Atari System II speech repair OR "now you have a friend in the paper business"

kstillin

Well-known member

Donor 2018
Joined
Aug 24, 2006
Messages
4,698
Reaction score
130
Location
Columbia, South Carolina
Atari System II speech repair OR "now you have a friend in the paper business"

I finally got off my arse to look at the Paperboy speech impediment my game was having.
All the sounds were pretty good except for the speech which came out as garbled noise.

First tried swapping the speech chip. No change.
Poke around with the oscilloscope and schematics. The speech signal to the op amp LM324 at 10B looks funky. I'm missing one channel of sound and I track that down to the output amp on the ARIII board (Q1, a TDA2030).

Now I have both channels, but still garbled speech.

Two of my resistor networks are cracked apart at RN1 and RN2. The schematics say that Paperboy only uses pins 5 and 7 at J102 for steering and speed, so I slip in 10K resistors in those two slots at RN1 and RN2 instead of ordering the resistor packs or trying to salvage one.

The data inputs to the speech chip at 7D (D0, D1, D2, etc) all look like they're getting data from the CPU. That looks good.

I've got another System II board that DOES have speech, so I do some comparing. Pin 18 on the speech chip is a RDY (ready) line. It appears to have regularly spaced little spikes on it during silence, and it looks like it's clocking out a pattern during speech. On the bad Paperboy CPU board, it looks like it's always clocking.

I keep poking around and zap the +15v supply.
Fix that and return the next night.
Looking at the good CPU board, I swap in the sound ROMS from 2A, 2B/C, and 2D. They work in the good board.
Measure some signals surrounding the speech chip on the good board-
TIVSS@pin 4 on speech = -5V
Look at the latch at 8D. Observe that the outputs follow the inputs and that they hold high and clock out data when speech is triggered.
On the good board, pin 27 (/WS) and pin 18(RDY) on speech chip have the pretty pulses regularly spaced. On sound trigger, pin 18 goes to toggling out a signal.

Plug back in the bad board. Measure pin 4 on the speech chip. -12V??
Damn- I zapped the +15V again.
Pins on the latch at 8D seem to match the good board though.
On the bad board, pin 18 @ speech never follows pin 27. Since the chip is known good, maybe the socket is bad?

Next night- try again. Replaced the 7815 for the +15v supply.
(oh- I replaced the socket, but I messed up and replaced a stupid ROM socket instead of the speech chip socket. DOH).
Plug back up, first thing I measure is the +15V. Read 6.3V? Zapped again?
Plug in a different ARIII I found. Measure 15V without connector attached. +15V OK.
Plug in connector to CPU board. BAM - Down to 6.3V.
So, something is bogging down the +15V. (it reads normal when connector is not plugged in)
I realize my 1st ARIII board +15V is still good. It was just loaded down when plugged into the CPU. I probably replaced those 7815 chips unnecessarily.

Note which chips are supplied by the +15V. Several LM324s on rows 9 and 10, and some LF13201s on row 9.

Ohm out the +15v to line on each board. Good board reads about 39K ohms. Bad board about 30 ohms. Figure I can pull those few chips to narrow it down.

Back at home, I probe around with voltage lines. (machine cabinet is at storage location, hence the back and forth). I find that the +15V line to the +5V line is less than 1 ohm!

I find a cracked diode CR7 between the +15 and the +5V lines. Remove and replace. Also, for kicks, I change the 2N3904 and 2N3906 in the circuit that supplies TIVSS voltage for speech.

Take back to cabinet, +15V is good. The TIVSS line is OK, but the VDD line is at -12V. Damn it- should be -5V. Realize the -5V is derived from -15V and generated by 7905 at VR1.

Take it back home again- SUCCESS!

So, the speech chip wants it's dual supply voltages to be nicely centered around 0V. +5V/0V/-5V at pin5/pin11/pin4
Without those, even though it's getting good commands from the CPU, the output was not a good signal to pass on to the amps.

In the end, just a bad diode and a regulator.
 
Thanks for posting. I really enjoy reading detailed troubleshooting threads like this.
Congrats on your speech! mine was broken when I first got it, but it turned out to be just a bad speech chip.
 
Thanks, gives me a good place to start. I appreciate you taking the time to document it.
 
Back
Top Bottom