Tetris (Atari) - No video, plays blind

I now have two Tetris boards. One is named AC-Fried - I accidentally ran AC through the video, (was a ChungHwa monitor isolation problem, thread in monitor section), wrecking its ability to sync, video is scrambled like the olden cable days. Other is Blind Tetris which has been the subject of this thread.

For AC-Fried, I'm walking back from the traces and finding continuity on both sides of a capacitor, having another tetris board is a big help in understanding what is and isn't supposed to be there.

In a pause, I decided to debug the original blind board by swapping some other ICs that were easy and see if anything would change:
Slapstick
2x Pokey
EEProm
CMOS Ram

No change, Still playing blind. The EEProm, when swapped over, didn't even play blind. Putting back the original one for that board returned it to playing blind.

I did notice one IC that is socketed on AC-Fried board, that is soldered straight on Blind Tetris: 137550-001, the SOS custom. MAME code says these are used to generate video signals, which I thought was intriguing. I have a working Rampart that also uses this IC, so I'll do a swap between Rampart's SOS and AC-Fried Tetris to verify that the SOS is not what got cooked on AC-Fried. If I got a working one I'll try to piggy back the SOS on the blind Tetris so see if anything interesting occurs.
 
Not sure if you ever figured out the diode thing (I'm guessing you did), but desolder and lift one of the legs to take it out of any circuit that may alter the result and then measure continuity between them, then flip the leads and test again.
 
You really need a scope so you can figure out exactly what isn't working. The one with no sync might have completely dead sync, or the composite sync might be missing the H or V components. On the board that plays blind, you might have good sync but be missing RGB, or everything might be dead.

As far as testing the video RAM, you can remove the chip and test with ex. a NeoLoch RAM tester. Or, since Tetris uses a 6502, you could hook it up to a Fluke to test in-circuit. There's only one RAM on this board, a 6264 on the edge opposite the JAMMA connector. I'd do this just to cover your bases, but because video and scratchpad RAM are shared, I think it's unlikely that this is the problem.

I'm not 100% sure about Tetris, but some many boards have driver transistors for the video, usually 2N3904 / 2N3906. I see a cluster of them near the "VID" header, so I'd look at those. In general, these parts are failure-prone, and they're usually driven by a common power rail, so if either the power or the drivers fail, you'll get no video.

The WD DIS test lug is present on most Atari PCBs; grounding it disables the watchdog circuit. The watchdog resets the game if the program stops running.
 
Not sure if you ever figured out the diode thing (I'm guessing you did), but desolder and lift one of the legs to take it out of any circuit that may alter the result and then measure continuity between them, then flip the leads and test again.

Thanks Otaku, I forgot to mention that, turns out the things I thought were diodes were really capacitors. The ones I pulled seem to be ok so far.
 
You really need a scope so you can figure out exactly what isn't working. The one with no sync might have completely dead sync, or the composite sync might be missing the H or V components. On the board that plays blind, you might have good sync but be missing RGB, or everything might be dead.

As far as testing the video RAM, you can remove the chip and test with ex. a NeoLoch RAM tester. Or, since Tetris uses a 6502, you could hook it up to a Fluke to test in-circuit. There's only one RAM on this board, a 6264 on the edge opposite the JAMMA connector. I'd do this just to cover your bases, but because video and scratchpad RAM are shared, I think it's unlikely that this is the problem.

I'm not 100% sure about Tetris, but some many boards have driver transistors for the video, usually 2N3904 / 2N3906. I see a cluster of them near the "VID" header, so I'd look at those. In general, these parts are failure-prone, and they're usually driven by a common power rail, so if either the power or the drivers fail, you'll get no video.

The WD DIS test lug is present on most Atari PCBs; grounding it disables the watchdog circuit. The watchdog resets the game if the program stops running.

Budget hardware debugging here - no oscilloscope, RAM testers or Flukes on my bench, yet. I agree that the o-scope is probably the right next step.

I'll check and note the transistors for that corner of the board.

I will say that trying to trace things on these boards is difficult - lots of paths under ICs, going from top to bottom multiple times, hidden stuff like resistors and capacitors under the POKEYs, for example. Reaaaally wish circuit diagrams for Atari Tetris would show up somewhere.

