Atom Colour Board Thread

[hoglet]

Hi all,

I thought I would start a new thread to record my experiences with various Atom colour boards. Since I got my Colour Atom a few weeks ago, I've tried several different ways of getting a proper colour picture:

- The original Acorn colour board that was fitted with an RGB output
- Phill's new design (based on the LMH1251) with an RGB output
- Using Y PA PB on connector PL4 directly as a YPrPb output
- My own design, that I've been playing with this weekend

My goal has been to drive a HP LP2065 monitor. As this won't directly take a 15.625 KHz video source, I thought I would try one cheap GBS8200 scan convertors from eBay.

In terms of picture quality, I'm looking for a reasonably sharp picture, that's noise/artefact free. And I'm also looking for accurate colours (i.e. reds that are actually red).

I'm going to use the next few posts to describe each of these options in more details, and try to post some pictures of the results.

Enjoy!

Dave

[hoglet]

Original Atom Colour Board manufactured by Acorn

Original Acorn Colour Board

IMG_2193.JPG (582.02 KiB) Viewed 31713 times

Here's a summary of this card:
- outputs Colour Composite Video (to the Atom UHF modulator)
- outputs TTL RGBS
- slows the 6847 frame rate down from 60Hz to 50Hz
- generates colour burst signal for the composite output
- decodes Y, PA, PB signals to digital RGB

The circuit diagram is here:
http://www.stairwaytohell.com/temporary ... rboard.gif

The technical manual is here:
http://electrickery.hosting.philpem.me. ... index.html

Although it's cool to have an original Acorn board, it doesn't work very well. I had the following problems:

- The 60Hz to 50Hz frame rate conversion is very complex. It involves stopping the 6847 clock for 1.7ms twice during each frame to extend the frame duration from 16.6ms to 20ms. While the 6847 clock is stopped, the colour board has to generate it's own sync pulses. The timing of these pulses is not quite the same as the 6847's, and this seemed to confuse my scan converter, so it would loose sync every few seconds.

- The composite output is not able to drive a 75 ohm load - it was designed to connect directly to the UHF modulator input, which is higher impedance.

- The composite sync signal is recovered from the Y output of the 6847 using a LM319N comparator. There is a lot of ground noise on the board, and this makes the sync recovery hard to adjust. Even when it's set optimally, the composite sync signal contains glitches in places where it shouldn't.

- The PA and PB signals are decoded to 1-bit R G and B. This means that orange is output incorrectly as red.

- The Acorn colour board cannot be installed as at the same time as Phill's excellent RAM ROM board, because they physically conflict. And I really wanted to be able to use Phill's RAM ROM board.

To be added later:

It's possible to disable the 60Hz to 50hz conversion.

[hoglet]

Phill's new Atom Colour card

Phill (Prime) on these forums has developed a new colour board, based on the LMH1251 YPrPb to RGB convertor.

Phill's V3 Colour Board

IMG_2234.JPG (530.99 KiB) Viewed 31694 times

There's some more pictures here:
https://www.facebook.com/media/set/?set ... 6286f612ae

I believe there are 20 of these in existence, and a further 10 of the earlier design, and I think Phill is now sold out.

The datasheet for the LMH1251 is here:
http://www.ti.com/lit/ds/symlink/lmh1251.pdf

This is a simpler and more elegant design than the Acorn card.

Here's a summary of this card:
- Based around LMH1251 YPrPb to RGB convertor
- Output RGB only, at 0.7v Pk-Pk
- Replaces the existing power hungry video RAM with a 8K CMOS RAM
- Include a Noise Killer circuit to eliminate the Atom's screen noise

Unfortunately, I hit a couple of snags:

- My GBS 8200 has 75 ohm terminated inputs, and I'm not convinced the chip Phill uses is really designed for driving these directly. The data sheet (to me anyway, page 11) indicates a buffer/amplifier should be used. I get a lot of noise on the picture. Adding an emitter follower transistor to the RGB outputs really cleans up the image. Many SCART inputs on TVs are actually higher impedance than 75 ohms, so I can fully understand why other folk might not have hit this problem.

- I can't get the colours looking quite right; the red is coming out quite purple. It's kind of arguable what is right here. The other thing I don't like is the dark green/dark orange background in text mode. This is down to the Y, PA and PB signals the 6847 outputs not really being accurate Y PR PB signals.

