Pole Position Billboard/Car Display Issue

[Vraz]

I recently picked up an Atari Pole Position. It is in fairly rough shape, but runs. Self-test passes, sound is good and basic game plays. However, there is something wrong with video display of billboards and cars. Attached is an image for inspection. It appears that only every other video scanline is showing and the colors are off. The basic background, character generator, road and mountains all look fine (other than color calibration is slightly off).

I added the sense bypass to the ARIIs and adjusted them for 5.13v each. Reseated all socketed chips on the video board. Cleaned the edge connectors, removed the RFI shield board (didn't want an extra set of connectors in use), etc.

My biggest challenge is that I dont understand the underlying game well enough to determine which subsystem is responsible for this issue (motion, playfield, or picture). I assume its something in the picture subsystem, but any confirmation or other pointers would be appreciated before I start swapping chips.

Thanks!

 

[Vraz]

Just an update for anyone else who might run into similar issues.

My guess is that the problem is somewhere in the Picture Memory sub-system. My first try was the swap the 13H Custom 02 chip for the one at 8N Alphanumeric (which I knew was good). No change which indicates its not that chip.

Second test was to pull the 13H 02 chip and insert pullup/pulldown resistors in its place (force a value on its nibble output). Pulling pins 12,11,17,18 to ground removed all the problem areas (and the cars and signs). This confirmed that the Picture Memory subsystem is my problem area (though its still fairly large). Pulling the pins to +5v created the attached image. The "solid" areas indicate where the car and banners should be. It still has the "every other line" look which makes me suspect the problem is downstream from this chip. By creating different values (different combinations of the pins to +5/GND) I could get some different colors, but not much more useful information.

I would really like to override the output coming from the PROM at 12H, but that needs to wait until I get some proper desoldering gear. I am not that great at desoldering and given the age of the boards, I want to be gentle. Am ordering a Hakko 808 which should allow me to desolder and add sockets with relative easy (or so I hope).

One thing that would be useful is knowing how good the self-check is. Does passing the self-check really mean all the RAM/ROM is ok, or is it limited in what it can find? It would certainly be a time saver to rely on its results.

Thanks!

 

[Witchboard]

Good deal. I have a PP board with a similar if not exact issue. The self test is pretty good. I ended up replacing 6 RAM chips on one until it finally completed the diagnostic. Everytime I replaced a RAM the error would move to the next RAM chip.

 

[Vraz]

Which RAM chips did you end up replacing? Was it some of the Motion Object RAM at 7F/8F/7H/8H/7J/7K/7K/8K? Interstingly, I got a RAM 01 error earlier this evening (first time) during gameplay which indicates a RAM 7F error (where 7F/8F had been socketed/replaced by some former owner). I swapped the two chips to see if the display pattern changed, but no luck.

I will order some spare RAM tomorrow so that once I get my desoldering tools, I can swap it out.

Thanks!

 

[Witchboard]

I just replaced the chips that the self diagnostic stated were bad. I do not recall off the top of my head.

 

[Vraz]

Updating this thread in case others run into a similar issue.

Yesterday I received my Hakko-808 desoldering gun (like a soldering gun with a vacuum pump built in). Works incredibly well and after some practise, put it to work. Based on my previous experiment, my problem is related to the picture memory or motion line buffers so I desoldered the 12H PROM to determine which. Since that chip interfaces these sub-systems, its an easy test point.

With the PROM removed, the picture/car images are missing. This makes sense as the pullups create a 1111 pixel value which the hardware treats as transparent. I then shorted the resistors to ground to force different color combinations. The results were consistent with my prior testing (removing Custom 02 at 13H). The two main findings were that every other line is still missing and the color palette is off. By analyzing the palette bit pattern, I determined that MP2 (bit 1) is being ignored since changing it does not ever change the color.

I had been studying the schematic and was conflused by a couple points which I finally figured out (and other might already know, but just in case):

1- Notations like "P,R97" stand for "pullup resistor R97". I kept thinking these were signals but could not figure out the source.

2- What looks like signal "IV" is actually "1V" and thus the odd/even vertical line indicator.

At this point, I believe my problem is in the Motion Object Line Buffer (schematic sheet 15A). Seems like three potential issues:

The 74S158 @ 8D which is the even/odd control logic. While this could cause the missing line, I dont think it could cause the color palette problem.

The 74LS298 @ 9D which recombines the even/odd pixel data. This chip could potentially cause all the problems.

The 2148L-2 @ 9F/10F SRAM which is used as even/odd line buffers. To account for both problems, one SRAM would need to be completely bad and the other would need a dead bit. Seems strange they would both fail, but certainly possible). Since these SRAMs have no direct connections to the CPU, their failure would never be detected during self-test.

I have to wait for replacement parts before proceeding, but my plan is to swap LS298 @ 9D first. If that doesn't resolve, then will swap the SRAMs next. Will update when that is complete.

 

[Vraz]

While I am still waiting on parts, I decide to go ahead and remove and add a socket for the 74LS298 @ 9D. This is the output from the Motion Object Line Buffers which gets combined with the picture and character data to create the final image.

