ukVac.com Homepage
Forum Home Forum Home > Technical > Fix logs
  New Posts New Posts RSS Feed - Missile Command - step by step
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Skin:


Missile Command - step by step

 Post Reply Post Reply
Author
Message
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (1) Thanks(1)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Topic: Missile Command - step by step
    Posted: 30 Oct 2018 at 11:12am
I'm working on a couple of MC PCBs at the moment, so I'll write them up here as I go. (So I don't forget!)

In general the MC self test is pretty good.... but quite often boards have faults that prevent them getting that far.
So the first PCB I'm working on won't self test. At this point I always start with the SYNC circuits (confusingly nothing to do with SYNC line on pin 7 of the CPU)
The SYNC circuits are responsible for all the timing signals on the board, anything wrong here and you won't get far.
So... I have a good 10MHz, 5MHz and H signals including H-Sync.
BUT I have no V-SYNC.
V-SYNC is based on the 32H and 4H (and _FLIP, but that's not a consideration right now)
Both of those (outputs on pin 14 of E5 and 13 of D5, inputs on A6 pins 11 and 12) are fine.
Following the circuit through checking with a logic probe I can see that a couple of outputs on counter at A4 are stuck low.
Replacing the counter at A4 means I now have V-SYNC and all the "V" signals seem ok (1V .... 128V)

Still no sound on self test.
I check that my test rig is working right - yep the "TEST" is low on the MC PCB.
Checking pin 37 on the Pokey shows the pattern I'd expect for self-test audio (it pulses in the sequence of the self-test "beeps"!), so there's something wrong with the audio output as well. That's my next step for investigation, probably starting around the LM324 on the audio output.


Back to Top
DanP View Drop Down
Admin Group
Admin Group
Avatar
Whipcracker

5 Years of Supporting ukvac.com!

5 Years of Supporting ukvac.com!



Joined: 17 Apr 2000
Location: Essex
Status: Offline
Points: 6096

Feedback: 5
Post Options Post Options   Thanks (0) Thanks(0)   Quote DanP Quote  Post ReplyReply Direct Link To This Post Posted: 30 Oct 2018 at 1:16pm
Nice fix Spammy.  Interesting to know that the sync circuit is actually for something else on MC...
Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 30 Oct 2018 at 1:58pm
.... the saga continues.
So I have a nice clean signal (clearly the 8 beeps and the buzz from the self test, but I can't tell which are high or low beeps with just a logic probe!) on pin 37 of the pokey.
BUT! Pin 5 on the LM324 at N10 is just sitting high!
Continuity check between pin 37 pokey and 1 leg of R38 indicates no connection. 
Turning the board over reveals a dry joint on pokey 37.
Re-flowing that joint gives me sound.

Now I have a buzz and 8 clear "beeps" - the ram test is working.
The 3rd beep is high pitch, but the other 7 are all low.
That indicates bad ram at M4.... so that's the next job!


Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 30 Oct 2018 at 4:37pm
one step forward three steps back.
So I removed M4 4116. Replaced it with a socket and added a 4116 chip.

In test mode the beeps still indicate an error on M4 (but that's not a massive surprise, MC sometimes gets the RAM check wrong.)
Out of interest I swapped M4 for one of the other socketed RAMs (previously replaced by me)
No change, beeps still indicate a fault at M4.

So I swapped the chips back.... clearly at this point I bggrd something up. As now the PCB won't self test.
SYNC signals all look ok.
Fluke indicated a RAM error, "RAM DCD ERR @ 0009 BIT 4" so it backs up the previous self test issue - something in the RAM address/data side of things is screwed.
BUT all the ROMs checksum fine in circuit.

So I'm left wondering why it's not self testing anymore. I'll go back over all the timing circuits and then start looking at the addressing/write status on the watchdog circuits.

Grrrr.... I like a challenge, but sometimes these boards go too far!
Back to Top
backflipper View Drop Down
Senior Members
Senior Members
Avatar

Joined: 26 Apr 2006
Location: Bow, London
Status: Offline
Points: 2617

Feedback: 5
Post Options Post Options   Thanks (0) Thanks(0)   Quote backflipper Quote  Post ReplyReply Direct Link To This Post Posted: 31 Oct 2018 at 11:50am
lol damn boards taking things to far, I like that.

