Williams Buffer Overflow/dot FAIL - RESOLVED

[PaulV]

Have a new (to me) Williams Dotmation Jackpot Stampede. Worked fine for a few weeks, now when I power on, right after the initial dot INIT and prom check, I get a dot FAIL on the led panels and on the DOTMATION screen is displayed Buffer Overflow and beneath it a hex address which updates every few seconds, like 0x0055, then 0x009A, etc. I found that if I leave the machine powered on then after about 10 minutes of warmup, if I shut it off and back on right away (with a few seconds delay between power off then power on) everything works fine. This sequence, if done in less than 10 minutes or so, continues to fail. I'm guessing power supply due to the dot FAIL, but not sure what area a Buffer Overflow is in. Can't find any loose cables etc.

Any ideas what a Buffer Overflow could mean?

SHORT VERSION OF RESOLUTION: Bad capacitor as part of RS232 interface chip circuit on main (lower) CPU board. Technical details in another reply below

[Slotmaster]

I have seen this in the past and the issue was the DOT EPROMS.    The topbox board does not do a checksum check like the CPU does. 

I think the XU4 was corrupt or something like that.  Was about 3 or 4 years ago as I no longer work on them and that is what I remember from the top of my head.

[PaulV]

Thanks Slotmaster,

I pulled the eproms (from both CPU and Dotmation Controller boards) and read them all, verifying the checksums, they all matched what was encoded on the eprom labels. Here's a more detailed history of this problem:

1. Machine worked fine for several weeks after purchase. It's my first slot so not knowing whether it should be left on all the time or turned off after use, I was turning it off at the power supply at the end of each evenings use. What's the recommendation: Leave it on 24/7 or turn it off daily? It is powered thru a heavy duty surge protector.

2. After several weeks of no problems, I turned it on one morning and got the dot FAIL message on the LED's and the Buffer Overflow on the Dotmation panel. I turned the machine off right away and powered right back on. It powered up normally at the point.

3. From day to day, it got worse, a longer delay between the time I would first power up and get the error, until it would power on OK. Finally, the other day it reached a point where it will not completely power up, it's stuck on the dot FAIL, Buffer Overflow condition no matter how long I keep the power on and attempt to reset.

Based on recommendations I first checked that all cables were fully seated by pulling and the reseating them. I then measured and scoped the upper powersupply, I have a good clean waveform with no ripple and it holds steady at +12 volts under load. Even so, I pulled the upper PS and replaced all the caps. Still failed. So then I used an ATX power supply and made the conversion, got similar voltage readings and still the same error. Put the original WMS upper supply back in for now.

Next check was eproms, all checksums match what's on the labels. Checked all boards to see that all socketed chips/cables were firmly seated.

Still no change, same error. 

Now have started performing a logical analysis of the problem. Taking into account that the LED error is dot FAIL, it appears the problem is the CPU board is not communicating with the Dotmation Controller board via the serial port. The indicators make it difficult to determine which board has the problem communicating. I'm now leaning towards the Buffer Overflow being an indication that either the incoming serial data stream from the CPU board is not getting processed by the Dotmation Controller, and is overflowing the input buffer, or that the outgoing data stream to the CPU board from the Dotmation Controller is not being transmitted, causing the output buffer to overflow. I see the processor on the Dotmation board is an MC68HC11 and I have a disassembler for that, so I am going to disassemble the code  and determine what conditions can cause the Buffer Overflow message to be displayed (I do see the message, along with many others, coded in the U10 eprom). This should help me track down the cause. Once, I caught a glimpse of a different message on the Dotmation display, it went from Buffer Overflow to what looked like Bad Packet and then flashed back to Buffer Overflow, I only saw this occur once although of course it may be happening so fast I don't catch it visually. I also see the message Bad Packet in the U10 eprom. I'm going to also try to stick a logic probe on the serial line and see if data is actually being transmitted/received. That may help me isolate it to the CPU or Dotmation Controller board.

If I had spare parts, I'd just swap out the Dotmation Controller board as that is probably the more likely place for the error to be. Is there anyone on here with a spare Dotmation Controller board to sell?

I'm certainly open to any and all comments/recommendations/suggestions!

[Slotmaster]

I think I have one to sell you, I will check later today.

However first change the topbox eproms (all three) as a first try.  I am almost sure that is the issue even if the checksum is correct. 

