Overview Features Instructions Performance Forum Downloads Products Reseller Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
VISIT APOLLO IRC CHANNEL



All TopicsNewsPerformanceGamesDemosApolloVampireCoffinReleasesLogin
Information about the Apollo CPU and FPU.

SAGA Flickering With Black Screenpage  1 2 3 4 5 

Satviewer G

Posts 3
08 Apr 2017 19:29


Jari Eskelinen wrote:

Ma Sch wrote:

      3. the used voltage regulators has too small output capacitors of 10uF. The datasheet set the capacitors to a nominal value of 2 pcs. 22uF (because lower ESR)!
      This is the problem, when the CPU do more at the core or at bus. The current will be increased at shortest time, the regulator get power from main PCB. Spikes results in trouble of DIGITAL-VIDEO signals.
   

   
    This was good info. I had some problems exactly like described in this thread - IDE activity causing screen flickering. Also PCMCIA activity did this. Even worse, some occasional system crashes, seemed to relate to flickering. This happened only after system had heated up, I initially suspected cold joints but now it sounds more plausible that heat causes voltage regulator capacitors to drop their capacitance even lower from 10uF which makes it things worse, no?
   
    Today I soldered 10uF 0805 caps (did not have any 22uF) on top of the existing ones. Ever since I've had no screen flickering nor any crashes at 1280x800@50Hz. So really, this nails it so I thank you A LOT!
   
    Ps. So I currently have 2x10uF caps installed. What would be the recommendation? 2x22uF like datasheet says is typical configuration or even 2x47uF?

Can you make a picture of your modification cause I have the same problem and I can't imagine what to do.


Simo Koivukoski
(Apollo Team Member)
Posts 468
08 Apr 2017 20:02


Please before you consider any kind of mods for your boards, make sure that you a) have high-quality DIGITAL-VIDEO cable b) use DIGITAL-VIDEO amplifier c) use as low as possible pixel clock from picasso96mode.


Jari Eskelinen

Posts 20
08 Apr 2017 20:46


Satviewer G wrote:

  Can you make a picture of your modification cause I have the same problem and I can't imagine what to do.

Sure. Please see this picture:

EXTERNAL LINK 
So I soldered 10uF 0805 ceramic caps on top of the C3 and C6 which are output capacitors for voltage regulators. So now there are 2x10uF caps per voltage regulator in parallel. A bit dirty hack (and seems my hand were not too steady, caps are misaligned), but hey, gets the job done.

Relatively easy thing to do if you are proficient with solder iron. You just need some 0805 caps (or 0603), temperature controlled soldering iron, flux and IPA to clean up. On the other hand, if you are NOT experienced with SMD soldering, please do not do it. You can easily rip off C3 and C6 etc.


Carlos Milán

Posts 51
08 Apr 2017 20:48


For me it worked using a lower pixel clock.


Jari Eskelinen

Posts 20
08 Apr 2017 20:57


Simo Koivukoski wrote:

  Please before you consider any kind of mods for your boards, make sure that you a) have high-quality DIGITAL-VIDEO cable b) use DIGITAL-VIDEO amplifier c) use as low as possible pixel clock from picasso96mode.
 

 
  Say, if your Ferrari is overheating due to lack of coolant, would you like to drive "as slow as possible", or add coolant?
 
  Edit: Yeah I get it though. No hardware modifications should be done without required skills or actually knowledge about nature of the problem - if the problem is bad cable then soldering new caps are madness. That's why Apollo Team should take this seriously and just let users know if such mods are necessary under some circumstances.
 
  Personally I was getting desperate, system crashes after bad flickering etc. Somebody who seemed to have the clue said that output caps are too small, after checking datasheets they really are smaller than recommended by manufacturer then what the hell, let's add some caps. And hey, it worked!
 
 


Manfred Bergmann

Posts 125
14 May 2017 19:32


Hi.

I can confirm a very frequent black screen flickering at a resolution of 1024x768 where the pixel clock is at 47MHz.
Reducing the resolution to 800x600 pretty much solves the problem.
It's a little unfortunate because 1024x768 is a nice resolution to work with.

Manfred



Martin Soerensen

Posts 230
16 May 2017 12:37


Manfred Bergmann wrote:
I can confirm a very frequent black screen flickering at a resolution of 1024x768 where the pixel clock is at 47MHz.

