Ghost characters on Atom video

emulators, hardware and classic software for atom + system machines
Post Reply
User avatar
roland
Posts: 4545
Joined: Thu Aug 29, 2013 9:29 pm
Location: Born (NL)
Contact:

Ghost characters on Atom video

Post by roland »

I am refurbishing an Atom. Basically it works fine but it has some annoying ghost characters on the screen and there is also the famous vertical line:
IMG_5504.jpg
In the circles you can see what I mean, especially in the right one. Is there any way to "remove" this? I already replaced Q1 and Q2, all the electrolytic capacitors, L1 and also the 6847 itself. I tried also connecting a small and large capacitor between pin17 and ground but that won't help. Has there been any investigation to this, how to solve?
FPGAtom: 512 KB RAM, Real Time Clock and 64 colours
MAN WOMAN :shock:
Volhout
Posts: 26
Joined: Mon Dec 30, 2019 6:15 pm
Contact:

Re: Ghost characters on Atom video

Post by Volhout »

Dat kan van de voedingsontkoppeling komen. Het is een dunne lijn, dus hoogfrequent. Kan 6847, of ram zijn. Plaats wat kleine c-tjes op de 5V. 10nf of zo.
User avatar
Multiwizard
Posts: 2257
Joined: Wed Jan 11, 2012 9:03 pm
Contact:

Re: Ghost characters on Atom video

Post by Multiwizard »

Volhout wrote:
Thu Jan 13, 2022 10:28 pm
Dat kan van de voedingsontkoppeling komen. Het is een dunne lijn, dus hoogfrequent. Kan 6847, of ram zijn. Plaats wat kleine c-tjes op de 5V. 10nf of zo.
I know I have autism,, but but, why in Dutch?... :-k
Volhout
Posts: 26
Joined: Mon Dec 30, 2019 6:15 pm
Contact:

Re: Ghost characters on Atom video

Post by Volhout »

@multiwizard

Excuses.
My remark translates to:
Looking at how sharp this line is, it is definitely high frequency. I suggest a ceramic cap 10nF at the 5V (close to the pins of the 6847) or the RAMs.

Volhout
User avatar
hoglet
Posts: 10867
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Ghost characters on Atom video

Post by hoglet »

I long time ago, in a place far away, I investigated this line artifact:
viewtopic.php?p=57310#p57310

I concluded it was crosstalk from pin 26 of the 6847 (DA4) to pin 28 of the 6847 (Y).

The Y signal is very prone to crosstalk because it's there is such a low voltage difference between black (0.72V) and low white (0.65V) (i.e. only 70mV). So a very small amount of crosstalk will be visible.

Increasing the capacitive load on DA4 seems to reduce it, which supports the above hypothesis.

What type of video RAM does this Atom have?

Dave
User avatar
roland
Posts: 4545
Joined: Thu Aug 29, 2013 9:29 pm
Location: Born (NL)
Contact:

Re: Ghost characters on Atom video

Post by roland »

This Atom has your noise killer board V4 installed with 32kB static RAM (AS6C62256-55PCN)
I'll try with a capacitor between pin 26 and GND this weekend.
FPGAtom: 512 KB RAM, Real Time Clock and 64 colours
MAN WOMAN :shock:
Volhout
Posts: 26
Joined: Mon Dec 30, 2019 6:15 pm
Contact:

Re: Ghost characters on Atom video

Post by Volhout »

@hoglet: good find.

That is a fast RAM chip, fast rise and fall times. If it is on a socket, maybe check if you have a slower chip (125ns or so).
But is the RAM chip directly connected to the 6847 on this V4 board, or is there a buffer in between ? Then the buffer could be too fast (sub ns rise and fall times).

But if it is crosstalk between pins (or traces on the board) a capacitor at the Y pin may be better. When you add capacitance to the digital signal you may induce high currents into the ground, that could also cause visible artefacts. Best is to put that cap at the buffer/ram, not at the 6847.
But a capacitor at the Y pin is more effective. Play with the value, so it does not degrade the video too much, start with 10pf, but you could go all the way up to 220pf (depending what the output impedance of the 6847 is). This cap is most efficient at the input of the buffer amplifier (basically you narrow the bandwidth of the amplifier).

Volhout
User avatar
Louis.m
Posts: 43
Joined: Fri Jun 07, 2013 12:36 am
Location: the Netherlands
Contact:

Re: Ghost characters on Atom video

Post by Louis.m »

Hey all,

I'm not familiar with the noise killer board, but looking at the original Atom scheme, I'd say that (if Y=pin 28 of the 6847 is still connected to C4 (22µ) and the other side of C4 to the base of Q1) a better place to put a (very tiny) capacitor would be between the base and the emitter of Q1, or in case that doesn't work, between the base of Q1 and GND (near where R10 is connected to GND).
Gr :D :D tings, Louis

[-o< May your wires be long and your nerves be strong❗
User avatar
hoglet
Posts: 10867
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Ghost characters on Atom video

Post by hoglet »

I don't think the way to resolve this is by capacitively loading the Y signal; that will affect the video quality significantly

It's better to add capacitive load to the DA4 signal (pin 26), or add an inline resistor close to the 6847.

