This blog deals with the fixes and mods I did to my Roland D-70.
2013-10-06 When I got the Roland D-70, it was in used and unserviced condition, with some problems to linger around the corner.
The lovely thing about the Roland D-70 is that it is completely serviceable. The mechanics and electronics are simple, low-tech, understandable. There are schematics available from Roland! Like many of the consumer devices ot the 70's and 80's were accompanied with their schematics. And there are spare parts available.
2013-10-06 First problem was a few keys no longer working perfectly, electrically and mechanically.
Cleaning the contacts on the PCB and the rubber strip with isopropyl alcohol fixed the problem.
While dealing with the red glue from hell problem below, I was bending the metal springs back into shape. Some keys needed a second iteration for a good result.
Oddly, there are noticable differences in the gaps between the white keys. Can't say if the gaps were there before, or if I should have paid more attention to place the keys to their exact original position. After shuffling a few keys, it now looks better. Odd.
2013-10-06 Sure, when I got the Roland D-70, it had already some 20 years. And sure enough the red glue problem started to show. The first weights from two white keys had just fallen off.
Not yet aware that the red glue is slowly dripping, I first tried to fix each weight in two corners with a bit of hot glue. That turned out to work somehow, but it's quite a mess and looks ugly. Takes loads of hot glue sticks. In hindsight, it was a bit a silly idea. Should have stopped earlier. Good thing is, nobody will watch the keyboard from below.
Having a closer look, I wanted to scratch off some red glue that was dripping down using a screwdriver. That didn't work well. Removing all the red glue seemed like an extreme, perfectionist task. I didn't come up with an easy way to remove the weights and glue from within the keys.
So I decided to cover and seal the glue with a new layer of yet another kind of goo.
I did apply slow hardening epoxy with a syringe into the gaps to cover and seal the read glue, and at the same time glue in the weights with the epoxy. Working on 76 keys with epoxy is only half the mess compare to hot glue. The epoxy gets transparent after some time, which now looks quite nice. I would recommend this way of fixing it. The only worry I have is how long-time stable the used epoxy will be... but that's for the next generation to find out.
And maybe next time I should do some research first, how others tried to fix this mess. Although, at first I was not even thinking that anybody else would have that problem too...
2013-10-14 It's the inner values that count. Bought a pair of EEPROMs with Roland OS v1.19 (the latest available) to replace the v1.14 ones. I think it was around 20 bucks, from ebay.
2013-10-14 The slider potentiometers had accumulated too much dust inside. Moving them was produding static noise. There are spare parts floating around on ebay.
Another infamous problem is the constant whining of the LCD backlight. This had to be fixed, someday.
While checking the compatilbility of replacement LCDs I came up with the idea to replace the black, sturdy flat cable and the strange connectors by a nice, standard flat ribbon cable. A 20 pin header on the new display, and two 20 pin IDC box headers on the mainboard, with the right pins removed. Just one advice: when pulling out the old cable sockets, check that you didn't destroy any of the PCB traces... BTW, the PCB layout and connector names CN2, CN3 in the service notes probably belong to an earlier version of the mainboard. Seems like they changed the physical connection to a top mounted through hole connecter at a later moment, requiring to swap the odd and even lines on the 10+9 pin connection.
2019-11-01 Buying a replacement display was easy. AFAIK there is only one factory left on this planet that produces them. I was opting for the Black-on-White.
The old display was around 7mm thickness, all the new ones are 9mm.
Type caption here
Type caption here
LED backlight supply (white should be K, GND...)
LED backlight supply (white should be K, GND...)
Upside down bracket and spacer
box header is quite tight...
The backlight LED can be run from 5V directly. The LCD module conveniently has a 30Ohm resistor integrated. At 5V it draws a maximum current of 120mA, so it's not a big deal to tap the Vcc pin to feed the backlight as well.
Well, at 5V the backlight is quite bright, and most of the time I like a moderate setting. So I came up with the idea to regulate the voltage.
