CDTV keyboard issues

[Pat the Cat]

... You might have CLOCK and DAT mixed up on the hacked cable? Or the pins are shorted on the hack cable?

[Pat the Cat]

No response at all with them swapped?

Well, you got the power and gnd correct, otherwise the LED wouldn't come on.

Apostrophe = raw keycode 000000000. Looks like something is grounding the keyboard data...

Maybe if you had the power connectors wrong, but the CN7 connector (which takes the signal to the motherboard) is upside down? That would ground out the data line but leave the clock intact.

https://ia801904.us.archive.org/2/items/CDTV_Service_Manual_1991-05_Commodore/CDTV_Service_Manual_1991-05_Commodore.pdf

On CDTV, the kdata and kclock go to the 6500 chip first (from CN7) and then somehow get routed to the even 8520 CIA.

Proceed with caution, always make certain the +5V and GND are connected right before powering on.

[Pat the Cat]

One other point - if you have a CDTV remote, turn it off when testing. (I forgot why the CDTV has an internal keyboard controller like the 6500).

I really do think it's a bad connection rather than faulty chip. Or maybe a cap has leaked and shorted KDAT to ground or similar? 

EDIT: Five pins needed on a CDTV? Keyboard detect sense pin? You could try grounding that.

http://www.rockus.at/Amiga/cdtv2a4000.html

https://aminet.net/search?query=cdtv+keyboard

[Pat the Cat]

No detail included on the iff on aminet I linked to?

I was half thinking you had the dat and clk swapped again.

Apart from that I'm out of ideas on this one.

[Pat the Cat]

It could be. One way to test would be thicker cables - perhaps if you have a spare 5 pin connector for the A3000 keyboard, you could rig up wires going straight into the port?

4.65 is very low, is the board on the CDTV that low too? It's not going to help.

I think tolerance limit is 5%, which would be 4.75. Just a touch higher and it might just work. Hence the thicker cables.

Mind you, if you've got less than 4.75 on the board (coming from the power supply) that's not going to help either.

Cable length is a factor too, if you've got a very long keyboard lead, the voltage will drop more than a short one.

A recap on the power suplly might well sort it. Try fix rather than replace.

[Pat the Cat]

 Hold it - just sticking a thicker piece of wire in between the pin and the current wires isn't going to make a difference.

Measure the voltage on the board. If it's at 5V, then fitting thicker wires to completely replace the present one will make a difference.

I suspect you'll find it's about 4.8V, marginal for operation. IE you were correct, it's the power supply, not you using thinner wires on the connection.

[Pat the Cat]

Check the DATA line on the keyboard. Hook up the scope with clips if you can, see if pressing a key gives you all lows or all highs no matter what key you press.

I'm guessing it would be all highs. Which would point to the DATA line being held high by something inside the CDTV as you figured already.

Especially if it's happening between the two keyboards, which work fine otherwise.

There is also a crystal oscillaor in the keyboard, you could check that to make sure the keyboard controller is getting a nice clean regular signal (unlikely to be that, but worth checking,

Now, assuming all the above is correct... to be positive it is the CDTV - repeat the test with the DATA line disconnected from the CDTV, but the clock and power lines connected.

You should see the data line wiggling up and down freely if it really is the CDTV. When different keys are pressed or released.

Finally, check with no power at all on the port. Is the data line shorted to the +5V line? Or does it float around?

If it floats, it's got to be a faulty component. If it's short, then it's a short on the board somewhere rather than a faulty component (harder to find, easier to fix than getting a spare chip).

It could just be a faulty resistor pack inside the CDTV,.. Might not be such a difficult replacement,

To verify that, again with the power off, check continuity between pin 39 on the even CIA and the DATA pin on the connector. then continuity between the CIA chip and the +5V rail.

Both the KClock on pin 40 and DATA on pin 39 of the even CIA are connected to a pull up resistor pack. If that's gone from being a 10K resistor to being close to a dead short (as far as the DATA pin goes) that would cause the fault.

EDIT: Resistor pack is labelled RP6 on the schematic.

EDIT EDIT: When you get to measuring resistance between KDATA pin 39 of the CIA and +5V (should be no lower than 9K Ohms allowing for tolerances) then do take the chip out of the socket. That way no circuits in the chip will interfere with the measurement.

If it is much lower, replacing the resistor pack should fix it. You might have trouble finding one with the right number of connections though.

[Pat the Cat]

I have individual 10k ohm resistors. I was wondering if there's a way to hack them temporarily on to the connections to see if that'll fix things before replacing RP6 outright. Possible?

Hmmm... would mean soldering to the pins of the chip. (2 pull up resistors, one for kclock, one for kdat).

If you had a IC socket to solder to, and plug the even CIA on top, then yes. I'd be wary about soldering direct to a CIA chip, they seem a bit fragile (very prone to static shock).

I've seen somebody blow up a CIA chip just doing a continuity test on a live Amiga (GadgetUK164). All he was doing was measuring voltage to the serial port.

If you have a socket, it's worth doing just to make sure it really does fix the keyboard problem.

Changing a resistor pack means dismounting the whole board, worth looking for any other faults first (battery? etc etc etc).

[Pat the Cat]

Oh hang on a minute - the +5V is going to take the path of least resistance, ie the RP6 resistor pack.

You will HAVE to remove it for testing with conventional resistors, which is going to mess up other pull signals on RP6.

Wait for replacement. I'm 95% sure a new resistor pack will fix it.

[Pat the Cat]

It looks like it's RP6. The resistance is pretty shot for 40 and 39 against +5V. It's around 2 ohm.

Is that still the case, pins 39 and 40 shorted to 5 volts with next to no resistance? Take the chip out, then measure with keyboard connected and unconnected.

EDIT: There has to be something shorting those pins out either to 5V or to 0V. It could be the lead after all... but you did check for no shorts between any of the pins.