Access Indigo 2 Repair
I recently purchased a non-working Access Indigo 2 synthesizer. It just showed a row of boxes on the display and did not make any sound. I knew it was possible that I wouldn't be able to repair it, but decided to take a chance. All of the control circuitry is contained on a daughterboard that mounts on the back of the front panel pcb. It seemed that the fault was probably on that board. I contacted the manufacturer to inquire about a replacement board. They replied quickly, and told me that replacement logic boards are no longer availble for this model. They suggested sending it to one of their two USA repair facilities, in case the problem was not on the logic board. I decided against this, as their nearest facility is 1400 miles away, and the Indigo 2 is pretty heavy due to its solid construction. Also, I was pretty sure the problem was on the logic board.
The first things to check were the power supply voltages, especially since there are a number of posts on the web about units being repaired by replacing failed capacitors. Unfortunately, all of the capacitors seemed OK, and the supply voltages were all correct. There are five chips which make up the heart of the Indigo 2 control circuit: An 8051-based "C515" master control chip, a 4MBit Flash chip, a 128K-byte SRAM chip, a PAL chip, and a Motorola/NXP 24-bit Audio DSP chip. I was pretty sure the issue was related to one of those five chips. There are a number of smaller chips as well, but I didn't think they could cause this type of issue.
The first thing I tried was changing the SRAM chip, as I had a similar part on-hand. The RAM chip is socketed, so it was easy to swap in the other part. Surprisingly, this actually made a difference. With the new RAM chip, I got a message on the LCD indicating that RAM contents were invalid, and then the unit seemed to boot and load a preset. Unfortunately, there was still no sound, but I could move pots and switches and see values changing on the screen, so it seemed that the main control chip was working. I ordered two SRAMs that were identical to the original part, as it seemed to be faulty, although not the main problem.
Next I removed the Flash chip and read and verified its contents several times to see if it was working reliably. It seemed to be fine. I found an image on the web of a newer version of the OS, erased the Flash chip, and programmed it with the newer OS. Maybe the Flash data had gotten corrupted somehow during an OS update. Unfortunately, updating the OS made no difference. I could still boot with the newer RAM chip, but there was no sound.
At this point, I realized that the only way I might have a chance to repair this unit was to learn a little about how it worked. I assumed that there was code specific to the Indigo 2 in a ROM inside the 8051 chip. So if that chip was bad, it was "game over". I decided the best way to proceed was to draw out the parts of the schematic involving the five main chips. Here is an archive containing the schematics that I created. I also drew out some of the the front panel circuits. These schematics were made using a DMM and my unit, and may well contain errors. They are not complete. Use them at your own risk! I had assumed that the 8051 would first execute code from its internal memory, and then jump to code in the Flash chip. As it turns out, the EAb pin is tied low on this chip, so the 8051 is always executing from the Flash chip. This meant I could replace the C515 chip if necessary. So three of the five main chips could be replaced if necessary, although the C515 chip is in a surface-mount PLCC package, so replacing it requires SMT soldering experience.
I checked the outputs of the PAL chip and they seemed to be working properly. Signals were pulsing, with good logic levels.
Next I turned my attention to the DSP chip. Was it possible that it was also running code from the Flash chip? As it turns out, I discovered that the 8051 reads a large chunk of data from the Flash chip at boot time, and writes it one byte at a time to the DSP. The DSP stores this data in the three SRAM chips connected to it, and this is the code that the DSP executes to make sound. I monitored the signals that the 8051 uses to write to the DSP with my oscilloscope, and could see that in my unit, only a small amount of data was being copied from the Flash into the DSP. Instead of ~128K bytes, it was about 250 bytes. So the 8051 would start downloading the code to the DSP and just do a little bit, then it would stop. With only a small amount of its code getting loaded, the DSP had no chance of working. At this point I was sure that the C515 main control CPU chip was defective and I ordered a replacement.
While I was waiting for the C515 chip to arrive, I decided to use the oscilloscope to monitor the signals used by the DSP and the 8051 to talk to each other during the download process. What I discovered is that the DSP was accepting data normally during the first 250 bytes or so, then all of a sudden it stopped accepting bytes. It was telling the 8051 "no more", and so the 8051 tried a few times to get it to accept more data, and then gave up. Now it seemed that there was nothing wrong with the C515 chip. That left only the DSP chip, and I wasn't sure I could replace it myself, as this is in a 144-pin SMT package and the lead pitch is 0.5mm, but I decided to give it a shot. I have worked with SOIC chips for years, and even some with finer pitch, and I have a fine-tipped soldering station. This DSP chip is still available, although it is being phased out of production.
Although I have a hot air rework station, I decided to use low-temperature solder to safely remove the DSP from the board. By adding this solder to every lead on all four sides, you can keep the solder on all leads molten (with just a soldering iron) for long enough to remove the chip. There are a number of YT videos showing how this is done. Of course what's important is to avoid damage to the pc board pads in the process. Once the chip was removed, I cleaned the pads to remove all solder, cleaned the pcb with alcohol, and added flux to the pads. Next I placed the new DSP on the pads and got the leads aligned perfectly on all four sides before tack-soldering two opposite leads to hold it in place. Then I soldered each lead individually with my fine-tipped soldering station and fine solder. (This is not how this type of chip is normally soldered, but it's the method I felt most comfortable with) Then I removed the few solder bridges I had created betwen adjacent pins, and checked all pins for shorts to the adjacent ones with my DMM. (My DMM has leads with very fine tips, not the ones that came with it, or this would have been impossible)
Replacing the DSP chip has fixed the problem and the synth is now working normally, so I just need to learn how to operate it!
Important Warning:
Soldering or unsoldering a 144-pin QFP package requires the right equipment and experience. A lot of the work must be done under a microscope. Unless you have been working with SMT IC's for years and have the necessary expertise, it would be a mistake to attempt it. You will certainly damage the pcb and probably make it impossible for someone else to repair your unit.
Also, DSP chips have a fairly low failure rate compared to some other parts. If you have a unit which does not boot, or even one with no audio, the problem is most likely something else. Please do not attempt to replace your DSP chip unless you have really good reason to and the necessary skills and equipment.
I will not be responsible for any
damage to any instrument caused by either proper or improper use of the
information presented here.