Thanks for the suggestions as always, Ieure. :)
 
I tried piggy backing the SOS chip, that did not appear to help at all (didn't even coin up). I'll shelf that idea for now.

There are 2 transistors models on the board - 2N 3904 and 2N 5306 as follows:
J10 - 2N 3904
J15 - 2N 3904
J20 - 2N 3904
K20 - 2N 5306
L20 - 2N 5306
B75 - 2N 3904
B05 - 2N 5306

One of my neighbors lent me his analog o-scope. I'll poke at the sync and RGB lines to see what's not showing. Any other suggestions on what to train the o-scope at?

For the transistors, how can I see the common 'power rail' to rule that out?

Thanks.
 
I tried piggy backing the SOS chip, that did not appear to help at all (didn't even coin up). I'll shelf that idea for now.

There are 2 transistors models on the board - 2N 3904 and 2N 5306 as follows:
J10 - 2N 3904
J15 - 2N 3904
J20 - 2N 3904
K20 - 2N 5306
L20 - 2N 5306
B75 - 2N 3904
B05 - 2N 5306

One of my neighbors lent me his analog o-scope. I'll poke at the sync and RGB lines to see what's not showing. Any other suggestions on what to train the o-scope at?

For the transistors, how can I see the common 'power rail' to rule that out?

Thanks.

I'd expect the power to be on the collector of the 2N3904s. The output on the emitter and input on the base. Not sure about the 5306s.

For the scope, you're going to want a 10x probe, or 1x/10x switchable in 10x mode. Hook the probe GND to a GND test lug on the board, set volts/div to 100mV/div and sec/div around 50ms/div. Touch the probe to GND — you should get a flat line — adjust position so it's centered on the display. Touch to a +5v test lug (or Vcc pin on any IC) and it should remain flat, but five divisions higher on the scope's display. Touch the probe to pin 3 of the 6502 and verify that you have a decent looking clock waveform. From there, start tracking back from the sync or RGB pins on the edge connector.
 
For the scope, you're going to want a 10x probe, or 1x/10x switchable in 10x mode. Hook the probe GND to a GND test lug on the board, set volts/div to 100mV/div and sec/div around 50ms/div. Touch the probe to GND — you should get a flat line — adjust position so it's centered on the display. Touch to a +5v test lug (or Vcc pin on any IC) and it should remain flat, but five divisions higher on the scope's display. Touch the probe to pin 3 of the 6502 and verify that you have a decent looking clock waveform. From there, start tracking back from the sync or RGB pins on the edge connector.

This is a GREAT step by step. Thanks! Will report back here when I go through it.

I've uploaded the video for AC Fried here:
https://youtu.be/3f_bOHqtbVw
 
Ok, got the bench all setup, scope dialed in as per recommendation, and started checking things.

On blind board: all video at Jamma edge is flat on scope. The activity at 2N3904 transistors mentioned above are also non-moving: C:+5v, B:0.8v, E:0.2v steady. The 2N5306s are also steady at E:0.2v, C:5v dropping on measure, B:0v

On AC Fried: RGB has activity on scope, Sync flat. The activity at 2N3904 transistors is also active: C:+5v, B:varies, E:varies. The 2N5306s are also steady at E:0.2v, C:5v dropping on measure, B:0v

Wild guess: The 2N3904s are controlling RGB, 2N5306s are for sync (H+V)? Also guessing my problems are not on these transistors but elsewhere.

Focused on the 2N5306s for now. I traced the path from the 2N5306s Emitters, both went to resistors, and then to an 74LS74 @ E15 (pins 5 and 9, separate each transistor). Looked at spec sheet for LS74, it's a flip-flop. (http://www.electronica60norte.com/mwfls/pdf/74LS74.pdf) Probing the pins on the LS74 showed the outputs (pin 5 and 9, respectively) same as transistor side. Looking at other pins on the LS74, noted that PREset (pins 4 and 10) are stuck High. Unsure if this is meant that way, I looked at other LS74s on board and their PREset pins have range of values. So I traced those pins (they're traced together on board) to pin 11 on 74LS197 @ E05. That is a Ripple Counter. Here's where I'm confused - pin 11 is an input according to the spec doc. So I must have missed a trace setting this to 5v elsewhere?

