Re: Plus/4 RS232 woes

Den Thu, 6 Sep 2018 20:29:29 +0200 skrev Gerrit Heitsch
<gerrit@laosinh.s.bawue.de>:
> On 09/06/2018 08:20 PM, Mia Magnusson wrote:
> > Den Thu, 6 Sep 2018 19:56:26 +0200 skrev Gerrit Heitsch
> > <gerrit@laosinh.s.bawue.de>:
> >> On 09/06/2018 07:40 PM, Nils Eilers wrote:
> >>> On Wed, 5 Sep 2018 16:39:44 +0200
> >>> Gerrit Heitsch <gerrit@laosinh.s.bawue.de> wrote:
> >>>
> >>>> The 6551 doesn't need capacitors or resistors. All you need is to
> >>>> connect the crystal to the two XTAL pins and you're done.
> >>>
> >>> The W65C51N datasheet shows a schematic at page 19 with a 30 pF
> >>> capacitor from XTALI to GND and a 1 M-Ohm-resistor from XTALI to
> >>> XTALO.
> >>
> >> Well, that's the 65C51... I was talking about the 6551 and its
> >> datasheet doesn't mention anything of the sort.
> >>
> >> And if you look at the schematics of the Plus/4, you'll see that
> >> there is only the crystal.
> > 
> > Not deriving the baud rate clock from the system clock must have
> > added to the cost.
> 
> Especially since back then, they used a rather large crystal:
> 
> https://www.thefuturewas8bit.com/media/sitebuildercontent/sitebuilderpictures/P4before.jpg
> 
> (See on the right side)

Yeah, those were huge back in the days.

> > Of course you'd need a divisor that isn't as nice as 16 to get
> > correct baud rates with a clock that can also be used to generate a
> > NTSC or PAL color signal. If the 6551 addition had been in the
> > plans when TED and CPU were designed for the Plus/4 series, they
> > might have had that buildt in.
> 
> Maybe.. but that would need something like a PLL, neither PAL nor
> NTSC clock will divide easily into 1.8432 MHz.

Yes, that would had required either a PLL or more likely an uncommon
divisor. (Many computers at the time used a clock that was a multiple
of the color carrier, so you had a higher frequency that you could
divide by some odd number to get a good baud rate clock).

> > Btw a "6552" which can take color clock might had been something
> > they could sell anyways.
> 
> They had something like this in the PC... I have seen ISA cards that 
> didn't have a crystal and one of the UARTs was a special one (Winbond 
> 86C451, you can find the datasheet online). It used the 14.318 MHz
> from the mainboard to generate the baud clock for itself and a
> possible second UART on the same card.

Interesting!

The original CGA cards used the system clock too. That's the reason for
the original IBM PC and XT motherboards to have a capacitive trimmer
near the 14.318MHz oscillator.

There is a post on the os2museum blog about a more modern (IIRC PCI, at
least some kind of (S)VGA) graphics card that uses the bus clock rather
than having an oscillator on board. There were issues with it not
working on some motherboards due to the clock signal not being good
enough electrically.

Amiga of course uses it's internal 28.xx MHz oscillator to drive
everything.

-- 
(\_/) Copy the bunny to your mails to help
(O.o) him achieve world domination.
(> <) Come join the dark side.
/_|_\ We have cookies.