Second change the topbox comm board (I will check today to see if I have one)

Also reseat the comm cable that goes between the I/O and the dot board.

Turn the machine off when done, you will not have an easy time finding the top DOT Matrix display and they gas out over time when left on.   You will get DIM spots and dark areas and the display will get darker and darker overtime.   

 

[StatFreak]

Once you get your machine running, you can also turn the top box off while leaving the machine on by turning the jackpot reset key and entering bookkeeping mode. These machines do not time out and return to idle mode like IGTs do, so it will stay in bookkeeping mode until you choose to exit, and the dot screen will be powered down during that time.

[Neonkiss]

You may also consider installing a 110V box fan in the top dotmation unit. I do this with all my machines.
The original design criteria for the top unit power supply from the data sheet reads,

"The PU110 series of compact, open PCB constructed,
AC-DC switching power supplies are capable of delivering
72 to 110 watts of continuous power at 25 CFM forced air
cooling or 52 to 80 watts at convection cooling"

Just using the convection cooling design for this unit is not enough. Also by using a 110V fan your not pulling power from an already overtaxed power supply.
This airflow provides additional benefits to the dotmation control board and those hard to find dotmation displays.

But as Statfreak mentioned, turning off the dotmation display when not in use is the best advise.

[StatFreak]

...

But as Statfreak mentioned, turning off the dotmation display when not in use is the best advise.

To be fair, Slotmaster said it first. I was only pointing out an alternative method of preserving the dot box.

[Buzz]

Just using the convection cooling design for this unit is not enough. Also by using a 110V fan your not pulling power from an already overtaxed power supply.
This airflow provides additional benefits to the dotmation control board and those hard to find dotmation displays.

But as Statfreak mentioned, turning off the dotmation display when not in use is the best advise.

Neon  another thing I do is move the upper power supply from inside the Dot. box to the back side of the box using the same screw holes. All the wiring is long enough so no mods needed to them. you do have to shim the power supply away from the metal box ( I use washers on the screws ) I look at like taking the PS out of a oven and putting it in fresh air.

[PaulV]

Thanks for all the comments,

Unfortunately I only have the one williams machine so no spare dotmation eproms to swap out on the controller.

I've been going thru the disassembled code and it does look like there is a test to see if either the input or output SCI buffers located on the 68HC11 CPU get full and if so it will trigger the buffer overflow message. Not positive about that, still going thru the code, but it is a strong possibility. Assuming the 68HC11 itself is OK, then it may be the LTC1384 RS232 chip is bad, preventing communications between the CPU and Dotmation boards, if so that would be a reason for the output buffer to fill up. I've checked and double checked the comm cable between the Dotmation controller and CPU backplane several times, it looks good. The LTC1384's not a socketed chip but it's out in the open where I can remove and swap it easy. I'll put in a socket if I end up removing it. I'll put the logic analyser on the input/output first to see if the chip is active before swapping it out.

I think I'll just go with a complete power off of the machine when done with it for the day. Once the newness has worn off won't be having it on all the time anyway.

[PaulV]

I believe this problem has been resolved, my machine is working fine now, powering up on the first try.

SHORT VERSION OF RESOLUTION: Bad capacitor as part of RS232 interface chip circuit on main (lower) CPU board

TECHNICAL DETAILS, FOR THOSE DIEHARDS INTERESTED: As mentioned previously, I had used a logic analyzer on the serial lines and decided that the Dot Controller was attempting to send data packets on the serial line to the main CPU but the main CPU wasn't responding, so the Dot Controller transmit buffer was filling up, causing the Buffer Overflow message (determined from disassembling the Dot Controller eprom). Wasn't able to actually get to the main CPU board to scope it during operation due to its being in the card cage, but tracing out the land patterns with the card removed, and focusing on the HIN232CP chip, which converts the 5 volt level to the 10 volts needed for serial communication with the Dot Controller, I found 4 capacitors used with the HIN232CP that are part of the charge pump circuit. This circuit is used to convert the 5 volt input to the HIN232CP chip, to the 10 volts output needed to go to the Dotmation Controller serial port. Figured that if one or more of these caps were bad, the charge pump wouldn't work, in effect shutting the serial port down. Unsoldered and pulled the 4 caps (they were standard 10uf 16v 85-degree caps) and checked each one on my cap tester (Extech EX530 DMM, I'd be lost without this thing) and found 3 caps were between 10.2uf and 10.6uf. However one cap was 168.7uf, way, way out of tolerance. Figured that was so far out of spec it was changing the timing of the charge pump to the point where it couldn't activate the HIN232CP chip properly, and resulted in non-funtion of the serial port. I figure that it was slowly creeping up in value over time, which is why my machine would work occasionally at first, and then when the cap got far enough out of spec, it just quit.