None of these are show stoppers, and I still think the board is incredibly cool.

Phill sent me the circuit diagram and the installation manual. Included here with permission.

PhillsAtomColorV3.zip

(90.35 KiB) Downloaded 463 times

[hoglet]

Component Video Direct from the Atom (almost...)

The GBS-8200 scan converter (more later) will accept component video directly via three phono plugs (Y Pr Pb).

The first thing I tried was just connecting the monochrome composite video signal that's present on connector PL4 pins 9/10 to the Y input. And amazingly, it worked, and gave a very crisp nice black and white picture. That got me wondering if there was an easy way to get Pr and Pb from the Atom.

The 6847 ouputs two colour difference signals:
- PA is approximately Pr (i.e. Y - R)
- PB is approximately Pb (i.e. Y - B)

Each of these is a tri-level signal (1.0v, 1.5v and 2.0v), and together they allow nine different values to be encoded. The 6847 uses these to encode eight colours (red, green, blue, cyan, magenta, yellow, orange, and buff).

Unfortunately, these are high impedance and cannot drive a 75 ohm video load directly. What's needed is a single transistor "emitter follower" amplifier (for each of PA and PB), i.e.

PL4 pin 9/10 (composite video/ground) -> GBS 8200 Y phono
PL4 pin 3 (PA) -> emitter follower -> GBS 8200 PR phono
PL4 pin 7 (PB) -> emitter follower -> GBS 8200 PB phono
PL4 pin 2 (+5v) supplies the power for the emitter followers

This worked pretty well, and maybe not surprisingly gave the same colours as Phill's board.

[hoglet]

Dave (Hoglet)'s Homebrew Colour Board (Part 1)

Being somewhat at a loose end this weekend, I thought I would have a go at building my own colour board, with the following requirements:

- Outputs RGBS, ideally at TTL levels (like the BBC does)
- works with old fashioned CRT RGB monitors (I don't yet have one of these....)
- works with GBS 8200 scan convertor to a plain old LCD monitor
- Connects to PL4 only
- no frame rate conversion
- outputs colours accurately
- output Orange as Orange, not Red (like the Acorn card does)

Well, here it is:

Dave's Homebrew Colour board on a breadboard

IMG_2224.JPG (469.17 KiB) Viewed 31686 times

It's made up as follows:

- Sync Tip Clamp to restore an known DC level to the Y signal from PL4
- Three LM319N's convert the analog Y, PA, PB signals to a 5 bit digital representation (L, AH/AL, BH/BL). This is pretty much lifted from the Acorn Colour board.
- An PALCE22V10 to encode these into R G1 G2 B
- Composite Video pretty much just passed through as Composote Sync
- Four 2N5551 transistors to buffer the RGBS signals enabling 75 ohm video loads to be driven

I used an old fashioned PAL, because this seemed more in period, it's very fast, and I happened to have access to an EPROM programmer that also programmed these.

There are two trim-pots
- One adjusts the sync tip clamp circuit
- The other sets the threshold for the Y to 1-bit Luminance conversion

Here's another picture where you can also see the Atom and the GBS 8200 scan converter:

IMG_2223.JPG (443.67 KiB) Viewed 31687 times

Here's the latest schematics, layout and PAL equations:

HogletColourBoardV3.zip

V3 Schematics and PCB

(132.47 KiB) Downloaded 548 times

Next up, what does the output actually look like?

[hoglet]

Dave (Hoglet)'s Homebrew Colour Board (Part 2)

Here's what it actually looks like.

First some test images,

Note the Orange is actually Orange, as distinct from Red.