Are you using an DIGITAL-VIDEO repeater? I used to have the same problem but after I started using an DIGITAL-VIDEO repeater I have not experienced it even once.


Manfred Bergmann

Posts 125
16 May 2017 20:46


Martin Soerensen wrote:

Manfred Bergmann wrote:
I can confirm a very frequent black screen flickering at a resolution of 1024x768 where the pixel clock is at 47MHz.

  Are you using an DIGITAL-VIDEO repeater? I used to have the same problem but after I started using an DIGITAL-VIDEO repeater I have not experienced it even once.

No. My cable is just 0.5m. Why would that need a repeater?

Manfred


Ian Parsons

Posts 194
16 May 2017 22:14


The Vampire output is not using a dedicated ASIC with DIGITAL-VIDEO compliance tested specs, it's an FPGA with a few resistors. It works with lowish frequency video clocks but with higher data rates it depends on how tolerant the input circuits of your TV/monitor are. A repeater is designed to be tolerant of waveform problems and clean them up to a "perfect" digital signal for the display.


Simo Koivukoski
(Apollo Team Member)
Posts 468
17 May 2017 05:45


Ian Parsons wrote:

The Vampire output is not using a dedicated ASIC with DIGITAL-VIDEO compliance tested specs, it's an FPGA with a few resistors. It works with lowish frequency video clocks but with higher data rates it depends on how tolerant the input circuits of your TV/monitor are. A repeater is designed to be tolerant of waveform problems and clean them up to a "perfect" digital signal for the display.
 

In fact, the Vampire does not have a hdmi, but digital video output.


Manfred Bergmann

Posts 125
17 May 2017 07:24


Ian Parsons wrote:

The Vampire output is not using a dedicated ASIC with DIGITAL-VIDEO compliance tested specs, it's an FPGA with a few resistors. It works with lowish frequency video clocks but with higher data rates it depends on how tolerant the input circuits of your TV/monitor are. A repeater is designed to be tolerant of waveform problems and clean them up to a "perfect" digital signal for the display.

OK. If it helps I'll buy one.
Looking at what Gunnar said:
Gunnar wrote:
In general we recommend you to use Screenmodes with Pixelclock up to 50 MHz maximum.
Screenmodes above 50 MHz Pixelcock can work but will
stress the video out of the Vampire to the maximum.

I'm wondering if 47MHz is long term acceptable or still too high. Thinking about the life expectance of the card (or the FPGA).

If acceptable I'd rather use a higher resolution than 800x600 using a repeater.

Manfred


Gunnar von Boehn
(Apollo Team Member)
Posts 4169
17 May 2017 08:55


Manfred Bergmann wrote:

  If acceptable I'd rather use a higher resolution than 800x600 using a repeater.
 
  Manfred
 

 
Hallo Manfred,
 
Which Monitor type do you use?
4:3 or 16:9 ?
How large in inches/zoll?

Many people have very good experience with 960x540 on 16:9 screens.
This res is very sharp and clean on many screens
and the text and WB-icon size is not to small to become hard to read.


Przemyslaw Tkaczyk

Posts 141
17 May 2017 11:19


Gunnar von Boehn wrote:

 
Manfred Bergmann wrote:

    If acceptable I'd rather use a higher resolution than 800x600 using a repeater.
   
    Manfred
   

   
  Hallo Manfred,
   
  Which Monitor type do you use?
  4:3 or 16:9 ?
  How large in inches/zoll?
 
  Many people have very good experience with 960x540 on 16:9 screens.
  This res is very sharp and clean on many screens
  and the text and WB-icon size is not to small to become hard to read.
 

 
I'm sorry to jump in like that - I know this reply is not meant for me, but I just wanted to let you guys know that I had the exact same problem after flashing GOLD2 - and I was just using 640x480 @ 60 - so not really an overkill for the SAGA I guess..?

After re-flashing back to GOLD1 the problem was gone. So is it really a hardware problem?


Nixus Minimax

Posts 341
17 May 2017 13:45


Manfred Bergmann wrote:
I'm wondering if 47MHz is long term acceptable or still too high. Thinking about the life expectance of the card (or the FPGA).

Driving too high a clock frequency doesn't hurt the FPGA, there is no need to worry about that. It's just that the pins of the FPGA aren't designed to drive such high frequencies and thus do a bad job at it. The digital image data is driven serially and there is some error coding on top of the payload which increases the total signal rate by another 25%. This leads to a very high bandwidth requirement which the pins can hardly provide (e.g. 47 MHz pixel clock x 8 bit colour depth per RGB component * 10/8 error coding = 470 MHz per base colour).

