Everyone Loves My Big D

The D-70 I got a little while back may not be small in physical stature, but in terms of cultural impact it really hides in the shadows of the D-50. While I don't have a D-50, what I did manage to pick up for the sweet price of "for parts or repair" is this lovely D-550, which is electrically identical to the D-50 but in a rackmount formfactor instead of being a self-contained keyboard.

It's not entirely healthy, though. It seems to have gone a little senile and lost all of its presets.

It's probably not because it took a bit of a bump to the head, but I'm sure it didn't help things.

That part is easy to fix, though, so I'll just take the rack ears off and remove the front panel.

Ok, the LCD window is coming out too, I guess that's fine. In either case, it's time to get medieval on this thing.

It didn't take too long to get things massaged back into place.

And while the LCD window is out, I think it could use a little cleaning.

Yes, that's looking much better after washing it.

While it's open, let's have a little peek inside.

Ok I'm definitely going to vacuum out that disgusting pile of hair in the back. But aside from that, let's check the battery.

3 volts is enough to save the SRAM contents, and usually the battery would be dead by now, so I wonder if someone tried to change the battery, got confused by why the synth no longer made noise, and then shoved it into a closet? Hard to say.

Looks like it's got a 1.0.1 ROM.

I think 1.0.2 is the most recent. I might bake a new ROM at some point but for now I don't see much reason to.

Anyway, I'll just do a quick factory reset to clear out the garbage in the SRAM and then reload the factory patches via MIDI... except the patches aren't loading. The synth just sits there waiting at the bulk transfer screen doing nothing. Worryingly, I'm not seeing the MIDI activity LED blinking when I connect a keyboard either.

I checked the physical MIDI functionality by connecting a cable to the MIDI-Thru port and that worked, so electrically that part of the circuitry is functional. I checked that the MIDI signals are making it all the way into the main CPU, and they are, but with the MIDI light not blinking, that points ominously to a CPU failure.

I mean it couldn't just be that the MIDI LED is burned out, and the MIDI I/O on the CPU is actually working fine.

You know, this MIDI LED right here, which suspiciously isn't conducting any electricity through it.

You know, just for giggles, I should probably try replacing this LED with a known working one. I don't have a pack of 2x5x7 rectangular LEDs on hand, but a 3mm LED will do in a pinch. It won't fit through the front panel slot but it'll only be temporary.

And oh hey, what do you know, it's working after all.

It looks pretty ugly with the face plate on, but again, a temporary fix.

So then why isn't the SysEx MIDI transfer working? Well, it turns out that Roland, in their infinite wisdom, decided to invent a protocol on top of MIDI that does a silly 2-way handshake with the remote device, which no modern SysEx tools know how to deal with.

But the solution to this is just as silly: if you hold down the Data Transfer button it bypasses the handshake protocol completely and you can blast the SysEx data over without any issues.

And there we have it, all the factory patches are loaded.

But importantly, does it make noise? Why yes, yes it does. Kind of scratchy noise, thanks to the volume pot being a bit scrungly after sitting in a closet for who knows how long.

But as per usual, that's nothing a little contact cleaner can't fix.

So while I wait for some parts to come in, I'll stick some feet on the bottom of the case so it doesn't scratch holes in every surface I place it on.

And then I'll deal with the packing material.

Usually styrofoam peanuts are the worst to get rid of, but fun fact: these aren't styrofoam.

And they're water soluble.

Fast forward to the next day, and we're going to deal with more Roland silliness.

This is an IEC C9 connector, more commonly known as a pignose cable, and it's used almost exclusively on Roland audio equipment from the 1980s. Ok technically it was used on more equipment than that, but this is where it shows up the most.

What's irritating is that it's ungrounded, even though the same equipment it appears on was sold with grounded IEC C14 connectors in other markets. Luckily that also means that the pignose connector uses the exact same chassis cutout as the C14.

First we clip off the soldered power wires and find an M4 nut and washers for the lonely, unused grounding stud.

Then we make a ground wire pigtail with a ring terminal on one end, slather up the other end with flux, and hook it through the ground terminal.

Using lots of flux, lots of solder, and a high temperature is critical here to get the solder to flow out quickly without melting the ground pin out of the connector.

Next up, it gets screwed into place and the pigtail gets bolted down.

Then the hot and neutral get some 3/16 spade terminals crimped on for easy servicing in the future.

And now it's no longer a high-voltage death trap.

Another bonus of it being the next day is that the 2x5x7 LEDs have arrived, so let's stick one in.

Except hmm, that's not good, one of the solder pads fell off when I was pulling the temporary LED out. I'm pretty sure I didn't overheat the pads, though.

Well, no matter, there's plenty of lead length on the new LED, so I'll just scrape off some solder mask and fold them over to solder them down to the remaining trace.

I think the problem was that the original LED was installed a little long, so it poked out just slightly past the front of the front panel, making it an inviting target to bump. That bump would go straight down the leads and push against the underside of the solder pads and... wait, hang on a second.

