• Please review our updated Terms and Rules here

CGA Snow

fatwizard

Veteran Member
Joined
Sep 28, 2012
Messages
525
Location
Indiana, USA
I know CGA snow has been covered on the forum before, but this is a little different. I have a 5160, 64 to 256K motherboard that I use on the bench for testing purposes. I took it out this week to test some hard drives, and I now have snow in the CGA display all of the time. It's there as soon as the display initializes.

I was using an original IBM CGA card, so I switched to an old Hercules CGA card I have and the constant snow remains. An MDA card, or a VGA card have no snow. It wasn't like this the last time I used it. Landmark diagnostic ROM found no issues. IBM advanced diagnostics didn't find anything either. When I run the memory test on the IBM diagnostics, you can see the patterns in the snow changing as the test progresses. I am kind of stumped on where to look for the problem
 
VGA is so much easier on the eyes I can't imagine why you would even consider CGA as an option for doing HD bench work. :)
 
Originally I setup an old Tandy CGA monitor to test my collection of IBM CGA cards, and the test jig has just remained that way. I don't have to stare at the display very long to run these kind of tests, plus this way I don't have to dig out a CGA or monochrome monitor to use the Landmark ROM.

My real concern is understanding what happened to the motherboard to cause this.
 
I know CGA snow has been covered on the forum before, but this is a little different. I have a 5160, 64 to 256K motherboard that I use on the bench for testing purposes. I took it out this week to test some hard drives, and I now have snow in the CGA display all of the time. It's there as soon as the display initializes.

I was using an original IBM CGA card, so I switched to an old Hercules CGA card I have and the constant snow remains. An MDA card, or a VGA card have no snow.
Is it really as soon as anything starts displaying or is it as soon as an OS loads?

Snow normally only happens if something is accessing video RAM without taking early CGA cards in to consideration. My first thought would be a TSR of some kind, for example an on-screen clock.

If it starts before the OS, then are there any odd ROM BIOS extensions in the machine?

Did the snow appear while running the landmark diagnostic ROM? That would be really weird.
 
Yeah, this isn't normal. The snow is there as soon as the video card initializes and is more or less constant. I was using an MDA card when I ran the Landmark test, so I just repeated the test with CGA and the snow is there. The rig is stripped down to just power supply, motherboard and video card. The board appears to be completely stock. I don't have another 5160 board loose to test, but all of my CGA cards work fine in my 5150 motherboard.
 
Yeah, this isn't normal. The snow is there as soon as the video card initializes and is more or less constant. I was using an MDA card when I ran the Landmark test, so I just repeated the test with CGA and the snow is there.

If that testprogram is constantly writing to the screen(memory) without checking for retrace then you will have snow on CGA
 
If that testprogram is constantly writing to the screen(memory) without checking for retrace then you will have snow on CGA

No, there is snow when the system first puts a cursor on the screen. With the original ROM or Landmark. Software isn't involved. I have been fiddling around with it tonight and have discovered that putting the CGA card in the slots nearer the center of the board, the worse the symptoms become. When I put it in the 7th slot I get big static purple dots and garbage in the memory count display. If I put a memory expansion card in the system I get 1 long and 2 short beeps (video), but the screen still comes up, but with big dots and garbage.

The VGA card works fine in any slot (leaving slot 8 out).
 
Very interesting. It seems you are running in to the true analog nature of digital electronics. Such problems can be hard to track down because things still mostly "work". It may be some voltage in/out of a chip that is a bit off or clock timing that is just a hair out of whack.

Can you further test the card more in that 7th slot? If it is actually corrupting the memory in there, perhaps a clearer pattern will emerge.

I suspect if you run a really aggressive memory test on the VGA card you might also see some memory errors.

Have you tried a different power supply just to rule that out?
 
No, there is snow when the system first puts a cursor on the screen. With the original ROM or Landmark. Software isn't involved. I have been fiddling around with it tonight and have discovered that putting the CGA card in the slots nearer the center of the board, the worse the symptoms become. When I put it in the 7th slot I get big static purple dots and garbage in the memory count display. If I put a memory expansion card in the system I get 1 long and 2 short beeps (video), but the screen still comes up, but with big dots and garbage.

The VGA card works fine in any slot (leaving slot 8 out).

Occam's Razor: Your CGA card's RAM is bad, or going bad.
 
"I was using an original IBM CGA card, so I switched to an old Hercules CGA card I have and the constant snow remains." Apologies, I misunderstood the issue -- the real issue is the board, not the video card.

If your board is constantly producing CGA snow, it is constantly accessing the RAM on the card (reads can do this as well). A short, perhaps.

Video of what you're seeing might jog some suggestions loose.
 
This video has 3 startups. The first with the Hercules CGA card in slot 1. 2 is with the same card in slot 7. 3 is the same card in slot 7 with a RAM expansion card installed. In the last scenario the system beeps a video adapter error code, but the display still comes up. I swapped power supplies too.

The board doesn't act right with any setup though. I wanted to try Checkit, so I setup for VGA with the RAM expansion and a hard card. With the RAM expansion in, the system wouldn't finish booting up. It would get to "Starting MSDOS", then just hang. A short? Yeah, there's a bad potato in this bin somewhere.

https://www.youtube.com/watch?v=T_JfvQUbmuU&feature=youtu.be
 
Congrats, you've managed to come up with something I've never seen before working with this hardware over 3 decades. Someone needs to get more electronics people into this thread to try to explain what's going on.
 
If you slow down the first part it looks like you found a hole in The Matrix :)

To me that doesn't look like "snow" snow. There is certainly a weak/floating logic line somewhere, and it seems as there is more load/distance on the ISA bus it gets worse. On the last two, at least there is a pattern to work with. It seems to place garbage characters every other 32 characters (64 bytes) and then the garbage appears every two characters (4 bytes) but is not specific to bit.

I'm just guessing we are probably looking at write/read operations that are intended for a different part of memory.

My wild guess would be a fault with one or more address lines on the ISA bus. Probably in the buffer chips.

It would be best if there were some test that could narrow it down further, but I don't know what specifically.

You might try "piggybacking" a known good chip on top of some of the ISA interface buffer chips. In the case of a bad floating logic output that might show the culprit.
 
Back
Top