End result is, I had four 10uf 16v 105-degree (high temperature/commercial use) caps in my junque box and replaced the 4 original caps with them. Reinstalled the main CPU board, powered up and the machine came right up. Went thru several power off/power on cycles for an hour or so and every thing seems to be functioning fine.

My thanks to all who offered suggestions, advice and support. I've learned a lot about this machine over the last few days. It would be easier if there were a set of schematics for the machine though, these old eyes were really straining to follow some of the land patterns on those boards to figure out what was going where!

[Slotmaster]

I believe this problem has been resolved, my machine is working fine now, powering up on the first try.

SHORT VERSION OF RESOLUTION: Bad capacitor as part of RS232 interface chip circuit on main (lower) CPU board

TECHNICAL DETAILS, FOR THOSE DIEHARDS INTERESTED: As mentioned previously, I had used a logic analyzer on the serial lines and decided that the Dot Controller was attempting to send data packets on the serial line to the main CPU but the main CPU wasn't responding, so the Dot Controller transmit buffer was filling up, causing the Buffer Overflow message (determined from disassembling the Dot Controller eprom). Wasn't able to actually get to the main CPU board to scope it during operation due to its being in the card cage, but tracing out the land patterns with the card removed, and focusing on the HIN232CP chip, which converts the 5 volt level to the 10 volts needed for serial communication with the Dot Controller, I found 4 capacitors used with the HIN232CP that are part of the charge pump circuit. This circuit is used to convert the 5 volt input to the HIN232CP chip, to the 10 volts output needed to go to the Dotmation Controller serial port. Figured that if one or more of these caps were bad, the charge pump wouldn't work, in effect shutting the serial port down. Unsoldered and pulled the 4 caps (they were standard 10uf 16v 85-degree caps) and checked each one on my cap tester (Extech EX530 DMM, I'd be lost without this thing) and found 3 caps were between 10.2uf and 10.6uf. However one cap was 168.7uf, way, way out of tolerance. Figured that was so far out of spec it was changing the timing of the charge pump to the point where it couldn't activate the HIN232CP chip properly, and resulted in non-funtion of the serial port. I figure that it was slowly creeping up in value over time, which is why my machine would work occasionally at first, and then when the cap got far enough out of spec, it just quit.

End result is, I had four 10uf 16v 105-degree (high temperature/commercial use) caps in my junque box and replaced the 4 original caps with them. Reinstalled the main CPU board, powered up and the machine came right up. Went thru several power off/power on cycles for an hour or so and every thing seems to be functioning fine.

My thanks to all who offered suggestions, advice and support. I've learned a lot about this machine over the last few days. It would be easier if there were a set of schematics for the machine though, these old eyes were really straining to follow some of the land patterns on those boards to figure out what was going where!

Good job!

We always just replaced the boards first before having the time to fix them.     Only if they are hard to find do we attempt to repair them, in this case the WMS Reel CPU was in no shortage or the DOT controller.

I sold about 20 - 30 CPU's to CFH who is on the board about a year ago including controllers too. 



It's been about 4 years since I have worked on a DOT and was going off memory.   

[PaulV]

Yes, swapping cards would have been my first choice before actually attempting a board repair. Unfortunately, I have no spares. I'd like to pick up one of each board, CPU, I/O and Backplane, I did find a Dotmation Controller board and have it on order. This way I'd have something for shotgunning a problem if anything occurs in the future.

Thanks for your help!

[Neonkiss]

Great work PaulV  
and some K+ to you for posting your resolution.

Always nice to hear back when a machine is up and running again along with the fix.
Enjoy your machine, you can't sell it now. You have too much time invested in it.

[StatFreak]

Excellent diagnostic work, and K+ from me as well for sharing the details with our members.

SF

[knagl]

K+ from me as well -- spectacular detail, great diagnosis, and it seems like it was right on track!  I could never do board-level repair, so I plan to send my broken boards to you.   

Congrats on getting your game up and running -- now go enjoy it!