IMG_2206.JPG (379.73 KiB) Viewed 31681 times

IMG_2207.JPG (386.68 KiB) Viewed 31681 times

IMG_2208.JPG (267.87 KiB) Viewed 31681 times

IMG_2209.JPG (246.03 KiB) Viewed 31682 times

And now some games,

IMG_2214.JPG (253.93 KiB) Viewed 31681 times

IMG_2215.JPG (316.1 KiB) Viewed 31681 times

IMG_2216.JPG (274.1 KiB) Viewed 31681 times

IMG_2218.JPG (303.65 KiB) Viewed 31681 times

IMG_2219.JPG (299.28 KiB) Viewed 31681 times

IMG_2221.JPG (344.3 KiB) Viewed 31681 times

[hoglet]

Dave (Hoglet)'s Homebrew Colour Board (Part 3)

Some more hardware geeky details

1) Getting the PAL right

It took me about five attempts to get the PAL equations right. I wanted to pare down the use of the L input signal to where it's the only way to distinguish two valid colours (e.g. black/white, dark green/green, dark orange/orange)

Code: Select all

/*
RED OUTPUT (1 BIT)

                 AL AH BL BH  L  R G1 G2  B
YELLOW   1.5 1.0  0  0  1  0  X  1  1  1  0
RED      2.0 1.5  0  1  0  0  X  1  0  1  0
MAGENTA  2.0 2.0  0  1  0  1  X  1  0  1  1
BUFF     1.5 1.5  0  0  0  0  1  1  1  1  1
ORANGE   2.0 1.0  0  1  1  0  1  1  1  0  0
*/

R = (!AL & !AH & BL & !BH & L) # (!AL & AH & !BL & !BH) # (!AL & AH & !BL & BH & L) # (!AL & !AH & !BL & !BH & L) # (!AL & AH & BL & !BH & L);
RZ = 'b'0;
RZ.OE=!R;
/*
GREEN OUTPUT (MS BIT)

                 AL AH BL BH  L  R G1 G2  B
YELLOW   1.5 1.0  0  0  1  0  X  1  1  1  0
CYAN     1.0 1.5  1  0  0  0  X  0  1  1  1
GREEN    1.0 1.0  1  0  1  0  1  0  1  1  0
BUFF     1.5 1.5  0  0  0  0  1  1  1  1  1
ORANGE   2.0 1.0  0  1  1  0  1  1  1  0  0
*/
G1 = (!AL & !AH & BL & !BH) # (AL & !AH & !BL & !BH) # (AL & !AH & BL & !BH & L) # (!AL & !AH & !BL & !BH & L) # (!AL & AH & BL & !BH & L);
G1Z = 'b'0;
G1Z.OE=!G1;

/*
GREEN OUTPUT (LS BIT)

                 AL AH BL BH  L  R G1 G2  B
ORANGE   2.0 1.0  0  1  1  0  1  1  1  0  0
*/
G2 = !(!AL & AH & BL & !BH & L);
G2Z = 'b'0;
G2Z.OE=!G2;

/*
BLUE OUTPUT (1 BIT)

                 AL AH BL BH  L  R G1 G2  B
BLUE     1.5 2.0  0  0  0  1  X  0  0  1  1
CYAN     1.0 1.5  1  0  0  0  X  0  1  1  1
MAGENTA  2.0 2.0  0  1  0  1  X  1  0  1  1
BUFF     1.5 1.5  0  0  0  0  1  1  1  1  1
*/
B = (!AL & !AH & !BL & BH) # (AL & !AH & !BL & !BH) # (!AL & AH & !BL & BH) # (!AL & !AH & !BL & !BH & L);
BZ = 'b'0;
BZ.OE=!B;

2) The quality of the output.

Here's a test pattern:

IMG_2231.JPG (92.46 KiB) Viewed 31677 times

And here's the Green output on a scope:

IMG_2230.JPG (126.11 KiB) Viewed 31677 times

Tektronix 2235, 200ns/division horizontal, 1V/division vertical

If you look carefully, you can see the black parts of the Inverse I have nice clean edges.

3) Dealing with phase delay between Y and PA/PB

An LM319N comparator is used to threshold PA/PB. I started out with the resistor values from the Acorn Colour board, but the result was poor. In the above test pattern the vertical bar had a leading white pixel that was very obvious. It turns out that my decoded PA/PB signals were lagging the decoded Y signal by about 200ns (1-2 pixels). Turns out this was because the 6K8/56K divider values were too high, resulting in very slow edges. I reduced these to 1K1 and 7K5 and the delay vanished, leaving nice crisp edges.

[hoglet]

GBS-8200 Scan Converter

Here's the GBS-8200 Scan Converter I'm using:

IMG_2225.JPG (491.62 KiB) Viewed 31655 times

I got this on ebay for £23 (inc postage) from 3d_professional_shop (not sure if this was a one-off)

Here's a link to the PDF manual:
http://info-coach.fr/atari/hardware/video/GBS-8200.pdf

This works well with the BBC as well (as long as you make sure to get one with the V4 firmware or later). The only problem is the output frame rate is fixed at 60Hz, so scrolling (e.g. Rocket Raid) is not so smooth.

