Autor: Gregorio Kus (Grego_at_RMnet.IT)
Data: Mon 02 Dec 1996 - 06:06:18 MET
On Mon, 2 Dec 1996 01:28:59 +0100, Andrzej Karpinski wrote:
>hi!
>
>na rynku dostrzeglem ostatnimi czasy karty i/o oparte o 16650 w miejsce 16550
>- ponoc charakteryzuja sie wieksza predkoscia i 32 b boforem zamiast 16. czy
>ktos mial juz cos takiego na blacie i moze cos wiecej napisac?
na blacie nie mialem
ale info z miarodajnego zrodla moge podrzucic
(Raymond Gwinn - autor driverow do OS/2)
16650
The 16650 by Startech seems a good logical step in the evolution of
the 8250 type UARTs. One of the many flavors of the 16650 is a
package that is pin compatible with the 40 pin (DIP) 16550s. This
means that the 16650 can be placed into the same socket as any 16550,
8250, etc.
The 16650 has enhanced the FIFOs to 32 bytes, added on chip flow
control ability, and increases the maximum bit rate to 460800. The on
chip flow control is a really nice feature. On chip flow control
practically eliminates the possibility of missed receive characters.
On chip flow control also means that some devices that require
immediate attention when it sends Xoff, like the Seiko Label Printer,
will work much better in environments like OS/2 and Windows.
The 32 byte FIFOs of the 16650 is an improvement over the 16550, but
is still a disappointment. I feel that 1k, or larger, FIFOs would
have been a much better choice. When I quizzed Startech about this,
they replied that the 32 byte FIFOs were chosen to keep the cost down.
They added that if the 16650 is successful, they plan a 16750 with
larger FIFOs.
The 16650 powers up in a 16550 compatible mode, meaning most software
written for the 16550 should work. I have discovered a serious
incompatibility in the current (Oct 1994) versions of the 16650. All
previous UARTS in the 8250 family will issue only a single transmit
interrupt if no additional data is sent to the UART. The 16650 on the
other hand, will issue continuous transmit interrupts making it
necessary to mask the transmit interrupt. I quizzed Startech about
this and was told that a new revision of the chip would be available
within a few months that would correct this incompatibility.
-- /----------------------------------------------------------------- Gregorio Kus Grego_at_RMnet.it Grego_at_cyberspace.org ROMA, Italy http://www.RMnet.it/~grego Grego_at_FreeNet.hut.fi Anonymous Mail Service - http://free.rmnet.it/~grego/AnonMail.html
To archiwum zostało wygenerowane przez hypermail 2.1.7 : Tue 18 May 2004 - 12:58:17 MET DST