Strange IndvisionAGA flicker issue

[TheCorfiot]

@rkauer

OK that's good, I was thinking R214 as well but why does ALICE not get fried in the A1200 when E127R is shorted, It's a similar arangement is it not ? CCK connects directly to LISA's CCK input once the resistor is shorted.

Thanks for the suggestion, please keep them coming

TC

[rkauer]

Checking again: on A4000 there is an inverter to act as a buffer...

After it there is the same resistor value...


Hmm... Better low its value too.

[TheCorfiot]

Back of the board R214 / R217


Zetty what are your thoughts my friend,

[rkauer]

My advice: low the values of both resistors to some like 27~33 Ohm.

Haven't been here in a while, I had to re-register my account.

After AmigaKit brought this to my attention, I started investigating what E127R is about, and found that it's equipped with ferrite beads (almost equivalent to a bridge) on almost all my boards. Only my very first A1200 board has a resistor in E127R, and that's 27 ohms.

E127R is a series resistor in the CCK line that goes to Lisa. Here's two measurements on two different boards. First the one with a resistor:



Looks very nice - no overshoots, no undershoots, pretty exactly 50% pulse width. Now the one with the ferrite bead, which is almost equivalent to a bridged resistor:



This one doesn't look very nice, but obviously fixes a problem. My assumption is that in higher resolution modes, the double-cas access is giving some trouble, so I measured the CCK line against the CAS line, which goes to the memory chips _and_ to Lisa for proper timing. Again, I started with the board that has the resistor in place:



...and compared it to the other board with the ferrite in place:



...but didn't really find a difference. Now, we're looking for a fault that shows only when the computer has warmed up, so it's very probable that there isn't really a difference on the signals outside, but _within_ the chips. The key variable is "threshold" - which is unknown. I could not find any specs on the input threshold of the CBM or the HP Lisa chip, but seeing that the voltage swing on CCK is different, I'm assuming that the positive width and the negative width is seen differently from what my oscilloscope is measuring.

AmigaKit asked me to translate the E127R fix to the A4000D, so I continued measuring that one. The CCK signal looks quite different, but the timing against a double-Cas access looks OK:



However, the voltage swing on CCK is considerably lower, which points to the threshold question again. I have to add that none of my computers shows any of the reported problems, so the following fix is just a shot in the dark:



This is a 1Kohm resistor on the CCK line to 5V. Without it, the voltage swing was hardly 3.4V, and with the resistor in place, the swing is well over 4V, which might lift it above a (so far imaginary) critical value of Lisa. Here's the result:



Again, none of my Amigas shows the problem, so if someone who can reproduce the problem could verify if this helps, I'd appreciate it.

Jens

I also have this problem in highgfx and 1024x768, i have a rev. 1B A1200 motherboard if this helps. In Highres PAL 640x512 is rock solid.

[TheCorfiot]

Jens

Firstly, I want to thank you so very much for looking into this issue which Amigakit have passed on at my request.

My A4000D is showing this effect as Matthew can verify, Can you give me your consent to adding the 1K Resistor onto my Indivision A4000

If so I will be happy to do so & let you & Matt know my findings.

Again thank you very much for your support my friend and I really appreciate all your efforts in trying to solve this issue.

Very Best Wishes
TC

Originally Posted by TheCorfiot

My A4000D is showing this effect as Matthew can verify, Can you give me your consent to adding the 1K Resistor onto my Indivision A4000

I don't exactly know your soldering skills, but if you are careful and have some experience with your soldering iron, you can hardly break anything. This will not affect your warranty, if that's what you mean by "giving consent".

Jens

Originally Posted by kouts28

I also have this problem in highgfx and 1024x768, i have a rev. 1B A1200 motherboard if this helps. In Highres PAL 640x512 is rock solid.

Lisa temperature rises significantly with SHires pixel clock (so does the Hires Denise!), which seems to affect bus timing. Since you have a very old board, your E127R might be equipped with a resistor instead of a ferrite bead. Shorting E127R will most probably solve your problem.

E127R is located on the bottom side of the A1200 mainboard. The two possible parts can easily be identified: If it's a black block with height about equal to it's width and no number printed on it, it's a ferrite bead. If it's a flat thing (much lower height than width) and some numbers are printed on it, it's a resistor. In that case, just short it.

