Faulty joystick port

[jimbob]

Hello, I have an A600 with an odd fault on the joystick port. When pressing up on the joystick, the amiga registers up and left as being pressed.

Left works normally, up+right works normally. If I use a mouse port switching utility, a mouse in the joystick port works normally.

Tried a bunch of different joysticks which all work normally on another amiga so it isn't the stick.

Up and left share a channel on the U34 chip so I swapped that suspecting a damaged input but the fault remains.

Any ideas?

[demolition]

Originally Posted by jimbob

Hello, I have an A600 with an odd fault on the joystick port. When pressing up on the joystick, the amiga registers up and left as being pressed.

I had a similar problem with my A600. When I pressed up, it would sometimes register a left as well (although not all the time). I couldn't find the error 100%, but it seemed to be a bad connection to the pull up resistor, although they measured fine from the joyport, so it couldn't be verified easily.

I put a 20k ohm resistor between pin 3 and pin 16 on U34 and haven't seen the problem since. You also just remove the corresponding pullup in R370 and put a 4.7kohm between pin 3 and 16.

[jimbob]

Great I'll have go at that. Pull ups all measure normal as you describe and I can't see any obvious problems with them on the board but it's worth a shot.

Do you mean pin 3 on the DB9 connector or pin 3 on the chip. I think it is pins 10 and 11 for the up and left inputs on the joystick. (pins 2 and 3 for the mouse port).

Also, I'm not sure why you used 20k as well as replace the 4.7k pullup?

[demolition]

I meant on the chip, but I don't have the Amiga right here and just had a quick look on the schematic and might have switched the two ports around.

I put a 20k on there for fault-finding purposes to ensure it would never be completely floating as the symptoms suggested a sometimes floating pin on the left input. If I just put a 4.7k on there, it would double the current through the joystick and I wasn't sure that would be good. But it turned out the 20k fixed the problem so I never got around to making a proper fix and replace the old pullup.

Problem is that I don't have any decent soldering equipment at home for SMD and have to bring it to work every time I want to change something and bring it back home for testing.

[Lord Aga]

Wow, just the thing that has been plaguing my A600. I can only play games with mouse and keyboard controls
I have searched some old threads, and this problem seems to affect a lot of A600s. Sadly, I couldn't find any definite answer. Some guy sorted it out with a better PSU. Some tried changing IC's but I don't know where that got him. All threads mentioning this problem seem to fade away without conclusion.

I sure hope this will be it. Also, I hope some of the tech gurus will visit this thread and have a go at this problem

[jimbob]

No joy yet. Removed R370C pull up resistor for left joystick and replaced with a 4.7k resistor between pins 10 and 16 of U34. Fault remains. I'll keep trying but not tonight.

[jimbob]

Success . . . kinda.

Seems to be a power supply problem in this case but must be some degraded circuitry too since other amigas work perfectly with the same PSU.

The PSU I was using works fine on all my other amigas but since LordAGA mentioned PSU I checked its 5V and it turned out to be on the low side ~4.6V.

So I tried one with an adjustable 5V and brought it up to 5V exactly on pin 16 of U34 and the joystick works normally. Turn it down past ~4.6 and the fault reappears.

If i had to guess, which I do, this is what is happening -

With a low 5V, U34 cannot bring its output up to a reliable logic 1 quickly enough when the clock switches the mulitplexor input from 'up' to 'left'. There are two low to high transitions here which will be slowed down by the combination of weak 5V and big long trace capacitances of clock to U34 and U34 to denise. The clock, and the output.

If I've read the schematic and U34 datasheet right, I think the reason the reverse fault, (left falsely triggering up), doesn't occur is because the clock hi to low transition will be quicker than low to high, giving the output more time to change its state before denise tries to read the level.

I haven't tried it but I bet if I lowered the voltage a little more, and if the machine didn't just crash, pulling down on the joystick would register as down and right.

Anyway, I am calling this fixed, or not broken, or at least not totally broken yet.

[Lord Aga]

All right So we seem to have some explanation at least.
Way to go jimbob !
Guess I'll have to find a better PSU then...

Maybe someone should think about making proper Amiga PSUs. With all these new accelerator cards I think we need some quality juice.

[demolition]

This is interesting and it could be the solution for my problem as well. My 5V supply is measuring 4.75V, so it could be right on the border line. The extra 20k pullup solved the problem, maybe due to the slightly faster rise time of the signal.

[jimbob]

Originally Posted by demolition

This is interesting and it could be the solution for my problem as well. My 5V supply is measuring 4.75V, so it could be right on the border line. The extra 20k pullup solved the problem, maybe due to the slightly faster rise time of the signal.

Well I was thinking the problem was with the output of U34 not changing fast enough, since when pressing up, the left input should be held high and not actually have to rise when the clock changes.

I was putting back the original smd pull up and tested it before removing the replacement I put in, (so the pull up resitor was halved in value) to see if this would help but it didn't. Good 5V still worked, poor 5V remained faulty. Could be we don't even have the same cause.

I also tried lowering the voltage more, nothing much changes until about 4V. Then the mouse in port 1 stopped responding to vertical movement and left on the joystick did nothing while up still registered left+up, (down and right kept working normally).

Not sure exactly how mouse signals are captured but I know that left and up are used together to make the vertical movement so it seems that the M0V and M1V signals to denise are effected by this worst. Maybe active pull ups internal to denise are degraded or just slightly weaker on some amigas which get this fault when 5V is poor.

Not sure this explains why some people got this when plugging indi ECS into the trapdoor, see related threads at bottom of the page.

Anyway, this one is going back to its owner today, hope he has a decent PSU.