You may be lucky and a "1" will get recognised as a "1" by the monitor and a "0" as a "0" most of the time. If it is a good monitor, it will also apply error correction (that's what the extra 25% mentioned above are there for, after all), if it is a bad monitor it will not correct errors even though the error coding would put it into a position to do so. Then there is inevitable noise on all supply lines. The noise can change from one core to the other because it will depend on the internal placement of core components which are determined by a mostly random place'n'route algorithm (Monte Carlo). Thus, in some core version the digital video signal may be more prone to errors than in others.




Manfred Bergmann

Posts 125
17 May 2017 14:42


Gunnar von Boehn wrote:

Manfred Bergmann wrote:

  If acceptable I'd rather use a higher resolution than 800x600 using a repeater.
 
  Manfred
 

 
  Hallo Manfred,
 
  Which Monitor type do you use?
  4:3 or 16:9 ?
  How large in inches/zoll?
 
  Many people have very good experience with 960x540 on 16:9 screens.
  This res is very sharp and clean on many screens
  and the text and WB-icon size is not to small to become hard to read.

I'm using a 4:3 19" monitor.
But I have other which I can try to have the best image.

Manfred


Manfred Bergmann

Posts 125
17 May 2017 14:43


Nixus Minimax wrote:

Manfred Bergmann wrote:
I'm wondering if 47MHz is long term acceptable or still too high. Thinking about the life expectance of the card (or the FPGA).

 
  Driving too high a clock frequency doesn't hurt the FPGA, there is no need to worry about that. It's just that the pins of the FPGA aren't designed to drive such high frequencies and thus do a bad job at it. The digital image data is driven serially and there is some error coding on top of the payload which increases the total signal rate by another 25%. This leads to a very high bandwidth requirement which the pins can hardly provide (e.g. 47 MHz pixel clock x 8 bit colour depth per RGB component * 10/8 error coding = 470 MHz per base colour).
 
  You may be lucky and a "1" will get recognised as a "1" by the monitor and a "0" as a "0" most of the time. If it is a good monitor, it will also apply error correction (that's what the extra 25% mentioned above are there for, after all), if it is a bad monitor it will not correct errors even though the error coding would put it into a position to do so. Then there is inevitable noise on all supply lines. The noise can change from one core to the other because it will depend on the internal placement of core components which are determined by a mostly random place'n'route algorithm (Monte Carlo). Thus, in some core version the digital video signal may be more prone to errors than in others.

So that's where the HDMi repeater could do it's job if the display can't?

Manfred


Nixus Minimax

Posts 341
17 May 2017 14:54


Manfred Bergmann wrote:
So that's where the HDMi repeater could do it's job if the display can't?

The repeater provides a clean output signal from a dirty input signal and thus can help with display quality.



Martin Soerensen

Posts 230
17 May 2017 15:19


It is much like the Gayle IDE controller. It cannot drive long cables or multiple devices very well but if you put a buffer on it, it solves the problem since Gayle now only have to drive a short cable and the buffer inputs. If you increase the speed of the IDE port, the problem would be even worse (perhaps a buffer would also help with Apollo FastIDE stability since it does not work equally well on all A600s?).
 
So anyone having error problems with the Vampire DIGITAL-VIDEO port should use a repeater/buffer and a short as possible cable between the Vampire and repeater.


Manfred Bergmann

Posts 125
20 May 2017 19:19


Hi.

So, I have gotten a DIGITAL-VIDEO repeater. But it doesn't help much.
I'm still having black screens.
I've tried three displays, two different 19" 4:3 and one 22" 16:10.
All three work fine on various other Amigas with Indivision ECS or AGA.
I've tried resolutions 1024:768 with 47MHz pixel clock on the 4:3 screens, which gives the most black screens. 800:600 with pixel clock ~40MHz is more stable. Though there are black screens as well.
I'm not certain what to try next. Working with unstable screens is annoying.
I'm tempted to switch back to Indivision ECS screen. Though that would be a bit disappointing.

Manfred


Carlos Milán

Posts 51
21 May 2017 21:49


I have a 60W PSU for my A600 with a 1.5 rev motherboard. Lowering the pixel clock worked for me.

posts 98page  1 2 3 4 5