JX3P keyboard fixing

After a quick attempt to fix my Midiverb II again (and failing) I pulled the JX3P out again to have an attempt at fixing the dodgy octave.

Thinking that it was possibly the keyboard decoder, I replaced IC45, which took a fair amount of time to do it cleanly with the desoldering vacuum.

JX3P - removing IC45

While I’ve got the board out…

Crinkly JX3P circuit board

not sure I really like the look of these crinkly paths on the back. Anyway onwards.

Changing IC45 made zero difference, so it wasn’t that. And anyway, I could see the signal appearing on the connector on the keyboard. It took me a while to work out what was going on, and it definitely made it easier yanking the keys out.

JX3P keyboard diodes

This is part of the dodgy section of the keyboard, and it looks a lot like my unhappy Poly61M from the same time. The rubbery circular things are pressed down by the keys – inside the dome is a carbon disc that presses down onto the board and completes the circuit.

One side of the switch connects to the top of the diode below it, with the other half connecting to the decoder IC45 via the keyboard connector at the far side. Pressing the switch should make the signal from IC45 appear on the other side of the switch (the top of the diode).

I can’t believe you’ve got this far, but aaaaanyway, the signal wasn’t coming up on the diodes for the broken section.

The soldered track on the far right here should have given the game away

JX3P keyboard connector and track fix

…essentially the track had failed for some reason. Scratching out the soldermask from either side of the break and soldering a wire across it sorted it. It wasn’t actually super-obvious where the break was from just looking at the track, it took me a couple of test scratches to work it out.

There was also one intermittent key which just needed the rubber switches refitting, and that was sorted too.

I still have lots of screws missing (in the JX, that is…), and the transformer hums like an electricity substation, but at least it all works now.

No Comments

JX3P MIDI in fixing

I bought a rough old JX3P a few years back for not much.

programmable preset polyphonic

It sounds surprisingly nice – whoever had it before me had programmed in some pleasantly woozy sounding patches into bank C and D.

The chap I bought it off was upfront about the MIDI in not working, and I naïvely thought they might have just not checked the switch that selects between the PG200 programmer or MIDI in.

Haha, no.

Tried the usual swearing, different cables, different sequencers, different swearing, nothing.

Let’s have a look inside, there’s a few things wrong.

Battered JX3P internal overview

Look at this transformer! It hums like a bastard, hence this odd arrangement to try and dampen it.

Wonky JX3P transformer

Needs replacing really, but it works OK.

Something heavy must have fallen on it at some stage and smashed into the last button on switch panel, it’s still a bit wonky looking.

Wonky 16 button

Here’s some wiring to fix the cracked part of the panel switch board; good effort previous owner. The “16” switch LED nearest this calamity doesn’t work any more but on a quick check, the connections beep out fine.

JX3P switchboard extra wiring

On holding an LED to the two contacts on the back of the board it happily burst into life. Closer inspection revealed that the impact had just pushed the LED pin down and popped the joint, you can just about see it here.

16 button LED dodgy joint

Reflowing sorted it.

Let’s play the keyboard, oh dear. Well it mostly works, some are a bit intermittent, but keys 17 – 24 don’t do anything despite heavy bashing. I’m pretty sure this is a recent problem.

Usually I’d assume that the key contacts are dusty (hello Poly61 and Polysix), but it’s a bit suspicious that exactly eight keys don’t work. It turns out that the keys are grouped in eights for scanning, so I’m suspecting either wiring or the demultiplexer IC45.

But I’m not looking at all that now, I’ll do nothing else this weekend otherwise.

Here’s the DIN board.

JX3P broken DIN board

Looks alright. Let’s switch the switch a bit, my Poly61M has godawful switches. Nope, nothing.

The optocoupler is the first thing that the MIDI connector hits, so maybe that got fried. I bought a replacement PC900 then thought should really test it rather than just blindly replace it.

As far as I could tell, I was getting a signal on the other side (pin 4). Booo. Here’s the schematic for the DIN board.

JX3P DIN board schematic

The switch for MIDI enabling is grounded for when switched to memory protect and programmer, and is unconnected when set to MIDI in – that line (RX MODE) is then pulled up by the 5V line through the 10K resistor at R5 – so high = enabled. The 10K resistor was definitely getting 5V, but the other side of it read as 0V, whether or not the switch was set to MIDI in.

Having temporarily chopped the jumpers at W4 and W5 to make sure nothing was pulling it down further down the line, all that was left was the 74LS00.

For some reason I’d bought a couple of these eight years ago – no idea why – so here’s the bosted one:

Duff 74LS00

…replaced it with a socket and the new one

JX3P DIN fixed

and now it works!

Score one for methodical troubleshooting rather than just blindly swapping stuff (which is never the answer, really).

I lashed up a slightly ropey CV to MIDI converter to play the JX from my MC-4, and then belatedly realised that it’s not possible to edit the synth from the front panel while it’s accepting MIDI input. Think maybe the Kiwi3P upgrade (or more likely the Organix MIDI expansion, given the cost of the Kiwi3P) might be on the horizon, it’s a nice sounding synth.

Done for now, anyway – although I’ll come back to the keyboard soon. Ish.

No Comments

Roland System 100M noise/S&H/ring modulator

100M 150 noise/s+h/ring mod stripboard

Out of all the various dodgy looking stripboards in my 100M clone, this one turned out the best (let’s just not mention the LFOs, they were a pain in the arse).

Determined to get it going first time, I made sure I’d done all the cuts, checked for shorts between tracks, powered up the board without ICs – was generally super careful, and then only later found that I’d missed out a couple of capacitors. Silly.

I made a daft mistake with the hold in the sample and hold. It worked, but the held voltage was drifting downwards rather quickly. Compared it to my System 700 board and realised my mistake – I’d used a 2.2nF (222) capacitor rather than the specified 22nF (223).

(Actually I was tempted to use System 700 boards rather than faffing with stripboard, but they’re so lazily massive that they didn’t fit in the panel space I’d designed).

I compared the white noise to the pink, and the white noise just sounded really quiet. Which is weird because it’s coming out of the same place, it’s just being filtered differently. Here’s the relevant bit of the schematic.

Roland System 100m noise circuit

I compared the outputs on the oscilloscope – didn’t take a photo, but the white noise was chopped off at 0V, whilst the pink danced around both sides of the ground point as you’d expect.

Anyway, it’s pretty obvious from looking at the schematic that the R31 (attached to the transistor output buffer, Q4) should be connected to -15V, and the track layout bears that out.

100M noise track layout

The ring mod is based on the LM1496 like the rest of the Rolands, and you’ve probably got three of those already, but anyway.

Here’s a tiny demo of clanking and white noise twiddling. The ring mod was in better calibration when I put it together, it’s bleeding a bit from one side at the moment.

There would be a demo of the sample and hold, but I’m presuming the lag pot has disconnected itself behind the panel, as it’s verrrrrrry smoooothed ouuut, barely even recognisable as a sample and hold right now.

Here’s the layout, and wonky BOM.

Roland System 100M 150 noise/S&H/ring mod stripboard layout

It includes the indicator for the clock out, which has separate ground and +V power supply connections, presumably in an attempt to keep the audio path clear of any clicks from switching the LEDs. On the original (and on mine) this is powered by 22V, but it should work off 15V.

No Comments