Announcement

Collapse
No announcement yet.

Out Run Repair Log

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Out Run Repair Log

    Had @WOKA's Outrun board on the bench this month, he had bought a complete cabinet some time ago, that worked fine except there was no sound. The seller had told him that the fault was the game board, so it ended up heading south for some TLC.



    Outrun is a pain in the arse to power up on the bench as it is a long way from JAMMA standard, and there are a number of boards that sit external to the main PCB, a power filter board, dashboard control board (steering, gears, buttons and lights), the brake/accelerator assembly, and an audio power amplifier. Basically all I have from this lot is the "CPU BD" and the VIDEO BD"



    So first job was to cable the board up on the bench to the point where I could power it up. At some point in the boards history someone had bodged the power wires onto the solder side pins of the 10 Pin AMP connectors labelled J and K, possibly it is an LAI original bodge. It actually made things easier though, as the original connectors are not something I have any of, but I could easily connect a new bodge to the old bodge and not feel any guilt for being bodgy.



    Video output is handled via pin header H on the lower PCB...



    ...and once that was hooked up it would boot and go into attract mode, but at this stage I was out of connectors that would let me hook up the audio neatly.



    My only option was to hook in my audio probe and headphones to see if I could hear anything.



    As expected, I got silence except for a slight chirping noise briefly. At this stage I fired up the scope and started poking around, there was no detectable signal going to the LM324 final stage op-amp, certainly nothing over the background signal noise. On the scope the Z80 had clock, and HALT and ENABLE were ok, but the signals on the Data bus didnt look clean enough to be perfect, but also didn't look trashed enough to be a dead set fault. Chief suspect was the TMM2015 SRAM at IC107, these ICs are a common failure and it that chip had failed it would break all the audio as it is the work RAM for the Z80 CPU. These black shiny Toshiba SRAMs are a pretty common failure so this was where I would have placed my bet, especially as I briefly got some weird chirping noises which you often do get when an audio subsystem has crashed.

    So as the Z80 was socketed the easiest option was to wheel out the Fluke 9010 and a Z80 pod.



    This plugs into the Z80 socket on the board and pretends to be a Z80 CPU, except it allows you to take control and run tests that are built into the 9010, rather than just crash on the 1st fault as a real Z80 would. This lets you check that the RAM and ROM are addressable, writable and usable from the perspective of a CPU in the socket on the board, much more conclusively than you would get by desoldering the parts and testing them in isolation.



    All you need to know is what the memory map of the Z80 on this board should be, a Z80 has 64KB of directly addressable space and the RAM and program code ROM will sit somewhere within that 64KB. This is where the MAME drivers come in very handy, as the driver file for the game will contain the information you need as below.

    From here ---> [IMG]https://github.com/groovybits/groovymame/blob/master/src/mame/drivers/segaorun.c

    /*************************************
    *
    * Sound CPU memory handlers
    *
    *************************************/

    static ADDRESS_MAP_START( sound_map, AS_PROGRAM, 8 )
    ADDRESS_MAP_UNMAP_HIGH
    AM_RANGE(0x0000, 0xefff) AM_ROM
    AM_RANGE(0xf000, 0xf0ff) AM_MIRROR(0x0700) AM_DEVREADWRITE("pcm", sega_pcm_r, sega_pcm_w)
    AM_RANGE(0xf800, 0xffff) AM_RAM
    ADDRESS_MAP_END


    In this case the Z80's work RAM lives between F800 and FFFF, which is 2048 bits of address space, ie 2KB, the exact capacity of the TMM2015, so by read/write testing that region you are hitting the full contents of that chip directly, as well as proving the read/write and output control system of the board is working.



    And it tested fine with the quick test, but also passed the long in-depth test.



    While I had the Fluke connected I pulled the ROM at 16F and dumped its contents, which verified as being the correct file for Outrun's audio CPU. With a known good dump I put that through the CalcSig application that creates the Fluke's signature checksum (F126 in this case), and with the ROM back on board I could confirm that the Z80 could access the data from the ROM chip, which proves the ROM, the bus, and the output enable controls on the board, and also proves that ROM chip is behaving.



    One slight trap for the unwary here is that the ROM only uses half the address space allocated for "ROM" in the memory map, as the board was designed to take a larger data set game at manufacture. You can see the jumpers (actually zero ohm resistors, noted by the single black band) that strap the board for either 27256 or 27512 ROMs, just to the left of the 16F ROM. They use jumpers that are built like resistors as it is easier for the pick and place machine that assembles the boards to treat them like any other resistor, instead of another part type.



