised another important point, I think, and that is single fault
conditions.
I think the connector should be designed in such a way that the communications/feedback
link makes contact before the actual power. This practically means that
it would be almost impossible to have a feedback signal, requesting power
while there is a short between any two connections. The feedback will have
a reasonably high impedance, compared to that of the power lines, which
looks like a short initially due to the presence of a discharged large
electrolytic capacitance.
Finally I think we are actually moving away from the possibility of having
communication over the DC power lines. So we would need at least a 3 wire
connection.
Best regards,
Rene Koch
Per Hassel,
I think we are _mostly_ in violent agreement. I now think that an analog
modality would be feasible and cost-effective, as long as it is a simple/feature-less
implementation. A good example is kelvin-sensing. As long as there is trickle
power, the adapter can sense the presence of a load and slowly increase
the voltage until the sensing lead reaches a specific voltage range. That
way the adapter does not need any further information from the load. The
adapter only has to track the feedback signal adjusting its output to remain
in the acceptable range. Alternatively, the current or impedance on the
'kelvin' lead could indicate the "power-class" of the load.
Note that any analog DC alternative (that I can think of) has the problem
of guaranteeing that the adapter does not become a hazard when two or more
random terminals in the connector are shorted together.
Regarding power, I agree on the approach of making the marking presence
and location a mandatory part of the standard. This can be complemented
with LED colors and patterns. A user could easily become accustomed to
an adapter's fast blinking red LED, just beside the power sticker, indicating
that it cannot provide enough power for the connected device. (Or a steady
yellow indicating that the user could be better served by a higher-power
adapter.)
I do see another conflict that has been brewing here:
- two wire adapters/plugs, vs. three or four wire plugs.
Edgar
On Jul 8, 2010, at 7:10 PM, Per Hassel Sørensen wrote:
Edgar,
I agree that it is detrimental if we have some various adapter classes,
but:
I believe it is almost impossible to not have incompatible combinations
of adapters and devices due to power requirements. The biggest cost factor
for a mains powered adapter is the power rating. It is also a weight and
size factor affecting portability and usage. I don't think people will
accept cost, size or weight of a 130W UPADM adapter to power a device like
a portable radio or charge a micro RC model helicopter. I think this
is not a big issue if handled properly.
With a good UPADM standard, it should be the possible to use any UPADM
adapter as long as it supplies the minimum power needed for the device.
Everybody (Almost) will understand that adapters need to come in
different sizes. This is not difficult as long as we keep adapter power
rating the only variable the user has to think of. It could be part of
the standard that power rating of adapter and consumer is easy to locate,
is concise and has a common labeling. And imagine a sales situation where
someone walks into a store in need for an adapter: The 130W 'brick' will
always work, (un)fortunately (for the store), it is the most expensive
one.
To save cost on the consumer side the UPAMD could support a basic analog
mode. Imagine having two separate communication wires in the UPAMD cable.
When in analog mode they are used for sensing a simple analog network in
the consumer. Certain properties (resistance, zener diode voltage etc.)
is detected by the adapter to select correct output voltage, current limits
etc. Further, as a transition solution, a UPAMD adapter cable with
built-in analog network could be built for a very low cost to be used with
legacy equipment lacking built-in UPAMD support. This way existing equipment
can use the new adapter.
For more advanced consumers the UPAMD would use the two communication wires
for duplex communication overlayed (or time multiplexed) on analog voltage
feedback (aka 'Kelvin clips') for reduced cable cost, better regulation
and other benefits as stated in my previous mail.
This way a full digital communication is present in every UPAMD adapter
from the beginning, without requiring a major revision for purely analog
devices. And every UPAMD adapter will work with all devices, as long as
the power requirements are met as described above.
Per Hassel
Fra: upamd@xxxxxxxx
på vegne av Edgar Brown
Sendt: to 08.07.2010 17:43
Til: LEI / Rene Koch
Kopi: upamd@xxxxxxxx
Emne: Re: Potential Goals of the group for discussion
Rene,
I see two conflicting goals in removing communications from the requirements:
(1) universality (one adapter fits all) and (2) low-cost (a.k.a. analog)
simplicity. Although a properly designed analog solution can nearly always
be the lowest _manufacturing_ cost one. It would probably be the most complicated
in terms of flexibility. As an analog and digital designer, I am always
confronted on where each technology fits best. I can clearly remember the
cases in which, after a grueling analog design has been put together, I
have wished to have gone digital instead. While the converse is rarely
true.
These are the issues as I see them:
- A single plug fits all: If a UPAMD plug is present in an adapter and
a device, we want the consumer to correctly assume that the two can be
plugged together. Having a class of adapters that would not power some
subset of devices would be very detrimental to this goal (this also applies
to power levels, B.T.W.). This is also one of the strongest arguments against
the highly non-standard and irritating family of barrel connectors.
- Providing enough flexibility through analog means, although possible,
would _probably_ complicate the circuit enough that a comparable digital
solution would end up being cheaper in the short (and long) term. (And
clearly cheaper from a design stand point.)
- A properly designed (and adopted) standard would create industry demand
for the right low-cost semiconductor solutions. As a sample, take a look
at the prevalence of ARM cores in TI's family of PMBus PWM power controllers.
An ARM core would be overkill for our intended application, but it is not
out of the question in the mid-term. A power supply that has to satisfy
multiple requirements, such as power-factor correction, low-power consumption,
stand-by modes, etc., presents enough functionality that an all-encompassing
integrated digital solution would probably be preferable.
