[sdiy] New to list - and DSP development

[Magnus Danielson]

From: René Schmitz <uzs159 at uni-bonn.de>
Subject: Re: [sdiy] New to list - and DSP development
Date: Sun, 12 Dec 2004 22:57:07 +0100
Message-ID: <41BCBEB3.7030403 at uni-bonn.de>

> 
> Hi Magnus, Rainer et al.,
> 
> > 1,8432 MHz = 96 x 19,2 kHz
> 
> Hmm, more like 115.2kHz * 16, higest rate generally used with RS232, and 
>   the factor 16 because the UARTs use to oversample at that rate.

Actually, back in the days, 19200 Baud was the highest noted. 1,8432 MHz was a
suitable frequency to make crystals from. 115,2 kBd came from the fact that
the standard PC design allowed it, so it became the "higest rate" because that
was when you set the 16450 UART to /1 instead of /6. The /16 is indeed the
oversampling ratio, used in the receiver to establish a number of sampling
points (odd number) from which a majority decission is made.

> > When IBM created the IBM PC they did these choices. I guess they had the
> > intention that graphical cards would easilly be able to be simplified by
> > already operate synchronous with the NTSC colour-burst. I have only seen one
> > graphical card able to output TV-signal, and then only in greyscale. Actually,
> > that card was sitting in the spare-part built PC I run my first UNIX ever on,
> > the SCO Xenix. It had a wooping 8 MHz 286 motherboard, a 20 MB HDD and a 360 kB
> > FDD. I used the familly 26" TV for that one, using a SCART contact to get in
> > there. It worked, but I never lifted with XENIX but kept dreaming about hacking
> > a real OS like UNIX. Today I am very happy hacking away on Linux I might add.
> 
> I had a PC where there was a 24Mhz Xtal for the 8086 CPU (div 3 = 8Mhz) 
> and a 28.636MHz clock for the video generation (CGA modes...)
> 
> (Infact you could hack it to use the video clock also for the CPU, to 
> get a whopping 9.54MHz...)

The speed-up kits and all that where really flowing all over the place.
I recall all the lovely stuff you could see in the adds of BYTE at the time.
Oh golly!

> >>>Most other units would also be able to handle this. It's a mystery why
> >>>this was not included into the standard, or maybe it's not as much of a
> >>>mystery :-P
> >>
> >>Interestingly, a number of internal buses of older machines run off
> >>"double MIDI".
> > 
> > 
> > Standards are good, everyone should have their own, but in reality everyone
> > run multiple standards for themselfs.
> 
> My guess would be that they used whole number MHz clocks in those early 
> MIDI synths, and they chose to run the UART from that too. And didn't 
> need to be compatible with teletypes anyway.

Exactly.

> The PC wasn't really a home machine back then.

When they started the MIDI work the PC just got out anyway.

> And homecomputers had often whole numbered multiples of 31.25k as clock
> frequencies, like 2, 3.5 or 4MHz, often without dedicated serial interfaces.

Indeed.

> Infact I have read that there was a game on the ZX Spectrum which could 
> put out midi on the cassette port...
> So at that time there wasn't the notion of an established standard.

The only standards you had was the normal range up to 19,2 kBd (which was
blitzing speed except for those using the big and strange things like 3270 or
Ethernet - then a wooping 10 Mb/s on a THICK cable).

The MIDI and PC took of at about the same time. If you got a Roland MPU 401
you had a separate box and a DB25 over to a card which you stuck into your PC.
The card was basically a few signalbuffers with address-decoding, nor very
different from the ISA cards that later came for IDE.

Cheers,
Magnus