Korg Forums

[enigmahack]

Okay, I've got a VERY bizarre issue here and I figured I'd come here where some experts roam for some suggestions.

I've had an issue with my joystick +X and -X. Sometimes, (often enough that it's not a reliable gigging board anymore) I'll experience issues where it sounds like the pitch bend decides to change it's portamento speed all on it's own.

Example: Pitch bend up, and it'll take it's sweet time coming back to center, or vice-versa, where I pitch bend up and it'll slowly work it's way up to the pitch I've got the joystick at. Sometimes it'll snap there, sometimes it jumps intervals. Sometimes it does nothing at all. Sometimes it gets stuck between notes.

So I was like "Maybe it's the pot going right?"
I hauled MIDIOX and it confirms that it's submitting MIDI data properly, and it's reflecting exactly what the audio output suggests. (Sometimes it'll jump pitches, sometimes it'll snap to notes, etc.)

I was thinking this was enough to conclude that it's a bad pot.

But then I tried something else: I moved the joystick +Y at the same time... And that magically fixes it. If the pot was actually going, I would think that it would misbehave any time it's in use, but if I use Joy+Y or -Y AND move the joystick left and right, it works every time. EVERY time.

Then I experimented more.

What if I get it to where it does slowly start re-centering, and then barely touch the joystick to +Y or -Y. Surely if it's a bad pot, moving it +Y or -Y will not affect it right?

Well, as soon as I introduce any +Y or -Y movement in the joystick, it immediately snaps to what it 'should' be where the joystick position happens to be.

To me, this is incredibly odd behavior. It doesn't matter what the program or combi I use is. I thought it was a misbehaving engine at first, but that's absolutely not it. Happens with any patch that supports pitch-bending.

What's more interesting: I have several clients whom I work with on the Kronos... One of them is having the *EXACT* same issue.

I'm not a hardware expert by any means, but the fact that Joy Y movement completely fixes the issue seems a little suspicious to me. Doesn't that sound more like a software problem?
_________________
Korg Kronos 88 2, Korg Kronos 73, Kurzweil K2600S
Sound developer, custom sound designer and trainer/Kronos support - www.audora.ca for details!

[tunaman]

Makes some sense to me... sounds like there is a wear band, and once it gets outside of that ‘track’ the contact is normal.

The pot is an electro-mechanical device, and wear can and does occur with use. Given what I imagine your Kronos usage has been, it wouldn’t surprise me that you’ve worn the pot contact down.

[voip]

Sounds like a dry joint or broken wire in the joystick assembly, or its connection to the main PCB. The wires coming from the pitch bend (X movement) section of the joystick are flexed by modulation (Y) movements, so are likely to become fatigued over time. If the connection becomes broken, Y movements could cause intermittent reconnection, hence the observed behaviour.

.

[enigmahack]

[HardSync]

I have this issue on my M3-88, albeit it was slightly different in behavior, with mine getting pitch bend blips when moving the joystick in the -Y axis. I also thought it was the pot in the joystick, so I bought a replacement. Put the new one in and the faulty behavior reversed so that I got pitch bend blips when moving on the +Y axis instead. (Note: The joystick in the M3 is the same as the one in the Kronos.) OK, so mind boggled.

Recently, I took apart the M3 for another reason, and disconnected the joystick cable at the time. When I put it back together, carefully re-seating the joystick connector, I still had pitch blips at power up, but now they were happening much more slower than they had been and a little less often, but still happening, still highly annoying. So I'd have to guess something in the cabling up to or including the connector on the small PCB is faulty.

I'd take it in for proper troubleshooting and service/repair, but there isn't anyone close to me and I really don't want to ship it to someone who can do it. It's heavy, and I don't enjoy moving it around or disassembling it. Anyway... just thought I'd share my experience as it's very similar to yours.

[voip]

The analogue to digital converter, used for "reading" the Kronos joystick signals, has a hight input impedance, and the joystick signal line has a simple capacitor filter circuit on the KLM-3006 PCB. This means that, every time the joystick signal goes open circuit, the capacitor effectively "stores", or holds the voltage from the last joystick position, albeit with a slow drift over time.

It's actualy a little more complex, in that the A to D converter is actually a single device on a chip that has 8 analogue inputs, so a sample and hold circuit, based around an analogue switch and a 35 picofarad capacitor, is used to scan each analogue line and hold its value whilst the conversion process takes place. This means that the analogue value of the previously scanned analogue line will impact on the voltage drift direction of the joystick X capacitor filter, which is 10 nanofarads, by drawing or dumping charge from the filter capacitor. This could be the joystick Y signal, or that from the ribbon or aftertouch circuits, or the positions of four different switches.

.

[enigmahack]

[enigmahack]

[voip]

PS I've added a further paragraph to my previous posting, in order to expand the explanation a little futher.

.

[enigmahack]

[voip]

The analysis of what might be going on comes from years of involvement in designing stuff like this :-) Korg have done a very neat job, design wise.

The joystick assembly is connected to the KLM-3006 PCB via a multiway cable. The voltage values come from the joystick potentiometer wipers and pass, via the 10nF joystick signal line filter capacitors mounted on the KLM-3006 board, and then on to the A to D converter. The A to D converter, which is actually just one functional block in a microcomputer chip that performs many functions, including sensing which keys are pressed, is also mounted on the KLM-3006 board, and contains the on-chip sample and hold circuit, including the ~35pF sample and hold capacitor (it's actually 20pF plus an additional 15pF in parallel). Normally, the voltage signal from the joystick passes to the A to D converter, as is, charging and discharging the filter capacitor as it does so. But, if the circuit from potentiometer wiper to KLM-3006 board is broken or, indeed, the wiper itself is making poor contact with the potentiometer resistance, there is very little to charge or discharge the filter capacitor, so the capacitor just sits there, still connected to the A to D converter, with its voltage drifting, through leakage and whatever else is going on inside the microcomputer chip.

.