Over Christmas I made a concerted effort to hack this:
- The microcontroller is a Myson MTV230M (I have a the datasheet)
- The video processor is a Tvia 5725 (I have a the datasheet)
- The firmware upgrade connector is P6 is I2C to the EEPROM and MTV230M
- The other connector (P5) is I2C to the TVIA 5725
- I was able to connect my Raspberry Pi's I2C pins to P6 and use i2cdetect
- I was eventually able to write a program to extract the firmware
- I had a go at disassembling this (it's 8051 assembler)
- I was able to twiddle some registers on the 5725 and change the output frame rate to 50Hz and scrolling from the Beeb was much smoother (my HP LP2065 syncs to 50Hz)
- I think I know how to write a program to write new firmware to the board
- I'm too scared to actually do this

[hoglet]

Useful links for the MC6847

A clean copy of the data sheet is here:
http://www.colorcomputerarchive.com/coc ... rola).pdf

One of Keith Howell's pages has some great information:
http://members.casema.nl/hhaydn/howel/l ... _clone.htm

There's a great discussion of the colours that the 6847 generates here:
http://forums.bannister.org/ubbthreads. ... ea6062aeab

They feed Y/PA/PB into the YPrPb->RGB color matrix and use these colours in the CoCo simulator. These colours are exactly the same as Phill's board generates.

Here's another link that shows the expected colours:
http://hcvgm.org/VDG_Colours.html

Other useful links:

This helped me understand sync tip clamping:
http://pdfserv.maximintegrated.com/en/an/AN3303.pdf
http://gamesx.com/wiki/doku.php?id=av:g7000_av

[hoglet]

Future work

There's a few more things need tweaking with my Home brew board:

- The luminance thresholder is a tad sensitive, and I'm wondering if there is any way to improve this.

- On the Alternative Colour Alpha screen (the orange one) I think the orange semi graphics should be darker than the inverse video text. Not sure if this is possible to do, it would involve a second luminance thresholder, and more bits of output. Given this mode is rarely used, I'm not sure I'm going to bother.

- I might have a go a designing a PCB for this. I've got a UV exposure box knocking about somewhere. If I did this, I'd probably include Phill's noisekiller and 8K RAM circuits.

Anyway, that just about it for now! If anyone is still reading

[hoglet]

Links to other Atom colour boards:

Ha! Used up all the reserved posts!!!

- October 1982 issue of Acorn User contains a design by Paul Beverly:

http://8bs.com/cgi-bin/magsel.cgi?http: ... er8210.zip

- Kees sent me a link to a Dutch design:

I've attached a document with a Dutch design rebuilding the Atom Colourboard to get digital RGB and convert it to a Scart-, CGA- or CVBS signal. There's also a supply for a 80-column board for the Atom with coloured text based upon the 9345 videoprocessor.

Here's the design:
http://stardot.org.uk/forums/download/file.php?id=5126

Does anyone know of any more designs?

[1024MAK]

My god!
He's reserved 9 places, how many women is he having as guests?

[oss003]

Hi Dave,

if your GBS 8200 has a VGA connector and can handle a Hsync of 15 kHz, maybe you can try this:

Code: Select all

Colourboard         VGA 
    CN2          connector   Remark
------------------------------------------------------------
     1 --------     1
     2 --------     2
     3 --------     3
     4 -x
     5 -x
     6 -x
     7 --------    13        On colourboard to IC3C 74LS02 pin9 (Hsync)
     8 --------    14        On colourboard to IC3C 74LS02 pin9 (Vsync)
     9 -x
    10 -------- 6,7,8,10

I tried it on a normal VGA monitor but the Hsync displayed 2 merged images.

Greetings
Kees

[MartinB]

I don't have an Atom ( ) but an interesting thread and some great work there. Nice

[hoglet]

Hi Kees,

if your GBS 8200 has a VGA connector and can handle a Hsync of 15 kHz, maybe you can try this:

Last week I made a 6 Pin DIN to VGA Cable (R, G, B and HS) without resistors and connected this to the GBS 8200. It locked onto the picture perfectly, and the colours were as described above.