    So to get a confirmed signature I had to test the lower half of address block the MAME driver states, with that confirmed the only thing really left to do was to put the Z80 into another board to test that, and it worked fine.

    One jaycar trip later and I could make up a proper audio lead so I could plug the audio output into an external amp, but the result was the same, silence. At this point I poked around for a while with the scope looking for where the audio died in the signal path.

    Attract mode audio on Outrun is mostly just overtaking noises and not much else, so for a full test you really need to coin up and start the game.

    Except I couldn't, none of the inputs were working. On this board the coin, start and other inputs connect via the 50 way edge connector but the actual input pins were all floating. A quick buzz around with the meter showed why, none of the resistor pull up arrays or any of the opto-isolators downstream had a 5 volt feed. They all take their power from the filter board, which I didnt have. I am guessing Sega did it this way to make sure the board was immune from the piezo clicker/lighter wielding kids trying to get free credits. With a few jumper wires hooked onto the right pins on the 50 Way connected to 5 volts I could then at least coin up the board.

    At which point I could very very very faintly hear the "Get Ready" and the count down beeps for the start of the race, with the volume on the external amp set to the max it was barely audible amongst the hum. The overtaking noises in attract mode are basically white noise so its not surprising they didn't stand out against the background noise.

    With the game sfx and music playing it was easier to watch signals in the audio path and I could see them hitting the final stage op-amp in the pre-amp stage before the output connector, but not leaving the op-amp output. So I took the obvious step of desoldering the LM324 at IC139...





    ...fitting a socket and a new LM324, at which point there was no change at all, exactly the same fault.

    The original LM324 now lives back on the board socketed, it's probably not a bad thing, this IC had been replaced before, or at least desoldered, and I am assuming it wasn't someone else trying to track down the same fault in the past.

    A closer look at the schematics would have helped to show my logic implicating the LM324 wasn't clear cut, as there is a 4066 analogue switch straddling the LM324 op-amps for the left and right channel, with the control line darting off to a hand-written "mute 7/7" label.

    The diagram below is one I have cleaned up a bit, with some of the pin labels re-done, as well as the dirty great SEGA secret squirrel stamp removed from the P connector outputs.



    Basically the 4066 is a chip that contains 4 switches, on the pin 3/4/5 element, when pin 5 is pulled high, pins 3 and 4 are connected together. This connects the output of the op-amp to its own non-inverting input pin, by-passing the resistor R22 and capacitor C100. When you do this to an op-amp you render it useless as an audio amplifier as the resistor is there to tame the insane levels of gain to something useful, and in this case as the inverting pin is tied to audio ground (AG) of the IC110 and IC111, the op-amp will try to drive the non-inverting pin to the same level using the output, so both inputs and the output go to ground - everything nice and quiet.

    Except not quite full-on mute in the case of this fault, as while poking around I could find that mute it further and make the very low audio vanish entirely by shorting pins 3 and 4 of the 4066 together, effectively bypassing the analogue (simulating it being switched fully on), and thereby fully disabling the op-amp.

    The 4066 could have been a suspect expect for the fact that grounding the control pin 5 correctly disabled the switch and disabled the amp muting function, the result being my test speakers nearly went through the wall backwards, the volume still being set to max.

    So the fault was with the mute signal going to the 4066 on pins 5 and 6, which on the schematic page "7 of 7" for the CPU board is on pin 47 of connector C, a quick buzz through with the multimeter showed this was connected. From there is is picked up by the same pin on the other end of the board interconnect and is fed from the output of a ULN2003 IC at IC122 on pin 11.

    So time to separate the boards and see what's what.



    Eurgh - mouse piss and shit is what's what! Finding dead insects and larvae cases in boards is pretty common, finding this on a working board is thankfully not. With the lumps picked off in the garden I could track the signal through to the right pin on IC122.



    According to the datasheet these are high voltage, high current darlington transistor arrays, 7 of 'em, in one chip. They would usually be used for switching high voltage, or high current loads on and off with a single low voltage input signal, which in the case of Outrun makes sense as the stand up cabinet has steering shaker motors and the sit-down motion cabinet has hydraulic pumps, and they all suck lots of amps, more than enough to blow TTL logic ICs sky high.

