Difference between IDE and SCSI ??

Les hlhowell at pacbell.net
Mon Feb 4 16:59:33 UTC 2008

On Mon, 2008-02-04 at 17:00 +0900, John Summerfield wrote:
> Alan Cox wrote:
> > On Sun, 03 Feb 2008 13:54:26 +1030
> > Tim <ignored_mailbox at yahoo.com.au> wrote:
> > 
> >> On Sat, 2008-02-02 at 11:57 -0800, Les wrote:
> >>> SCSI is a serial system, or at least it can be.
> >> Pardon?  Usually, when one has a data bus for several parallel data
> >> lines at once, one refers to it as parallel.
> >>
> >> Serial - one data line, that sends bits sequentially.
> >> Parallel - several data lines, that send bits simultaneously.
> > 
> > Of course its never that simple. Most "serial" busses are actually
> > multiple wires which send information in parallel..
> More likely
> txdata
> rxdata
> Maybe a pair for the other direction (eg RS0232C)
> Ground
> For a modern cable that has fair throughput, view your ethernet cable:
> http://www.ertyu.org/steven_nikkel/ethernetcables.html
Actually Ethernet cable is only modest throughput.  Twisted pair limits
frequency due to included inductance.  The bandwidth in bits is due to
the encoding which is moderately complex.  And as the encoding envelope
gets more and more complex and higher in bandwidth, the relevant
transmission effects, radiation loss, envelope delay, phase errors
introduced by lumped changes in impedance and so forth, become limiting
factors in transmission distance.  That is one of the reasons for the
change to PoE which is to allow the use of modular repeaters inserted
into the line.  

Encoding is a somewhat exciting area of research these days, with things
like PRML (Partial Response Maximum Liklihood) and RLE (Runlength
Encoding), giving way to analog Quadradure Amplitude Modulation (QAM) in
one of several forms.  WLAN and Cellphones are using Orthogonal
Frequency Division Multiplex (OFDM) and within the OFDM envelope are
several standards of BiPhase, and QAM modulation.  Modern DSP techniques
allow envelope recovery, phase leveling (wideband signals tend to have a
phase creep with distance, even when sent wirelessly), and frequency
tracking using pilot frequencies within the envelope.

In addition, digital frequency synthesizers coupled with some fancy DSP
allow phase shift control at low jitter points on the signal (zero
crossing and peaks) which lessen the intermodulation products,
minimizing the inherent noise, and maximizing the signal to noise ratio
in a controlled environment like cable.  Further improvements like
coherent signal tracking allow predictive reception, lowering the
bandwidth of the receiver, which in effect reduces the reception of
radiated noise yeilding signals that can actually be plucked from below
the open band noise floor.  It is an arcane area of study, but
bandwidth, transmission means and noise studies for modern signal
processing is really growing by leaps and bounds.  Lots of things that
have been known to theorists for years are now finding practical and
useful application. 

	The real cutting edge is in light manipulation, and means of applying
these DSP techniques to light, yielding greater transmission distance
and lots more bandwidth than can ever be obtained by RF.

Sorry, I get carried away a bit on this stuff.

Les H

More information about the fedora-list mailing list