Dave
User avatar
roland
Posts: 4545
Joined: Thu Aug 29, 2013 9:29 pm
Location: Born (NL)
Contact:

Re: Ghost characters on Atom video

Post by roland »

I have tried several capacitor values between DA4 and GND but none of them provided a good picture. With a value of 270pF the vertical line at the right side almost disappeared but a new one showed up at the middle of the screen. I had a look at another Atom to see if it had also some artefacts but that one had even a worse picture. I guess it's just one of the characteristics of our Atom :lol:
FPGAtom: 512 KB RAM, Real Time Clock and 64 colours
MAN WOMAN :shock:
User avatar
Louis.m
Posts: 43
Joined: Fri Jun 07, 2013 12:36 am
Location: the Netherlands
Contact:

Re: Ghost characters on Atom video

Post by Louis.m »

Hey Roland,

Did you try my suggestion of placing a very small C between the base and emitter of Q1 or a small C between the base of that transistor and GND ?

An other possible remedy may be found here: http://acornatom.nl/atom_nieuws/1990/nr2/19902030.htm
Gr :D :D tings, Louis

[-o< May your wires be long and your nerves be strong❗
Volhout
Posts: 26
Joined: Mon Dec 30, 2019 6:15 pm
Contact:

Re: Ghost characters on Atom video

Post by Volhout »

@Louis

The article you refer to has a basic flaw. The requirement is 2V with an output impedance of 75 ohm to match the 75ohm terminated video input.
This results in a video amplitude of 1Vpp (in 75 ohm).

The Acorn Atom video amplifier does the best it can (it actually outputs 1Vpp in 75 ohm based on the output voltages of the MC6847).

However, and there in lies the problem, these voltages are not the best levels to amplify into a composite video signal. For a composite video signal the black (backporch) level is reference. The sync must be -0.3V, and absolute white must be +0.7V (total 1V).

The video from the MC6847 is (after amplification) more like -0.4V (sync) and +0.6V (white). When your TV uses sync to reference (it will gain adjust to reference everything to the -0.4V), the white will not be pure white, but more like light grey. When it does not adjust the gain (some monitors/TV's don't) the white will be brighter, but still not pure white.

See attached simulation.
The video circuit is actually quite well designed, it will accept a wide range of transistors (2N2222, BC108B etc..) and still deliver good 1Vpp signal, even at 4.5V on the 5V bus. Many (inverting) amplifier design are more critical and will start clipping. Compliments.

About the crosstalk (artefacts in the video). DA4 and Y are separated by the pin GM2. Can a capacitor at GM2 improve ?
Attachments
Capture2.JPG
User avatar
roland
Posts: 4545
Joined: Thu Aug 29, 2013 9:29 pm
Location: Born (NL)
Contact:

Re: Ghost characters on Atom video

Post by roland »

Volhout wrote:
Mon Jan 17, 2022 10:18 am
About the crosstalk (artefacts in the video). DA4 and Y are separated by the pin GM2. Can a capacitor at GM2 improve ?
No, a capacitor at GM2 does not have any effect on the video quality and artefacts. I tested it although I didn't expect any improvements because the GM lines are very static compared to address and data lines. The only change when switching graphics mode.

I also didn't test with capacitors "after" the Y-output because the artefacts are generated in the 6847 so the cause is at the inputs and not after the output. Adding a capacitor to DA4 and noting that the display changes is enough evidence for that (IMHO). Also adding capacitors after the Y output just decreases the video quality in general.
FPGAtom: 512 KB RAM, Real Time Clock and 64 colours
MAN WOMAN :shock:
Volhout
Posts: 26
Joined: Mon Dec 30, 2019 6:15 pm
Contact:

Re: Ghost characters on Atom video

Post by Volhout »

@roland

I understand you are satisfied with the performance, but if you would like to try, some improvement may be achieved by lowering the bandwidth of the video amplifier.
Generic composite video as used by TV sets should be satisfied with 5-6MHz bandwidth at Y. The video amplifier has around 15MHz bandwidth. See simulation dotted line. The best controlled way to limit the bandwidth is at the base of the second transistor. (the first is driven from the MC6847 chip and there is little know about the output impedance, so changes are less good controlled).

Adding 22pF get's you to 5MHz. Lower values may be better, experiments may be needed.

Another thing that could cause artefacts is the video amplifier power rail (ferrite bead filter L1 with electrolytic capacitor 22uF). That power rail could definitely use a ceramic cap added to filter higher frequency noise from the power rail. Note that the inverting amplifier Q1 output is referenced to the +5V rail, meaning that every bit of noise on that rail will be buffered by Q2 and visible at the monitor output.
Attachments
Capture.JPG
User avatar
roland
Posts: 4545
Joined: Thu Aug 29, 2013 9:29 pm
Location: Born (NL)
Contact:

Re: Ghost characters on Atom video

Post by roland »

I will try that as soon as I have a 22pF capacitor …. I don’t have one at the moment.
FPGAtom: 512 KB RAM, Real Time Clock and 64 colours
MAN WOMAN :shock:
Post Reply

Return to “acorn atom and acorn system series”