    Using a ULN2003 to switch a low voltage, minute current, mute signal is hugely overkill but I guess there was a spare output on IC122 so it made sense to use it.

    I've tidied up the relevant section, and filled in some of the details about what the other ULN2003 pins are for.



    The ULN2003 takes it's inputs from an LS273 octal flip flop, at IC121, and the LS273 output pin 9 was held high, this signal should turn on the transistors in the darlington pair at pin 6 on the ULN2003 and thereby connect pin 11 to the emitter common, which in this case is ground. Basically these chips are usually used to complete a DC circuit by connecting the DC ground when required. The input signal to pin 6 was at 5 volts, clean and continuous, which means the pin 11 should be tied to ground, but it was actually hovering around 1.7volts and was very noisy.

    I was obviously in the right place as when the scope probe was on the output pin the chip would have a fit, causing the output signal to pulse high and low, repeatedly muting and un-muting the audio, very odd.



    It's not unusual to find a fault is unstable when you have tracked it right down to the pin and start poking about.

    So out it came...



    ... and one from a track-rot write-off Sega System 16 motherboard...



    ...was soldered in...



    ...to fix the fault.



    Or at least the fault on the game PCB, as I am slightly suspicious as to what caused the issue. It is uncommon for a board to actually mute the audio like this, and in fact Outrun has two mute systems, the 4066 to disable the pre-amps, and another mute signal that goes to the power amplifier board elsewhere in the cabinet. I assume they did this to stop the amps pumping heat when in silent attract mode, and also probably to remove any hum when the cabinet is supposed to be quiet. It is just a bit suspicious that the only three functions on the bad ULN2003 are the on-board mute for the pre-amps, the bulb in the start switch, and the mute signal for the power amp.

    The wires to the volume control in the cabinet had all broken off and the potentiometer itself was pretty mangled. Woka has fixed that and confirmed the amp board is getting 12V but there is no sign of life from the amp, no pop from the speakers and no hum that can be used to confirm the amplifier is actually working and responding to the volume control. There are some more tests to do to see if the amp has any life in it, but I am wondering if a dead power amp and a blown chip that controls the muting both the pre-amps and off-board power amp is more than just a coincidence.

    Then again it could just be down to the mice!
    Last edited by Womble; 7th October 2017, 07:22 PM.
    Sic transit gloria Atari!

  • #2
    Great to read your process. Thanks very much.
    This is where it ends!

    Comment


    • #3
      Love your work mate, always an informative read.

      Comment


      • #4
        Great detailed repair log. How long does something like this take you?

        Love the retro test gear too

        I have a working Sega Hang-on cab so probably similar.

        Thanks.

        - - - Updated - - -

        I went googling for more info on the 9010a and found this https://www.aussiearcade.com/showthr...62-fluke-9010a

        Comment


        • #5
          Always good to see someone able to work on an Outrun board and bring it back to life [emoji106]

          Comment


          • #6
            Always enjoy your write ups mate - great thread and thanks for sharing!
            Eschew the standard. Turn the paradigm upside down.

            Comment


            • #7
              Originally posted by Claude View Post
              Great detailed repair log. How long does something like this take you?

              Love the retro test gear too

              I have a working Sega Hang-on cab so probably similar.

              Thanks.

              - - - Updated - - -

              I went googling for more info on the 9010a and found this https://www.aussiearcade.com/showthr...62-fluke-9010a
              It really depends on the board and the fault, this one was probably about 7 or 8 hours all up, spread over a couple of weeks.

              Yeh that's the Fluke 9010 and the Z80 pod I used in this repair. It came with a Z80 pod and a pretty useless 8085 pod, but I have since added a 6502 , a 68000 and a 6809 pod to the fleet, basically the full set for repairing arcade stuff.
              Sic transit gloria Atari!

              Comment


              • #8
                What a great read.
                I love how the schematics for the IC say TOP SECRET!

                Keep up the great work.

                Comment


                • #9
                  Nice, been away for a while so this was a good read when I managed to find the time to have a look at the goings on. Adds a bit more information to my collection of info on these boards. I find these things a pain to get working, the 40mhz osc seems to fail as two out of my three PCBs needed them replaced. And I have now one of them that just cant seem to find the fault with a data line stuck high so now placed into the to hard at this point basket.

                  Thanks for the good read.

                  Comment

                  Users Viewing Topic: 0 members and 1 (guests)
                  Working...
                  X