VGA
ZXNet echo conference «real.speccy»
_Skar from I.O.G._
Greetings, All!
I have a VGA monitor (with the help of programs it understands 800*600). So that's the question
- is it possible to connect a real spec to it, or rather the Pentagon?
May _Speccy_ be with you, All!
_
' )--. [ARIA] [ZX.SPECTRUM] [486+DOS] [s...@bbs.spez.com.ua]
/'|_M_|` CUSTOMS FEES WITH UPS (DHL etc.) Must _DIE_!!! =:E~E~~~
The question is not an idle one.
Now, judging by fido7.zx.spectrum, many people are thinking about this. I'm trying to make a prototype of an RGB2VGA adapter at home, the idea is that you need to memorize the image row and repeat it twice. At the same time, pixel-by-pixel synchronization must be made internal, so as not to interfere with the spec. Opinions differ about Pixel CLC, 7.5 or
7 MHz, I’ll check it. In the meantime, I'm playing with the Pentagon, I connected an EGA to it.
You can still buy an EGA for 5-10 dollars. on the internet on Molotok, for example.
Success.
Konstantin.
"Alexandr Skoropisov" <Alexandr....@p34.f68.n461.z2.fidonet.org>
wrote in message news:10411...@p34.f68.n461.z2.ftn...
04 Jan 03 11:38, Konstantin Tuzov wrote to Alexandr Skoropisov:
KT> Now, judging by fido7.zx.spectrum, many are thinking about this. I'm trying
KT> to make a layout of an RGB2VGA adapter at home, the idea is that you need to remember
KT> the image stack and repeat it twice. In this case, pixel-by-pixel
blinking will be too noticeable. It is better to remember the frame and output it at double frame frequency of the Spectrum. You also need to somehow blur every second line,
otherwise everything will be square, like in emulators.
KT> synchronization must be done internally, so as not to interfere with the spec.
KT> Opinions differ about the Pixel CLC, 7.5 or 7 MHz, I will check.
Of course, 7.0 MHz in most Soviet clones. Well, I screwed up... But in the bourgeois models there is a little more than 7.xxxMhz, a number that is a multiple of the carrier frequency of the PAL encoder. Some Soviet Spectrum models also had quartz close to
14Mhz, but not exactly 14Mhz (due to shortages). Usually these were 13, 13.5, 14.5, 15,
16 Mhz and by the digits of the counter inside the ZX this frequency was divided so that the lowercase
remained 15625Hz.
To obtain a high-quality image (and in 512x192 mode) in any case
digitization must be done at least at 14 Mhz and with a sufficiently large bit
of each colors, bit 5 per color for example.
Imho, it’s easier to buy a ready-made TV tuner for a monitor, cheaperwill be released.
Greetings, Kirill!
KF> Imho, it’s easier to buy a ready-made TV tuner for a monitor, it will be cheaper. And a PAL encoder for it “for focusing” :) - Divorce the scheme, big and small... [Sprinter] Sprinter-II [Forth-CPU] ... ivan...@mail.ru * http://winglion.spb.ru * http://www.petersplus.ru
The protoss escaped. The zerg dug in. It's time for me to call it a day. Ivan.
It’s more difficult to remember a frame in hardware, and there’s no point in doing so. It’s easier to record a line with a 7 MHz clock and play it back with a 14 MHz clock. However, what is there to digitize?
Colors are output to the connector as a three-digit code (or am I confused?). This is what we need to
remember. But what to remember is also a question. Of the available static memory chips, only 155RU7, also known as 93425 or 74S214, works. In short,
address sampling time should be no more than 50 ns.
08 Jan 03 13:01, Konstantin Tuzov wrote to Kirill Frolov:
KF>> To obtain a high-quality image (even in 512x192 mode) in
KF>> in any case, digitization must be done at least at 14Mhz and with a sufficiently
KF>> large bit depth of each color, 5 bits per color for example.
KT> It is more difficult to memorize a frame in hardware, and there’s no need for that.
The amount of RAM is needed more and the bit capacity of the counters too.
KT> It’s easier to record a line with a clock of 7 MHz and play it back with a clock of 14 MHz.
On modern monitors, with 50Hz frame scanning, tears from the eyes
flow in streams after only 5 minutes.
KT> However, what is there to digitize?
And all sorts of Flash-Color (14Mhz Flash)?
KT> Colors are output to the connector as a three-digit code (or
KT> am I confusing?).
An analog signal is output to the connector. In reality, there are 3 gradations of each
signal and to remember the color of a pixel, 4 bits (RGB and brightness) are enough.
KT> This is what needs to be remembered. But what to remember is also a question.
KT> Of the available static memory chips, only 155RU7, aka
KT> aka 93425 or 74S214, works. In short, the address sampling time should be no
KT> more than 50 ns.