Break here for now. Next time, I'll check that LS74 pins 4+10 lead elsewhere that may be stuck high. I'll also work back from the 2N3904 transistors to figure out where things connect.
 
I mapped the lines that are connected to the 2N5306 transistors to a huge pile of stuff, and looked at each one of the data sheets for any possible outputs - there aren't any obvious ones! Which means I'm probably looking at that 5306 transistor incorrectly - I got confused on the Emitter function of the 2N5306, so whatever is driving that transistor may be compromised?

Atari Tetris PCB Shared Trace by ICs Map

Location, IC Identifier, Pin Connections
E15, 74LS74AN, (4, 10) [4: Input 1PRESET, 10: Input 2PRESET], Flip Flop
E10, 74LS197N, (11) [11: Data Input], Presettable 4-Bit Binary Ripple Counter
C10, 74LS374N, (8) [8: Input 4D], Edge-Triggered Flip Flop
A15, 74LS374N, (3) [3: Input 1D], Edge-Triggered Flip Flop
C25, 74LS74AN, (4, 10, 13) [4: Input 1PRESET, 10: Input 2PRESET, 13: Input CLR], Flip Flop
D30, 74LS74AN, (10) [10: Input 2PRESET], Flip Flop
C30, SN74LS257BN, (6, 10) [6: Input 2B, 10: Input 3B], Quadruple 2-Line To 1-Line Data Selectors/Multiplexers With 3-State Outputs
F35, 74LS245N, (1) [1: Input DIRection], OCTAL BUS TRANSCEIVERS WITH 3-STATE OUTPUTS
D60, SN74LS257BN, (10, 11, 13, 14) [10: Input 3B, 11: Input 3A, 13: Input 4B, 14: Input 4A] Quadruple 2-Line To 1-Line Data Selectors/Multiplexers With 3-State Outputs
F70, 74LS245N, (1) [1: Input DIRection], OCTAL BUS TRANSCEIVERS WITH 3-STATE OUTPUTS
E70, 74LS245N, (1) [1: Input DIRection], OCTAL BUS TRANSCEIVERS WITH 3-STATE OUTPUTS
 
I'm not sure you are tracing right - that seems a lot of stuff attached to that transistor :)

I think in actual fact the resistor networks near the transistors are what is driving the transistors. Think of it this way - those transistors drive the final analog RGB outputs. The game works internally with digital TTL. The resistor networks convert the digital palette RAM into analog signals for the transistor. On this basis I wouldn't expect any TTL chips to directly connect to those transistors. (EXCEPT - looking at the Gauntlet schematics I see a 7406 is used to detect when a color channel is all black - then it blanks off the output in the analog section).

I know this doesn't really help you! I would probe the TTL side things of more as that fails much more often than the transistor side. Specifically the palette RAM chip - if that failed and always outputs zero or is floating, then naturally your transistor section will always get black to display.

I don't have this board but maybe someone can identify the palette RAM as putting the scope on it's output would help.
 
I'm not sure you are tracing right - that seems a lot of stuff attached to that transistor :)

You're right Tendril, to clarify, each of the 2N5306 Darlingtons connect to resistors that are (driven?) by the outputs on E15 74LS74 pins 5 and 9 each. When probing the leads on that IC I noticed 2 input pins (4 and 10) are pegged at 5V, which made me think there was something stuck high, leading me on that goose chase.

There was another Tetris PCB repair log on the net that mentioned the video ram as an issue. I shotgun replaced that on my blind board some time ago, no change.

I found a copy of the manual that has some interesting info here http://www.aurcade.com/games/manuals/00000230.pdf:
The 'Atari Video Connector Pin Assignments' located at 15K is as follows:
1 - Red
2 - GND
3 - Key (No connection required)
4 - Green
5 - GND
6 - Blue
7 - GND
8 - GND
9 - Negative Composite Sync
10 - Positive V sync
11 - Positive H sync

I'll check these for activity with the scope and back trace them?