Jens

[FOL]

Originally Posted by Schoenfeld

Lisa temperature rises significantly with SHires pixel clock (so does the Hires Denise!), which seems to affect bus timing. Since you have a very old board, your E127R might be equipped with a resistor instead of a ferrite bead. Shorting E127R will most probably solve your problem.

E127R is located on the bottom side of the A1200 mainboard. The two possible parts can easily be identified: If it's a black block with height about equal to it's width and no number printed on it, it's a ferrite bead. If it's a flat thing (much lower height than width) and some numbers are printed on it, it's a resistor. In that case, just short it.

Jens

Thanks for the info jens. I still need to contact you regarding another problem I have noticed.

A noob question, what does "shorting" means? I have to put a blob of solder on it in order to connect the two points?

[TheCorfiot]

Originally Posted by kouts28

A noob question, what does "shorting" means? I have to put a blob of solder on it in order to connect the two points?

Absolutely, but a short thread of copper wire (with tweezers) would be the preferred option & less untidy.

TC

[Cammy]

I'm not sure if this is the same problem other people get, but this is what happens sometimes on my A1200 when I'm running in HighGFX 1024x768. I have noticed that it usually happens more often if there is some audio playing from one of the programs on Workbench (Eagle Player or sounds from SabreMSN). The screen keeps flashing black, like it's constantly losing the signal. Here's a video of what happens -

[ Show youtube player ]

Is this the same problem other people get or something different?

[Peter]

That is weird Cammy... it does seem that the indivision has some strange habits to say the least... my A1200 is set at the same screen mode as yours and when I first turn on my Amiga I experience some picture flickering and a little break up in areas... after only a minute or so the picture settles and is rock solid - it is almost like it has warmed up a bit like an old (really old) TV set...!

Do you have the same problem in other screen modes?
I wonder if the indivision testers have an opinion as to which screen mode is most reliable??

[ajacocks]

I have a similar issue with my Indivision AGA, with my rev. 1B A1200. The picture steadies, and becomes rock-solid after a couple minutes of use. I have to believe that this issue is somehow heat-involved.

[Cammy]

With this A1200, when it starts up everything is rock solid, and about 90% of the time it stays this way. I have only noticed that sometimes after leaving it on Workbench for a long time, like half a day or more, with several applications running, it starts getting this flashy problem. And I'm pretty sure it always only happens after one of the programs has played some audio. It will continue flashing like this until I reboot, then it's fine again.

[NovaCoder]

Originally Posted by Cammy

With this A1200, when it starts up everything is rock solid, and about 90% of the time it stays this way. I have only noticed that sometimes after leaving it on Workbench for a long time, like half a day or more, with several applications running, it starts getting this flashy problem. And I'm pretty sure it always only happens after one of the programs has played some audio. It will continue flashing like this until I reboot, then it's fine again.

How bizzare, this is very different to the flicker issue that I had

[TheCorfiot]

Jens / Matthew

I fitted the 1K resistor mod this AM to my A4000 / Indivision AGA, removed added cooling from ALICE & refitted the lid.

The system has now run SuperHires & HighGFX 1024x768, 28MHz screenmodes perfectly for six hours. The display is stable & perfect.

So it's fair to say it has eliminated my problem & is a suitable fix.
I thank you very much for that

On a different symptom can I ask If you experinece any problems with the following config.

1 640x512 64 colour mode
2 Classic Workbench OS3.9
3 8MB SIMM fitted in ChipRAM SIMM slot.

I'm experiencing screen redraw / update corruption with the ICONS in 640x512 PAL mode.

Thanks again

TC

[NovaCoder]

Originally Posted by TheCorfiot

On a different symptom can I ask If you experinece any problems with the following config.

1 640x512 64 colour mode
2 Classic Workbench OS3.9
3 8MB SIMM fitted in ChipRAM SIMM slot.

I'm experiencing screen redraw / update corruption with the ICONS in 640x512 PAL mode.

Thanks again

TC

Are you running any patches apart from bb2? Are your icons setup to run in fast ram (workbench prefs)?

[TheCorfiot]

@Novacoder

Only BB2 installed & no ROM update as the current HD is only 4GB in size.

Please expand a little on the "settings" you have mentioned as I usually use Classic WB as setup by default.

What is interesting / annoying & suggests an Indivision problem is that when the IndivisionAGA is removed from the A4000 all operation is normal & everything works just fine when using onboard LISA.

TC