As the inverter module was gone, there now was some free space to put a potentiometer and a tiny Emitter follower circuit, and recycle the 2-wire backlight power cable of the previous LCD module. The Emitter follower has a dropout of around 0.7V, so there is around 4.2V max voltage at the LED of the LCD backlight, which is still sufficient to me. The LED's start lighting at 2.5V, so you need a 1k potentiometer connected to 5V and a series resistor of 1k to ground.
LED backlight at 4.2V
LED backlight off at 2.2V
LED dimmer circuit
LED backlight dimmer
Why? because I can.
Reset switch with 5V light
Reset switch is shorting the cap near the reset IC.
There is no need to remove the mainboard battery. Oh, uhh, too late. Unfortunately, there is no easy "press the factory reset key combination" -- why didn't they come up with that?
You loose the RAM contents with the sound configuration. AFAIK, the factory set of PCM waveforms are still in the six ROMs on the mainboard, but the configuration contains pointers into the PCM data and assembles them into Tones, Patches, and Performances. And the names of these sound settings are gone too, leaving you with non-readable stuff on the display.
Two options to restore the factory settings:
1. If you are lucky you will get hold of a RAM card containing the Roland D-70 factory settings. It's said that from time to time (every decade?) such a card could show up on ebay, but I have yet to see one. Source: TheMusicGuyUK.
A RAM copy of another D-70 could possibly work too, but I don't own a RAM card and don't have access to another D-70 to test this hyopthesis. This leaves us with the other option:
2. Put the synth into "Bulk Load all data from MIDI" mode.
This will load SysEx messages from a plain .syx file, or from a MIDI file containing the same SysEx messages wrapped in a playable MIDI file.
For me, the Bome SendSX at slowest possible speed, and D70_ORIG.MID file did work. I didn't succeed with the d70_orig.syx file or other software (MIDI-OX), but didn't try very hard.
Looks like if you connect two D-70 you could Dump the data on one, and directly Load it on the other D-70.
You will find the details and files at roland.com and on the net. A possible starting point is this blog here: Roland D70 Factory reset
This is another infamous problem every D-70 owner should be aware of, before you destroy the underlying plastic mechanics by pressing the button ever harder.
After the LCD finally worked -- PCB traces repaired -- and keyboard was assembled back together, I noticed that some of the microswitches were close to failing completely. Annoying, but meanwhile I am pretty quick in disassembling down to the LCD board.
The 57 tactile micro-switches are on three single layer boards. Luckily, the switches are standard, dirt cheap and readily available. They are 6x6mm with height of 5mm. Finally, a good use for my desoldering gun, a Pro's Kit SS-331, pimped of course (power switch on front, quiet fan. Though the back of the heating element tends to clog fast, probably because its not hot enough.)
Inspired by bradthx's youtube video on restoring a Roland D-70, I took apart a replaced switch, side by side to a new switch. Clearly, they failed due to oxidization, not because of wear.
There is a bad crackling on the headphones output when the power is switched on or off. This is especially annyoing if you feed that signal to a pair of active monitor speakers.
The schematics show that there are some muting measures taken by Roland on the direct and mix out. A quite clever circuit is designed to clamp the signal by a transistor until the supply voltage is stable, and to immediately clamp on power loss.
When hooking into the same muting control signal, I didn't get satisfying clamping from my small signal transistor BC548. So I was thinking of using a relay, and designed a control circuit around it.
The same MUTE signal from the power supply board, having a 100Hz ripple of 6V at around 20V, is used to slowly charge a capacitor of 1uF and after about half a second pull the relay to open the output lines. The 11V zener diode gives an offset to the capacitor voltage until the darlington transistors switch on.
On power loss, the MUTE signal will pull down the voltage within a few milliseconds through the diode. The relay is fascinatingly fast and has a drop off time of 1.3ms. I put two of these low voltage relays in series to use the 12V supply next to the circuit.
The 1uF and 10kOhm on the left are the capacitor from the power supply board, and the integrated base resistor of Q7 of the original clamping circuit.
Here is a quick stripboard hack attached to the audio board: