Subject: Re: HSTL standards
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Wed, 25 Dec 2002 02:37:29 GMT
NNTP-Posting-Date: Tue, 24 Dec 2002 18:37:29 PST
Organization: EarthLink Inc. -- http://www.EarthLink.net
It's interesting to note that the higher-class levels of HSTL do not
recommend that their VREFs and VTTs be at VCCO/2. Apparently this is due to
the non-symmetrical drive characteristics of the real-world HSTL drivers
(any insight into this, Austin?).
I've used HyperLynx to simulate a QDR ram with Virtex-II. If the VTTs are
not set as specified by the HSTL spec, then the simulation shows a distorted
duty cycle for the data.
Life used to be so much simpler when everything was at 5V. Sigh...
"Austin Lesea" wrote in message
> Starting on page 178
> You will find the topologies of the four interfaces. The class I and
> III are intended for unidirectional high speed interfaces, wheras classes
> II and IV are intended for bidirectional high speed interfaces (mirror
> symmetric so they would support a tristate IO for both TX and RX).
> Classes I and II are weaker, and intended for shorter runs, and III and IV
> are stronger and are intended for longer (more heavily loaded) runs.
> Both 1.5V and 1.8 volt versions of all four classes of HSTL exist, and the
> 1.8 volt versions are just a bit faster than the 1.5 volt versions in most
> multi-purpose IOs (ie programmable IOs). 1.8 volts came about when many
> ASIC implementations just didn't work at the intended frequency, so the
> voltage was increased to make it work.
> All use a separate Vref supply of 1/2 Vcco at the receiver which is a high
> speed comparator.
> All are parallel terminated standards that are suitable for multi-drop
> (in the unidirectional case), and have excellent signal integrity
> The lower voltage swing leads to less cross talk, and less EMI, but not
> less ground bounce, as the currents are about as large as other strong IO
> The disadvantage is that external resistors are required (unless you use
> internal termination feature, such as the Virtex II and II Pro DCI), and
> that such parallel termination, internal or external burns power.
> Many people use HSTL without the resistors on very short runs, but to do
> violates the standard, and one must simulate and test to be sure that you
> will be safe, and the interface will work as intended.
> John McMiller wrote:
> > Hi,
> > What is the main differences between the variuos HSTL I/O technologies:
> > HSTL-I
> > HSTL-II
> > HSTL-III
> > HSTL-IV
> > ?
> > John