My dad handed me another item to repair this weekend, a Roland GP-100 Guitar Processor, circa mid-1990’s. He reported that it had been acting strangely, so he replaced the backup battery in the unit, but after that, it wouldn’t boot (screen would light, but then it wouldn’t show anything, or respond to controls). To me, symptoms like this always scream power supply problems, and given the age of the unit, to me that meant capacitors. So I agreed to repair it, and brought it home. After confirming the problem myself (you never know how someone’s AC power can influence issues like this), I started looking quickly for the service manual, which I found quickly at this wonderful site! And quickly found the schematic for the power section of the board. Holy crap, I quite literally explained, they put 25V capacitors on the 21V rails. That’s a pretty small margin, and bound to lead to excessive heating1, and while the ones on the 7V rail are at least better (15V), it’s still a bit closer than one likes to see2. Throw in there’s no ventilation for the enclosure, and these things are bound to be bad… let’s take a look.
So, first thing is, the caps aren’t actually bad. Scoping the output rails from the power supply board shows a pretty solid ±21V, and a pretty solid +7V. Sure, there’s ripple, but after the voltage regulators, that’s gone. Huh. Doesn’t change the fact the caps are leaking, so they’re still going to be replaced. So, I broke out my Hakko FR-300, and desoldered the caps real quick3. The new caps were ordered to fit the circumference footprint, but I didn’t notice the height on the 2200uF caps. So, unfortunately, I had to install them on their side. The 1000uF caps fit fine though. I see why Roland did what they did as far as cap size, but it’s still bad design. The board is easily big enough to space stuff out a bit more and make room for properly derated caps.
Next, I took a good look at the front LCD, and I noticed that on power up, it does show some messages, just very dimly, and can only be seen at a very high viewing angle. After not being able to find the contrast adjustment, I just did a factory reset of the “System” table, and that brought the LCD back. The rotary encoders are pretty crapped out though. Turning them causes them to jump around and in some cases, they’ll jump around even when you’re not touching them. A quick search online shows others have had this issue, and you can pull them apart to clean out the gunk, but since I was already ordering caps for the unit, I decided to go ahead and order replacement encoders. Then after I get those replaced, I’ll try to clean up the old ones and throw them in the parts bin. =)
The encoders were pretty easy to remove with the Hakko as well, and installing the new ones (Alps EC11B15242AE) was equally painless. The only difference between the stock ones, and the new ones is the shaft is longer, so the nobs stick out a bit further than they used to (which isn’t a problem). And they work better (they don’t skip around, or do random things when untouched). All and all, huge win, and the parts only cost about $12 total (about $4/each for the encoders, and about $1/each for the caps). I did test the old caps after removing them, and all were at or near 20% under their stated capacitance, which is within tolerance, but they obviously were wearing out. I’m still not 100% convinced the “gunk” at the base of the caps was leakage vs. really old and crusty glue (the unit dates to early/mid 1995 judging by the datecodes on the ICs).
Thanks for reading!
- Be warned, I’m actually going to include some pictures in this repair [↩]
- Derating capacitors according to the US Navy. TL;DR: at 75C operating temp for a 105°C Aluminum electrolytic capacitor, the max voltage should be 70% of it’s rated value. 21V across a 25V capacitor is 84%, and 7V across a 15V capacitor is ~46%, so while the 15V ones were okay, the 25V ones were WAY under rated. [↩]
- Have I said how much I love this thing? [↩]