It also mentions doing additional horizontal shifting by removing R20 and putting another low-resistance resistor, less than 100 ohm, or straight wire at R30 or R40 (not both) I'll check those resistors as well.
 

Attachments

  • AtariTetrisPCBDiagram-ResistorShorts.jpg
    AtariTetrisPCBDiagram-ResistorShorts.jpg
    54.7 KB · Views: 22
Agree with tendril, I think you ended up on a goose chase. His description of the resistor network sounds correct to me.

Looking at other pins on the LS74, noted that PREset (pins 4 and 10) are stuck High. Unsure if this is meant that way, I looked at other LS74s on board and their PREset pins have range of values.

Hard to tell without a schematic, but it's fairly common for inputs to be pulled high like this & not change. Usually there's a resistor connected to then +5 rail on one end and the input on the other.

The Atari video connector is going to connect to the same places as the JAMMA edge, except there are two additional pins that supply discrete positive sync as well as the negative composite sync JAMMA uses. You might check those out, if you have sync there but none on the /CSYNC, that's a good indication that the actual sync signal is working, but the component that combines it into /CSYNC is faulty.

The resistor matrix is pretty easy to understand. There are a bunch of resistors of different values in a ladder (exL 47Ω 100Ω 150Ω 200Ω — all totally arbitrary values). All the outputs are connected together. The other sides of the resistors connected to the output of a decoder like a 74LS139 or similar. The decoder takes a binary input representing a color intensity (ex 2 inputs (bits) for four levels of intensity), and enables a single output line. Due to the resistors, this makes the output of the network produce a different analog voltage (representing the analog intensity of the color) for each binary value input into the decoder. Then that feeds into the 2N3904s to amplify the overall signal up to a level appropriate for the monitor. Multiply that by three & you have a basic RGB output stage.

While there isn't a Tetris schematic, I was looking at the Klax manual, which does have one & also uses the SOS-2 custom. Maybe you can use that as a guide for what to expect from the SOS-2 on Tetris. Per that schematic (page 5-11 of the manual), /VSYNC comes out pin 24 & /HSYNC on pin 23 of the SOS.
 
The resistor matrix is pretty easy to understand. There are a bunch of resistors of different values in a ladder (exL 47Ω 100Ω 150Ω 200Ω — all totally arbitrary values). All the outputs are connected together. The other sides of the resistors connected to the output of a decoder like a 74LS139 or similar. The decoder takes a binary input representing a color intensity (ex 2 inputs (bits) for four levels of intensity), and enables a single output line. Due to the resistors, this makes the output of the network produce a different analog voltage (representing the analog intensity of the color) for each binary value input into the decoder. Then that feeds into the 2N3904s to amplify the overall signal up to a level appropriate for the monitor. Multiply that by three & you have a basic RGB output stage.

This is a great explanation resistor matrix. Learning a lot from trying to get this working again.

H Sync, V Sync and Neg Composite Sync are all connected to F10, SN7406N (Pins 10, 12 and 2 respectively), a Hex Inverter to convert TTLs->MOS (which I think Ieure said was the function of the resistor matrix). Those pins are driven from inputs at Pins 1, 11 and 13.

The inputs for H sync (pin 11) and V sync (pin 13) for the F10 IC track all across the board to D50, 74LS00N (pins 1 and 2) where I think they're combined via NAND gate to make the positive sync out of Pin 3 (guess to confirm). Pin 1 on D50 is also connected to pin 24 on the SOS. That's as far as I got tonight.
 
This is on ACFried Tetris. I scoped out pins 1, 2 and 3 on the D40 LS7400. Pin 1 matches activity seen on the SOS @ Pin24, which according to Klax docs is /VSYNC; its pegged high with that one drop of noise to 0v - that seems wrong? HSYNC I guess seems ok (high and low)? I figured these would be wave forms but no matter how I set the scope I get these pegged high low type views.
 