And while writing this I thought of a way to accomplish some of the conflicting
analog requirements. A simple feedback mechanism, in which the device requests
power from the UPAMD through a simple lower/higher analog signal (essentially
a slow remote regulation loop) could satisfy some of these conflicting
requirements. An AC signal (e.g., AM-modulated voltage feedback) can close
the device-adapter loop. Only requiring simple diode/resistor/capacitor
circuitry to extract the return feedback signal. And this would not even
require extra wires in the connector. This choice would somewhat complicate
any communications between adapter and device, but it still can be done
with decades-old modem technology.
Regarding 'greening' technologies. Although it is true that large equipment
is the easiest target for conservation efforts, do not underestimate the
amount of power being wasted on tens of unused but plugged adapters in
each home and thousands in each office building. Also don't underestimate
that any progress towards these types of standard would also ease the development
of equivalent standards in directly-plugged AC devices. Any progress to
be made in removing power through greener adapters I believe to be worthwhile.
Especially in a standard that we hope would be still useful a few decades
into the future.
Edgar Brown
On Jul 7, 2010, at 4:51 AM, LEI / Rene Koch wrote:
I think I sense that extended communication is wishful; I agree that would
be a nice addition for the future.
However, I still have cost concerns to this when we absolutely need this
communication in order to get some power from the supply.
Therefore I would vote for a solution that would not restrain communication
in any way, but that we also find a way for simple and low-cost power adapters
to be used, without communication ability.
Adding smart-grid integration would make sense, although this, at this
time, is more focussed on big power users like aircons, water heaters and
other luxury type of products. When the device is smart enough and goes
to a power down or standby mode whenever possible, there is no need for
smart-grid to intervene, I do not think that EISA 2007 will go so far that
even laptop computers printers and networks will have to go down when actually
being used; I think it will remain at aircons, heaters, waterheaters, (dish)washing
machines and (public)lighting; most of this type of equipment consumes
a lot more than what the adapter in question can provide.
It is not difficult anymore to get below 100mW in stand-by, it is just
a matter of price at this time, in a couple of years I foresee most adapters
to be able to draw less than 50mW in no-load.
This is a great achievement from most power supply manufacturers and power/PWM
IC design houses. Destroying that by adding another 100mW or more, just
to listen or communicate in order to go down to an overall power down for
a period in time does not make sense to me.
Would the overall energy usage really bring us below 50mW? If so, is this
enough to justify the extra waste that we add? (controller, manufacturing
of that controller, distribution of that controller, extra wires for the
communication needs).
Ecodesign in Europe already announced to be looking at the total picture,
including design, manufacturing, testing and distribution involved in putting
any product to market. (total green concept or cradle-to-grave)
As I mentioned in one of my earlier comments, having a power supply that
could also perform as a bridge to Power Line Communication or Smart-grid
could be a great addition and would make sense in the future when the different
types of network communication will get more and more integrated into each
other to increase and secure the total available bandwidth.
Best regards,
Rene Koch
Some additional suggestions regarding "green" capabilities.
A) Support for Load Shedding / Demand Response / Dynamic Demand
functionality.
The power adapter-to-device communications should support primitives
for demand shaping, incorporating mains-side interface/definitions to
support the applicable protocols and standards. To reduce cost, and
avoid restrictions over the multiple mains standards, the specific
load-shedding support should remain optional.
Use cases:
(1) The utility generates a load-shedding request that gets coupled
into the home network. Compatible UPAMDs (the ones implementing a load-
shedding protocol mains-side) send the request across to the connected
device. The connected device responds by modifying its power
requirements or removing them altogether. (The user could set the
power-saving settings similarly to existing battery-conservation
settings in laptops).
(2) The user requests device charging only during off-peak (e.g., low-
cost) utility periods. A home-based controller instructs compatible
UPAMDs to provide power only during those periods (note that this
might not require intervention from the device side, i.e., forced
power-shedding.)
(3) The user connects an specific UPAMD (to be used with a different
device) to a computer and configures its power-saving capabilities
through a local program or web page. A real-time clock inside the
UPAMD indicates at what times it will provide power to the attached
device.
Possible applicable standards: IEEE P2030, IEEE P1547.8, ISO/IEC
14543-3 (KNX), IEEE 1815, IEEE P1901, IEC 61334, ITU G.9972.
The most relevant standards seem to be the ones from the KNX
association.
Industry standards: HomePlug Command & Control, HomePlug GreenPhy,
http://www.prime-alliance.org/
Related technologies:
load shedding lighting ballasts: http://www.lrc.rpi.edu/researchAreas/pdf/LoadShedBrochureNYS.pdf
,
home control: http://www.Insteonsmartgrid.com/about-us.html
)
B) Sleep capability requirements.
To reduce power consumption, the mains-side of the UPAMD should
disconnect itself from mains power for pre-defined periods of time
(e.g., 5 sec, 10 sec or even longer). When not in use, the excess
capacitive power reserve should be able to power the UPAMD electronics
for a considerable period of time. The mains-side switcher can be
turned-off (for several seconds) in those conditions providing near-
zero power consumption. (A slow-blinking LED, for power reasons, can
indicate this condition.) This would make it much easier to satisfy a
<100mW standby mode.
Alternative 1: If trickle power is provided to the powered device, the
UPAMD only powers-up to keep trickle-power regulation.
Alternative 2: The UPAMD only provides trickle-power to the connected
device for a brief period every n seconds.
Alternative 3: The sleep period becomes longer if the device has not
been used for an extended period of time (a reserve super cap can
provide sleep power under such conditions).
Note that EMC issues can arise due to this start/stop condition, and
additional restrictions (e.g., a soft start-up envelope) might be
required to avoid these.
Cost considerations: given that some intelligence is to be required to
provide the communications layer, the additional controls required to
implement this capability would be minimal. Alternative
implementations (e.g., timers, or voltage comparators) can allow for
further cost reductions.
EB
|