Good to read another fix log.
Back to Top
Equites View Drop Down
Senior Members
Senior Members
Avatar
Chief Sheesher®

5 Years of Supporting ukvac.com!

5 Years of Supporting ukvac.com!



Joined: 28 Jun 2010
Location: Warrington
Status: Offline
Points: 21807

Feedback: 5
Post Options Post Options   Thanks (0) Thanks(0)   Quote Equites Quote  Post ReplyReply Direct Link To This Post Posted: 31 Oct 2018 at 12:06pm
Check your ROM sockets on these.
Atari PCB's (and most non-Atari), AR2's, Vector Deflection/HV boards repaired & refurbished.
Wanted: Non-working Atari PCBs.
Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 31 Oct 2018 at 1:20pm
All ROM sockets on the board have been changed for turned pin sockets and the ROMs verify fine with the fluke plugged in.

I found a damaged track DEAD4 on the dram addressing circuit (common to all pin 11s on the 4116s) I meant that M4 onwards didn't have DEAD4 connected.
I've bodged a bit of solder onto the track to reconnect.


Back to no sounds on the self test, but it looks like the watchdog is barking, so I tied WDDISABLE to ground and got this...

I'm thinking there's something screwed in the ram addressing post A->BA in the addressing (as the ROMs all address and read fine), probably somewhere around the 157s where they go from BA->MAD in addressing, but I'll start doing some checks later.

EDIT.... having checked the schematics, my money is currently on the 74-42 at E8 as it affects _WDOG and _COLRAM... but we'll see!

EDIT #2: Here's a video of the board running with WDOGDIS tied low:


Edited by yorkshire_spam - 4 hours 53 minutes ago at 3:07pm
Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 05 Nov 2018 at 10:29am
A quick update on this although it's not actually fixed yet....

With the fluke in the processor socket the board now verifies in terms of bus, rom and ram. (Long ram test passes) but clearly there's still a fault with the WD barking

I'd previously tested all the timing circuits but only with a logic probe. At this point it's time to break out the 'scope.

I started working through all the timing lines and there's something not quite right on the 64H line (E5 pin 13), so I swapped the 74ls160 at E5. BUT the new known good IC gives the same "quirky" output. So I suspect I have a track/pin short or a seriously faulty IC downstream. As 64H (and other H signals) feed the logic in the DRAM address decoder I think it's a reasonable candidate for the fault I can see on-screen.

I'll post a proper update with some scope traces etc. when I finally fix the fault.

Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 05 Nov 2018 at 8:36pm
I am such a muppet. I forgot that E5 is an LS160 and D5 is a LS163....

So the output pattern on 14,13,12,11 on E5 will not be the same as D5.... it should look like this:

Which it does.

Back to the drawing board on this one!

Back to Top
yorkshire_spam View Drop Down
Groupie
Groupie


Joined: 12 Jan 2002
Location: United Kingdom
Status: Offline
Points: 343

Feedback: 0
Post Options Post Options   Thanks (0) Thanks(0)   Quote yorkshire_spam Quote  Post ReplyReply Direct Link To This Post Posted: 4 hours 55 minutes ago at 3:05pm
A couple of steps backwards on this board (but that's just the nature of these things!)

I came to do some more work on the board the other night and it had reverted to the typical Missile Command blue screen of death.

So I pulled the CPU and ran some basic checks with the fluke - rom, ram and bus all seemed ok.
I checked the _RESET (pin 40) on the CPU and it was stuck low.
A bit of tracing back through the schematic revealed a dead counter at D4 (74-93)
Replacing that got the screen flicking again, but still blue. Watchdog is now barking, but looking at the DRAM circuits I can see that the MADSEL is stuck.
I checked pretty much all the logic that feeds the MADSEL and it all seemed fine, but a check of the 74ls74 flip-flop at F7 using the HP comparator reveals it's outputs are goosed. 
Piggy-backing a working IC at F7 get's me back to where I was 2 weeks ago - the board is running but with the glitches you can see in the video above.

I'll stick the faulty IC and the replacement on the scope and try and get a screenshot to post here for reference.
Then it's back to trying to figure out the screen issue. I'm going to dig out the Data I/O 29B and get it running so I can dump the prom at L6 (I socketed it earlier in this PCBs life, so it's an easy check)

Back to Top
 Post Reply Post Reply
  Share Topic   

Forum Jump Forum Permissions View Drop Down



This page was generated in 0.234 seconds.