The minimum you need to do is one write and two reads in 140 ns. An 8-bit memory with an access time of less than 100 ns or a 32-bit memory with an access time of 400 ns is suitable - you can install a regular long SIMM...You just can’t do this, the image will be of crappy quality, as if it had been slightly enlarged or reduced in Art-Studio... There is no synchronization with the pixel clock spec, the bit capacity of the ADC is low, the digitization frequency is the minimum possible (equal to pixel clock). That's
that's why I initially proposed 14Mhz and 5 bits per color.
But 155RU7 is not serious, it's better not to meddle here without specialized microcircuits, or without
FPGAs. Of course, it can also be done on the 155 series, but it will be
a device larger than the Spectrum.
KF> And all sorts of Flash-Color (14Mhz Flash)?
Digitization will not help - at any given time there will be only one color on the output
: either paper or ink.
KT>> Colors are output to the connector as a three-bit code (or am I confusing it?).
KF> An analog signal is output to the connector. In reality, there are 3 gradations
KF> of each signal, and to remember the color of a pixel, 4 bits (RGB
KF> and brightness) are enough.
Exactly - 4, but you don’t need to digitize anything, the signals need to be taken from the board
Spectrum, and the DAC should be placed at the output of the scandabler.
KT>> It needs to be remembered. But what to remember is also a question. Of the
KT>> available static memory chips, only 155RU7 rolls, aka
KT>> 93425 or 74S214. In short, the address sampling time should be no more than 50
KT>> ns.
KF> The minimum you need to do is one write and two reads in 140 ns. A
KF> 8-bit memory with an access time of less than 100 ns or a 32-bit memory with a
KF> access time of 400 ns is suitable - you can install a regular long SIMM... Only
Bourgeois statics: 6116... 61256 15 ns. you can remove
from dead PC motherboards, and you can also take our analogues:
*** Domestic analogues of STATIC RAM
STATIC RAM Analog
---------------------------------------
6116 (2K*8) KR537RU8
---------------------------------------
6116(2K*8) KR537RU10
------------------------------------
6116 (2K*8) KM581RU5
----------------------------------------
6264 (8K*8) KR537RU17
C courtesy Sergey Moskalev [Grianyov. [Cruise] [White Guard] | [ZX-Spectrum] [Amiga]
KF>> bits (RGB and brightness).
SM> Exactly - 4, but you don’t need to digitize anything, the signals need to be taken from the
SM> Spectrum board, and the DAC placed at the output of the scandabler.
You can simply redo the spec at 31 KHz. Well, multicolors will not work,
who needs them... For demos you can also connect a TV...
SM> you can remove it, and you can also take our analogs:
SM> STATIC RAM Analog
SM> ---------------------------------------
SM> 6116 (2K*8) KR537RU8
SM> ---------------------------------------
SM> 6116 (2K*8) KR537RU10
SM> ---------------------------------------
SM> 6116 (2K*8) KM581RU5
SM> ---------------------------------------
SM> 6264 (8K*8) KR537RU17
To store a frame of a television image you need ~1-2Mb.
KF> The filter is needed at <7Mhz.
define 7MHz and let the scandabilizer mix the colors itself.
KF> You can simply redo the spec at 31KHz. Well, multicolors
KF> will not work, but who needs them... For demos, you can also connect a TV...
Yeah, thanks. Then why is all this necessary if the TV
is available at any time?
KF> To store a frame of a television image you need ~1-2Mb.
Let's calculate:
screen in Scorpio (in mine) 2x232x312=144768 (pixels)
depth 4 bits, i.e. 144768x4/8=72384 bytes
In general, we don’t need the entire frame. Two line buffers
of 464 bytes each are enough, then 6116 is enough.
store the even line in buffer 0
output the even line from buffer 0 and store the odd line in buffer 1
output the odd line from buffer 1 and store the even line in buffer 0
and etc.
KF>> ~1-2Mb.
SM> Let's count:
SM> screen in Scorpio (in mine) 2x232x312=144768 (pixels)
SM> depth 4 bits, i.e. 144768x4/8=72384 bytes
I think like this: 5 bits per color (actually 8 is better), ~1000 pixels horizontally, ~700 lines vertically. A meter is not enough...
SM> In general, we don’t need the whole frame. Two line buffers
SM> of 464 bytes each are enough, then 6116.
50Hz is enough. Why connect a monitor? I think so, so as not to break your eyes on TV.
KF> I think this: 5 bits per color (actually better than 8), ~1000 pixels
These are the costs of your “universal” option
KF> horizontal, ~700 lines vertical. A meter is not enough...
On the Amiga you need to connect to the RGB output, because soldering 24 legs is very
problematic, but on the Spectrum there are only 4 - do you see the difference?
You need to connect to the digital outputs.