Is connecting to CN2 any different?

Although it's not clear from the data sheet, all the different video inputs of the GBS 8200 are actually joined directly together (apart from the YPrPb phonos).

Dave

[hoglet]

MartinB wrote:

I don't have an Atom but an interesting thread and some great work there. Nice

There are two on ebay at the moment:
http://www.ebay.co.uk/itm/Acorn-Atom-Vi ... 0833934440

http://www.ebay.co.uk/itm/Fully-Expande ... 1137518326

Price of the second one is a bit LOL!

Dave

[retroclinic]

http://www.ebay.co.uk/itm/Fully-Expande ... 1137518326

Price of the second one is a bit LOL!

Dave

I congratulate that seller on his clearly impressive derangement.

Mark.

[oss003]

Hi Dave,

Nice work, looks good!!

Is connecting to CN2 any different?

No but I added the Hsync and Vsync signals to 2 unused pins of CN2 needed for the VGA output.

Greetings
Kees

[Prime]

No but I added the Hsync and Vsync signals to 2 unused pins of CN2 needed for the VGA output.

Out of interest Kees, which sync did you connect to which pin, and I'll add it to the board, incase there is a next version....

Cheers.

Phill.

[oss003]

Out of interest Kees, which sync did you connect to which pin, and I'll add it to the board, incase there is a next version....

Code: Select all

CN2  To
 7   Colourboard IC3C 74LS02 pin9 (Hsync)
 8   Colourboard IC3C 74LS02 pin9 (Vsync)

Greetings
Kees

[Prime]

Phill's new Atom Colour card

Phill (Prime) on these forums has developed a new colour board, based on the LMH1251 YPrPb to RGB convertor.

I believe there are 20 of these in existence, and a further 10 of the earlier design, and I think Phill is now sold out.

That is correct.

Phill sent me the circuit diagram and the installation manual. I'd like to include these if Phill's OK with that.

Yeah sure.

Phill.

[Prime]

Dave (Hoglet)'s Homebrew Colour Board (Part 1)

Being somewhat at a loose end this weekend, I thought I would have a go at building my own colour board, with the following requirements:

I used an old fashioned PAL, because this seemed more in period, it's very fast, and I happened to have access to an EPROM programmer that also programmed these.

Yeah programmable logic (of any sort) generally helps when you're not actually sure what logic you need, as in this case.

To be added later:

The schematics, when I get around to drawing them up.

I'd also be interested in seeing these, my first attempt at a colour board was basically a copy of the Acorn board but without the scan rate converter. But like the Acorn board it was difficult to adjust and didn't produce very satisfactory results.

It might be usable as well with the Dragon, Tandy CoCo, MC-10 etc as these also uses the 6847.

If you do make a board, feel free to borrow the RAM / Noisekiller circuits.

Cheers.

Phill.

[hoglet]

If you do make a board, feel free to borrow the RAM / Noise killer circuits.

Thanks Phill, I'll try to get the schematics drawn up soon.

What is the origin of the noise killer circuit?

I've been trying to figure out how it works, and failing.

Dave

[hoglet]

Updated earlier post with Schematics:
http://www.stardot.org.uk/forums/viewto ... 123#p56123

[hoglet]

Updated earlier post with V2 Schematics and PCB Layout

http://www.stardot.org.uk/forums/viewto ... 123#p56123

The PCB is 4.7" x 3.2" in size, and single sided with seven short links on the top side. It is designed to sit outside the case, so I can carry on using Phill's board internally for the 8K RAM and Noise Eliminator. I laid this out manually using DipTrace:

HogletColourBoardV2.PNG (91.6 KiB) Viewed 31366 times

Plan is to dig out the light box from the back of the garage this weekend, and have a go at etching it.

[Prime]

Dave,

Would it be possible to post an archive with the files you used to compile the PAL ?

Cheers.

Phill.

[hoglet]

I had a fun day yesterday etching the circuit board then building it up. The PCB design is not particularly ambitious (in terms of density). This was really just an experiment to test the process. It's the first time I've done this in about 10 years.

Anyway, I thought it might be interesting for folk if I posted a few pictures.

