Pinball CPU data lines pulled up to 1.2V

[drhex]

Hi,
I have a pinball CPU on my desk where D0-7 are being pulled up to about 1.3V with CPU, RAM and EPROM removed. Lots of logic chips on the data bus, already identified and removed a leaky buffer which helped somewhat. Any smarter way to identify culprit(s) than snipping off 5V from all chips and see what happens? Nothing gets hot or looks particularly untoward. Some battery damage but affected chips aren't in the area. Post removing the buffer signals look better but there still are times when I get undefined states.
Thanks,
Florian

[floobydust]

It's not easy when so many parts are connected together on the bus. All you have left are the peripherals?
But in removing the CPU careful the address bus is not floating and making new symptoms that are a goose chase.
Bus contention can also be cause by a bad address decoder, where it is enabling two or more peripherals at once. Look at the chip selects/enables as well. And that the (non-driven) address bus is OK too. I wouild use a weak pulldown if it's CMOS when probing.
You can give us the model/schematic for better help.

If it's ugly, I use freeze-spray on each of the IC's.

[up8051]

What pinball machine?

[drhex]

It is a Sega board used in X Files and Lost World e.g. Without CPU all data lines are at 1.2V constantly, with CPU in it seems like they are at 1.2V whenever the lines aren't actively driven. Something is wrong. Anyway, there are about eight logic chips on the data bus, will cut their power pins one by one and see what happens. Surely the last one :-)... Board seems to be somewhat running as there is traffic on data and address. Also, the run LED is lit, sadly there is no test EPROM or any debug information (blink codes or such) for this board. I have a display connected so should see when it does something sensible. Chip addressing is happening via a PAL which also isn't particularly helpful. That is new though so should work.

[up8051]

Try adding Pull-Up resistors and check if the voltage is forced to 5V.
The voltage on the unconnected input of the TTL gate can be about 2V

[floobydust]

OP be careful you're not going down the wrong road of cutting power pins when it's something else. Is that new PAL any good and in the the socket as well?

You've gotta pay attention to CMOS/NMOS/TTL logic family differences and who floats what when nothing is driving the bus.
This board is dual 6809? and a shit ton of parts.

[CaptDon]

Actually I would expect a float voltage of about exactly what you are getting. What did you expect it to be and why? Try pulling up with 470 ohms and pulling down with 100 ohms and see where the voltages shift to. Those values should provide an acceptable hi level and a marginal low level.

[drhex]

I‘d expect the same as on the address bus where I have either zero or 4.5 volts. I‘ll post a few pics later.

[Ian.M]

If there is any bipolar TTL connected to the data bus, as others have explained, their inputs are *NOT* very high impedance and normally float to this range of input voltages.  Look for chips with a part number like <letters>74<letter(s)><2 or 3 digits><letter(s)>.
If the middle group of letters does NOT include 'C' or 'LV' its probably bipolar TTL - see https://everything.explained.today/7400-series_integrated_circuits/ for details.

You'll probably find there is a TTL octal buffer or latch on the data bus somewhere.

Fun historical fact: Only a few vintage families of 'TTL' are actually Transistor Transistor Logic, with multiple emitter input transistors.  Anything with a 'S' in the middle group (or F on its own) is Diode Transistor Logic using Schottky diodes for the input gate, however their input bias levels and bias currents were designed to be compatible with real 'TTL'.

[drhex]

Ok, definitely lots of discreet logic there. Would that only affect the data and not the address lines?

[Ian.M]

That depends on what its connected to.   The address lines probably only go to the CPU, RAM, ROM and PAL, as if you have a PAL for address decoding why would you also use a mess of TTL MSI logic for that function?    The odds are there are some TTL latches for output and tristate buffers for input, connected to the data bus and their addressing controlled by the PAL.

[floobydust]

Is it the MPU board for The Lost World Jurassic Park pinball machine?
Which schematic do you have? I found one but too low rez scan.

[drhex]

Good quality schematic is here: https://pinwiki.com/wiki/images/e/ec/CPU_Sound_Bd.pdf
A few pics attached, two of the data lines, one of the address bus
Data:


Address:

Data really looks wonky to me.
This board was used in plenty of machines, currently has X-Files ROMs on it (which I read out, they are good). Doesn't initiate things properly, PRES is always held high e.g. so the display is never initialized. It is running though as per the images, just probably not the right stuff. The guy owning the corresponding pinball said it went all crazy when testing. Found a short between two data lines in the first go which got LED200 lit - this is the only debug info the board creates. I also took out U206 as that was leaking through the voltages from the noise supressing caps/pull ups - this cleaned up things considerably. I don't know the history of the board, it was sold as defective on the Bay apparently. Could well be that someone took 50V coil power into it. Also, there is some battery leakage damage, sockets were replaced by the owner (this is where I found a short) and some corroded traces repaired. 

[floobydust]

Those waveforms are the databus with the CPU U209, SRAM, EEPROM removed? Not the plasma controller databus, not the BSMT2000 databus? I see a mix of 'LS and CMOS logic.
Strange to see some clock modulated? signals there. I suspect the PAL16V8Q might have internal pullups and it should not be enabling anything if the address bus is dead off. But somebody drives it to 5V once in a while. Who is that?

[drhex]

No, everything in, three logic chips powered off. U5, U206, U208. This is straight from pins 23, 24 and 25 on the CPU. Also, I think everything on that bus is LS or HC, so should be 5V?

[drhex]

Ok, CPU socket had an issue R/W didn‘t make contact. I hate boards someone has worked on before. More there, still not running right. :-(

[drhex]

And found another corroded through data line in the end. That killed a few bus drivers but left CPU/RAM/ROM working. Board is now OK as far as I can test.
Now left with a conundrum with the original PAL chip - it works but any "new" ones I am burning don't. TIBPAL16L8 and NS PAL16L8 seem to be the same thing though so don't know why. Have read the original and used the files available on the net. Must be missing something along data levels or such.