On Tue, 07 Mar 2017 23:37:04 +0000, Graham. wrote:
> On Tue, 7 Mar 2017 22:21:58 -0000, "NY" <
m...@privacy.net> wrote:
>
>>"Bill Wright" <
wrights...@f2s.com> wrote in message
>>news:o9n5la$3t0$1...@gioia.aioe.org...
>>> So many things run on 12V DC. Firstly, why has that become the
>>> standard?
>>> Is it all because of the nominal voltage of a car battery?
>>
>>I would imagine so. 12V can easily be made out of six 2V lead-acid cells
>>(for a car) or eight 1.5V batteries (for a portable radio etc).
>>
>>I'm not sure how 5V became the standard for USB power.
>
> TTL logic must figure strongly in the answer.
It sure does! Since the days of the very first hobbyist kit computers in
the mid to late 70s, the logic components have all been sourced from the
famous 7400 TTL series "Logic Family". The very first IBM PC had very
few, if any, dedicated VLSI (or even just LSI) chips being made up almost
exclusively of TTL logic chips to provide the 'logic glue' for the Intel
8088 cpu that IBM's cash starved fledgling PC division chose as a cheap
way to fabricate a 16 bit class desktop computer using cheap off the
shelf 8 bit data bus components.
Although the technology has moved on to cmos based chips, initially a
logic family given the base number 4000 (and still available) and able to
work on supply voltages ranging from a low of 3 to a high of 15, the cmos
versions of the 7400 TTL logic family were characterised to operate at
the standard 5v logic levels to retain compatibility as were the later
dedicated VLSI chips used to replace the original TTL logic glue chips to
reduce fabrication costs of the later motherboards where most of the
costs lay simply in the physical size of said motherboards and the
mounting of dozens of TTL chips where a single VLSI chip could now
suffice.
The 5 volt rail has been a standard supply rail in all versions of the
PC PSU from the earliest PC PSU through the AT and into the still current
ATX class PSUs. The addition of a 3.3 volt rail which was a mobile chip
voltage standard used by notebook (laptop) computers, didn't show up as a
standard rail voltage until the advent of the ATX class of PSU where it
appeared as an extra rail voltage rather than a replacement for the 5
volt rail.
Although modern CPUs have been using low voltage cores for some two
decades now (give or take a year or three - it's definitely been more
than 15 years), there's no point in getting the ATX PSU to provide a
voltage lower than 3.3 in order to directly power cpu cores which are now
typically using voltages as low as 1.2v at currents that can peak above
100 amps.
Such low voltage high current supplies are best generated local to the
cpu socket via programmable high efficiency switching converters drawing
their power from the 12 volt rail which can provide the required voltage
regulation at current demands that can exceed 100A and at voltages that
can be programmed in 5 or 10 mV increments over the range 0.8 to 1.5
volts or thereabouts. A 5% variation of a 1v supply is a mere 50mV.
There's no easy or efficient way to provide such low voltages at such
high power levels directly from the PSU.
The 5 volt standard is unlikely to vanish from the desktop (or even
notebook) PC any time soon since it's such a useful voltage to power USB
kit and internal peripherals. The time of its eventual demise in whatever
new PSU standard that will eventually replace the now venerable ATX
standard, is likely to coincide with the loss of the 3.3 and -12 volt
rails (the -5v rail disappeared from the ATX spec a few years ago now),
reducing the supply rails down to a single 12 volt rail (probably split
into separately monitored groups to allow detection of overloads that
could potentially melt a wire if such protection was solely reliant on
detecting, for example, the 30 odd amp's worth of overload that would be
needed to trip a 360W rated PSU).
Enthusiasts of low power ITX kit have been using this "Single 12 volt
rail PSU" concept for a few years now (and it's a technique that's not
just limited to ITX kit, it can be applied to low and medium power ATX
machines just as effectively).
In this case, special ATX 20 and 24 pin MoBo psu connectors complete
with switching converters to provide the -12v, the 3.3 and the 5 volt
rails powered by the single 12v supply which also directly feeds the 12v
pins allows the machine to be powered from a high efficiency 12v power
brick not unlike a high power laptop charging brick (usually rated
somewhere in the region of 90 to 120 watts output).
Additional switching converter power modules may be needed to power up
peripherals such as SSDs and HDDs, I can't recall whether the ATX adapter
header allowed for the extraction of the 5 and 3.3 volt (and 12 volt) to
power such peripherals or not. In a low power setup, this could be a
useful option otherwise excessive demands on the 3.3 and 5 volt rails
would best be served by additional converter modules (assuming the 12v
power brick can take the additional strain, of course).
An updated single rail PSU scheme would see main boards with power
conversion modules already built in to locally generate any 5 or 3.3 or
-12 (or even -5) volt rails to power their circuitry. Such main boards
could therefore still provide the standard 5 volt power to their USB
ports.
A change to single rail PSUs won't necessarily signal the demise of the
5v standard. There are no clear signs of it disappearing for another
decade or three since it's such a useful low voltage DC voltage source to
power the logic chips inside of most peripheral devices or gadgets with
modest power requirements.
Any such USB connected peripherals with power demands in excess of the
classic 2.5W limit are probably better powered from a dedicated PSU
rather than be allowed to add additional stress on the PC's PSU anyway.
The USB3 power management protocol that allows (IIRC) for up to 30 watts
to be drawn by a peripheral is a feature I find rather troubling. I
suppose if the feature can be limited or disabled in the MoBo setup when
there's an obvious overload risk to the PSU chosen to power the system,
it would be ok otherwise I'd be concerned that plugging in a 'greedy' usb3
peripheral could end up triggering a catastrophic power outage event.
--
Johnny B Good