The starting point is to print out the layout on some kind of translucent film. It's important that this is as dense and crisp as possible. It also helps to print the mask mirrored such that the printed side ends up being placed against the PCB.

In the past I've had excellent results with LaserStar drafting film and a trusty old HP Laserjet 2100 that I picked up at a Computer Fair for £5.

IMG_2239.JPG (139.96 KiB) Viewed 31303 times

Here's the end result:

IMG_2236.JPG (177.79 KiB) Viewed 31303 times

And a slightly blurred close up:

IMG_2238.JPG (107.25 KiB) Viewed 31303 times

The next ingredient is a UV Light Box. I made this about 20 years ago out of four 15W UVA Tubes from RS:

IMG_2241.JPG (207.65 KiB) Viewed 31303 times

The mask is placed on the glass (toner side up), the PCB goes next, (photosensitive copper side down), a bit of foam and close the lid. Cook for 4 minutes, then down the the Wet Lab (aka the Utility Room):

IMG_2240.JPG (164.72 KiB) Viewed 31303 times

The board needs to be developed in a solution of Sodium Metasilicate (RS 690-849). This dissolves the photo resist coating where it was exposed to UV:

IMG_2244.JPG (122.86 KiB) Viewed 31303 times

After washing, the board is etched in a solution of Ferric Chloride Hexahydrate (RS 551-277). Careful not splash this, as it stains and is generally unpleasant. It helps speed up the etching time if the solution is warmed first:

IMG_2247.JPG (137.35 KiB) Viewed 31303 times

The etching process takes about 15 minutes. Once complete, the board is washed in water and dried, and the end result looks like:

IMG_2252.JPG (227.32 KiB) Viewed 31303 times

I was pretty happy with this. Here's a close up of area around one of the IDC connectors.

IMG_2253.JPG (93.16 KiB) Viewed 31303 times

The board is then drilled by hand (using a dremmel and a 0.7mm bit) and given a good going over with wire wool to remove any photo resist.

[hoglet]

Here's a few pictures of the build.

First the wire links and resistors go in:

IMG_2263.JPG (217.05 KiB) Viewed 31303 times

Then the rest of the components:

IMG_2268.JPG (209.61 KiB) Viewed 31303 times

Finally the ICs:

IMG_2269.JPG (214.72 KiB) Viewed 31303 times

And here it is, working:

IMG_2270.JPG (202.02 KiB) Viewed 31303 times

I've still a bit more debugging to do. With a longer cable between PL4 and the board I'm getting a bit of ghosting on the screen. I think this is crosstalk from some of the other signals on PL4 onto PA/PB (specifically INV and HS). It doesn't happen with a short cable. It's not bad, but I'd like to fix it somehow.

[hoglet]

Hi Phill,

Would it be possible to post an archive with the files you used to compile the PAL ?

Here you go:

HogletColourPalV6.zip

(5.33 KiB) Downloaded 304 times

The .pld file is the one with the equations in it.

I used WinCUPL, which is freely available from Atmel
http://www.atmel.com/tools/WINCUPL.aspx
(You need to register for to get a license key but it was quick)

I used Cypress PALCE22V10 PALs from Quarndon, because they were cheap (£1.56), FLASH Erasable and supported by the Datamann Pro40 I borrowed from work.

[Prime]

Hi Phill,
I used WinCUPL, which is freely available from Atmel
http://www.atmel.com/tools/WINCUPL.aspx
(You need to register for to get a license key but it was quick)

Yeah I have WinCUPL, as I started out using Atmel ATF PALs (before I started working with CPLDs), so that seemed the best choice

I used Cypress PALCE22V10 PALs from Quarndon, because they were cheap (£1.56), FLASH Erasable and supported by the Datamann Pro40 I borrowed from work.

I have some Atmel ATF22V10s so I would imagine those would work.

I've re-layed out your board, as I can do double sided, and I'll probably panelize it with some other small boards onto a DS Eurocard. Really all I've used the top layer for is routing the power tracks

I'm thinking that I'd also like to try this circuit with the Dragon, which also uses the 6847, I tried an early version of the Atom colour board and didn't think it was that good, worse than on the Atom.

Hummm I may have to add the inverter circuit from the Atom motherboard for Y will fit a jumper to enable it so I can do initial tests on the Atom.

Cheers.

Phill.