Attachments

  • AtariTetris-VSYNC-SOS-Pin24.jpg
    AtariTetris-VSYNC-SOS-Pin24.jpg
    196.1 KB · Views: 9
  • AtariTetris-HSYNC-D50-LS7400N-Pin2.jpg
    AtariTetris-HSYNC-D50-LS7400N-Pin2.jpg
    192.2 KB · Views: 10
  • AtariTetris-D50-LS7400-Pin3.jpg
    AtariTetris-D50-LS7400-Pin3.jpg
    192.8 KB · Views: 10
Last edited:
This is on ACFried Tetris. I scoped out pins 1, 2 and 3 on the D40 LS7400. Pin 1 matches activity seen on the SOS @ Pin24, which according to Klax docs is /VSYNC; its pegged high with that one drop of noise to 0v - that seems wrong? HSYNC I guess seems ok (high and low)? I figured these would be wave forms but no matter how I set the scope I get these pegged high low type views.

I too would expect vsync to be pulsing. Looks the SOS is a timing generator - see if you have any pulses on pins 32 - 37 (vertical line counters). If not check the clock input pin which is 10.

If 10 is not pulsing then trace back from that. If 10 is pulsing, but nothing comes out of 32-37 then probably the SOS is dead.
 
I too would expect vsync to be pulsing. Looks the SOS is a timing generator - see if you have any pulses on pins 32 - 37 (vertical line counters). If not check the clock input pin which is 10.

If 10 is not pulsing then trace back from that. If 10 is pulsing, but nothing comes out of 32-37 then probably the SOS is dead.

Thanks Tendril, I checked out the SOS, pins 32-37 are active (all of them) and pin 10 is beautiful clock cycle at 100ns.

I'm guessing the electrical violence shorted something upstream to peg the VSYNC high-ish. I'll continue investigating.
 

Attachments

  • AtariTetris-SOS-ClockPin10Scope.jpg
    AtariTetris-SOS-ClockPin10Scope.jpg
    194.1 KB · Views: 6
Discovered SYNC is grounded out on AC Fried, i.e. continuous from SYNC to Video GND *no bueno*. C39 has continuity on both sides, which my other Tetris board does not. I pulled C39 from circuit and it's not shorted according to my tester, and it's not shorted out of circuit. Both sides of the holes for C39 test grounded.

Trace one side from C39, and it just leads to the CSYNC pin, and the 7406N at F10. Noticed pin 2 (CSYNC) and pin 7 (GND) are continuous. Not the case on my other board. Desoldered and lifted pin 7, and sure enough, it is shorted from pin 2 to pin 7. SYNC at edge is no longer grounded! Clip off F10 7406, replace with socket, install fresh 7406. Victory #1!

Looks like SYNC signals flow from SOS -> D40 LS7400 -> F10 7406N -> SYNC at edge/video pin connector. No transistors in the mix for sync (previous guesses on 5306 being sync are incorrect).
 

Attachments

  • AtariTetris-F10-SYNC-IC.jpg
    AtariTetris-F10-SYNC-IC.jpg
    210.4 KB · Views: 23
  • AtariTetris-ACFriedBoardRepaired.jpg
    AtariTetris-ACFriedBoardRepaired.jpg
    201.5 KB · Views: 20
On Blind Tetris, I looked at the SOS pins, and it doesn't have activity all over (pin 24 and pins 32-37 are high not fluctuating). Pin 10, the clock, looks good on scope.

I'm guessing dead SOS based on Tendril's suggestion. It's soldered on board (not socketed). Debating clipping it off to put on a socket and try one from the other, working board - unfortunately don't have a re-work/desoldering station. Only site I found is asking $8 for a pulled one, so I'm risking a bit on trashing it if its not the problem on the board.
 
On Blind Tetris, I looked at the SOS pins, and it doesn't have activity all over (pin 24 and pins 32-37 are high not fluctuating). Pin 10, the clock, looks good on scope.

I'm guessing dead SOS based on Tendril's suggestion. It's soldered on board (not socketed). Debating clipping it off to put on a socket and try one from the other, working board - unfortunately don't have a re-work/desoldering station. Only site I found is asking $8 for a pulled one, so I'm risking a bit on trashing it if its not the problem on the board.

Hooray - well done! :)

Double check that +5 and GND on that SOS too before declaring it dead. A heat gun to the back of the pcb would remove it with no damage to either chip or board, but if you've never done that before you might want to practice on a less important board first! :)
 
Back
Top Bottom