Though I didn't have a replacement part, I went ahead and added pullup/pulldown resistors on the output to force patterns to determine where the missing line is coming from and where the color problem is. Unexpectedly I found that a 4.7K pulldown resistor won't pull the outputs to ground. For anyone else trying to pull down signals for testing, it appears that 4.7K works as a pullup, but going direct to ground is the best bet for generating a logic low.

Once that was resolved, I was able to force a color. As expected, the display was solid confirming that the missing line is within the motion sub-system. However, it looks like the color palette problem is further downstream. The generated values simply did not make sense. The most likely explanation is that one of the 2148 SRAMs is bad causing the missing lines and there is an unrelated color problem. I have also noticed that the road looks more black than gray which is likely the same issue.

 

[Vraz]

Finally received replacement parts and removed the old 2148s at 9F/10F, added sockets and added new SRAM. The missing line problem appears to be resolved though I still have a palette problem (certain colors draw wrong for cars/signs).

Thanks to the sockets, I was able to pull each SRAM individually to see which has the palette problem. Turns out its something in the flow of 10F. It appears there are only a couple of possible suspects. Either there is an electrical problem (cold solder joint, etc) between 10E, 10F or 9D or 10E is bad. Since there are only four data bits, it wont be hard to check all the paths with a meter tomorrow. If it appears to be good electrically, then I will remove 10E, add a socket and replace with a new part. With any luck, my adventure should be drawing to a close.

 

[Vraz]

My PP1 board has been repaired and the video display now looks good. To resolve the ugly looking image attached to the first message, it was a two-part fix:

1- The high-speed 2148 (or 2149 depending on the board) line buffer SRAM at 9F was bad. Replacing it eliminated the every-other blank line problem.

2- I had a bad solder joint on the LS257 at 10E. Removed the old chip, added a socket and tried the old chip in the socket and it worked. That was causing the color palette problem in every other line.

While doing this work, I also took the opportunity to remove the old battery which was surprisingly still working and had not leaked. With my PP1 & PP2 boards both working properly, the final project is to deal with high score backup. Will probably just use a 1.5F supercap and see how that works.

 

[Phetishboy]

Well that was one hell of a thorough job you did my man. If you are ever up to looking at Pole boards on the side for others, you will be a popular guy. I may have one for you to look at if you're interested. Sounds like you have the gift.

 

[Vraz]

Right now time is really tight for me and I just do this for fun. Am happy to offer advise to others trying to fix their PP boards, but not ready to get into the repair business myself. I have seen a couple places offering to do PP repairs out there (though they are not particularly cheap).

The biggest challenge I encountered fixing mine was getting replacement parts (I didn't have other boards to salvage from) and getting the equipment to desolder successfully. My removal of parts and replacement with sockets was certainly not the cleanest job out there (even with the desoldering gun), but functional. If you are willing to sacrifice the part, then its not too bad (just cut all the leads and desolder the pins one at a time). The problem is that since at least a few guesses will be wrong and some replacement parts don't work, you can actually dig yourself a bigger hole this way.

 

[Space_cowboy]

My PP1 board has been repaired and the video display now looks good. To resolve the ugly looking image attached to the first message, it was a two-part fix:

1- The high-speed 2148 (or 2149 depending on the board) line buffer SRAM at 9F was bad. Replacing it eliminated the every-other blank line problem.

2- I had a bad solder joint on the LS257 at 10E. Removed the old chip, added a socket and tried the old chip in the socket and it worked. That was causing the color palette problem in every other line.

While doing this work, I also took the opportunity to remove the old battery which was surprisingly still working and had not leaked. With my PP1 & PP2 boards both working properly, the final project is to deal with high score backup. Will probably just use a 1.5F supercap and see how that works.

or go with an off board battery solution

 

[Vraz]

>> or go with an off board battery solution

I considered that as well as I have a bunch of AAA NiMH around from a prior project that would probably work. The simplicity of a single 1.5F SuperCap soldered to the board has a certain appeal. I will order battery holders at the same time as the SuperCaps and see which works better.

 

[Mikebetz42]

>> or go with an off board battery solution

I considered that as well as I have a bunch of AAA NiMH around from a prior project that would probably work. The simplicity of a single 1.5F SuperCap soldered to the board has a certain appeal. I will order battery holders at the same time as the SuperCaps and see which works better.

Wasn't there a way to do it with a Dallas or Simtek chip and skip the battery all together? I searched for the post, think Mark talked about this one.......

-Mike

 

[Vraz]

>> Wasn't there a way to do it with a Dallas or Simtek chip and skip the battery all together? I searched for the post, think Mark talked about this one.......

Yep. You can replace the 6116 with an NVRAM compatible chip. The Simtek is the coolest as it uses a unique "write to flash-like memory at power-down and reload into ram from flash-like memory at power-up" system. It is supposed to last for 100+ years or something crazy and there is no battery of any kind in it. Both Dallas and ST make versions which have an integrated lithium cell. So its essentially a 6116 with active backup and they last around 10 years. The Simtek is no longer available and the Dallas/ST run $10-15 each.

The SuperCap is different in that it would have a more limited time between charges (am guessing around 3+ months at 1.5F) but still has a relatively long life (15+ years). Like other capacitors, its charge capacity will diminish somewhat over time but they have a long rated life. At $4/ea, they are the least expensive solution and unlike a remote battery pack, you can mount it right on the board.