Oh hey, the original LED still works after all, it was just the broken trace that messed it up. Well, I sure wish I'd found out it's supposed to be green before soldering in a red one.

Hmm and speaking of, that's sitting a little too deep for my liking now. Putting a few washers on the back of the board would space it back out nicely, but the question is how to hold them there while I'm installing the board?

The answer is, apparently, "use a dab of not-so-super glue."

And now it's sitting perfectly.

Speaking of sitting perfectly does anyone else think that these LC pi filters are a little uhhhhhhh...

They could have made them any colour they wanted to, and this is the colour they chose.

Anyway, looking back at a reference photo I snapped earlier of how the original LED was spaced, it's pretty clear why it would be vulnerable to being shoved back into the unit.

If it had been sitting against that plastic support, then the force wouldn't get transferred up the leads. But I guess then it would sit too far back from the front panel. Oh the trouble Roland could have saved by putting a few washers on the back of the board.

Well anyway, does the new LED work?

Why yes, yes it does.

So, that wraps up bringing this D-550 back to life. I wasn't expecting quite this level of fuss but the important thing is that I got there in the end.

I Can SE Clearly Now

When I picked up my Boss SE-50 and SE-70, the displays on both were doing pretty poorly. The SE-50 had a backlight that was much more of a backdim, and the SE-70 had a completely roasted polarizer and so the text that the LCD was trying to display was basically invisible. I had combined the two displays into one fully working and one fully failed display, but that left me one short.

Luckily new displays that work in these units are less than $5 and are a pretty easy swap, so I upgraded both of them to white-on-black displays.

And the results look pretty great.

Flashing Floppies

So I installed a Gotek on my Roland S-750. It worked, but it didn't automatically boot the system image when powering on. I had to select the image and then it would boot, which is peculiar.

I spent a lot of time tracking this down. I made extra sure that the flashfloppy config was absolutely correct and pointing to the right image file. I even copied the exact same config from my Roland S-760 that was working correctly and booting automatically. But no dice.

What eventually ended up working was setting pin34 = nchg in the FF.CFG file, though I'm not entirely sure why. In the schematics, it's quite clear that pin34 is active-low, and so "nchg" should be correct.

Or, well, I suppose "D_CANGE" is supposed to be active low. The overbar is clearly visible, at least.

But what gets me is that it's always supposed to be active-low, and I don't really understand why flashfloppy very clearly sets the pin to active high for the IBMPC mode.

This is very peculiar. I'm glad it works, I guess, but it feels unsatisfying that this contradiction doesn't really make sense.

It's All About Control

As mentioned in my previous S-750 blog post, my unit is reading a ghost-mouse on the EXT CTRL port, which results in a phantom cursor that slides down to the bottom-right of the screen at all times, disrupting the UI control. This is due to IC14 on the mainboard failing, and while I don't yet have a fix for that IC, I can at least try to work around it.

First off, the simple test: does bypassing the blown fusible resistor help?

Well, it would be nice, but no it does not. It still detects a phantom mouse and will not speak to the RC-100. So, it's time to get drastic. Let's have a look at the backside of the connector.

It's a bit grimy with flux from when the board was put together back around 1991, but other than that nothing looks too amiss here. I did see what looked like a blob of conductive something that might have been shorting pin 8 to ground.

But cleaning it off, as I expected, did not yield any improvement.

So the drastic step I'm going to take here is to completely disable this port. I'm doing this by shorting out pins 1, 2, 3 and 4 to ground. These are the pins used by the MSX mouse to report its movement, and by shorting them to ground rather than leaving them high, the movement will be reported as 0x0,0x0 instead of 0xF,0xF. We'll still end up with a mouse cursor, but it won't move, and thus won't interfere with the button-based navigation.

It's not the most beautiful solder job, but it'll do.

While I'm in here I'll also pull the blown fusible resistor, since it's not doing anything at all useful.

And then leave a note for whoever opens this next wondering why the EXT CTRL port isn't working and half the pins are shorted to ground.

Also while we're here, since this is now in a usable state, let's get rid of the floppy drive.

I do actually have the physical boot floppy for this, which is nice, but I'd rather not have to rely on a 35 year old mechanical drive. So in goes a Gotek.

Conveniently the set of thumbdrives I got to use with the Goteks was a 3-pack, so I only needed the floppy emulator itself.

Next up, the feet need some love. The front two are in good shape, but the rear two have exited the chat.

A little heat and scraping gets the remains of the tape off.

And after cleaning the remaining adhesive with some alcohol, the new feet go right on.

Weirdly this set of feet came in a set of 18, which is very much not divisible by 4, so I'm glad I found a use for the remaining two.

On the good news front, the video-out works, and gives a much better view of what's going on than the built-in front panel LCD.

The composite-out circuit is a bit primitive, though, and only displays a B&W image. It looks a bit fancier in colour, like when I hook up to my S-760.

Though you do only get 8 colours, since the display has only 1 bit each for red, green and blue.