Oh, the Korg Polysix. A wonderful piece of vintage awesomeness, except for the unfortunate tendency for the on-board battery to self-destruct, taking the programmer board with it. I’ve done several of these in the past; this was by far the most challenging to work on.
This unit was not unusual in that the battery had indeed leaked, and I could immediately see several components were damaged. It was unusual in that not only did it have MIDI, it had an original Korg MIDI retrofit installed from the factory. I honestly didn’t know that Korg even made an interface for this. I also had no idea how difficult it was going to make the repair.
The first hint came in trying to access the KLM-367 programmer. The retrofit sits on top. Two of the larger CMOS chips that are normally on the programmer are instead installed on the retrofit, with ribbon cables leading back the the original sockets. There are also hard-wired (soldered) connections to the front panel controls, the key assigner, and the voice board. I pulled the ribbon cables, but had to leave everything else connected.
When I first looked at it while the customer was present, the damage didn’t look too bad. With the board out of the unit, though, it was as bad any I’ve seen previously. One of the quad logic chips and one electrolytic capacitor were completely shot. The cap would be removed for the battery modification, but the chip had to be replaced. In addition, many of the traces were bad.
These repairs are not normally difficult, just time-consuming. I removed the old battery and the damaged components, then cleaned the board. Using the component diagram from the service manual, I marked the traces that would need to be repaired. For the new battery, in addition to removing the capacitor there’s a resistor that has to be replaced with a diode. I also installed a socket for the new logic chip. Once all that was in place, it was time to start adding the wires to bypass all the bad traces.
Double-checked everything, installed the new components, put everything back together, hit the power, and…success! Sort of.
The programmer seemed to be working. Kind of. The synth section, though, was not. There were several dead keys. Some of the controls worked and some didn’t. I could create a patch and it would store, but they all sounded horrible. Powered it off, pulled the board, and went over it again. I found one wire that wasn’t soldered correctly, and also noticed that the metal mount for the retrofit was shorting a couple of resistors. A quick dab of solder, a minor change of position, and after I pulled the keyboard and took care of the bad keys it seemed to be working – almost.
This time, the synth section was perfect and it sounded fantastic. The arpeggiator didn’t work, though, and while I could still store patches the behavior of the programming section seemed odd. Realizing that the retrofit might be causing an issue, I tried to locate documentation. A web search turned up a user’s manual, but no schematics. At least I was able to disable the external sync for the arpeggiator, so now the whole synth side worked as it should.
Ignoring the odd behavior, I tried reloading the factory patches from known-good WAV files. No dice. They’d partially load, but then I’d get an error. I ended up entering the missing patches manually. I also verified that not only does the MIDI interface work, but the additional patch storage locations are accessible.
In the end, the customer picked up the unit and was extremely happy, even with the quirky issue saving patches. It doesn’t affect playability at all, or patch selection. I told him that if I can locate the service information for the retrofit, he can bring it back.
And of course the other positive note is that after this one I’ll never be afraid of doing a Polysix battery repair again.
Just had an interesting repair.
Customer had a Prophet 600. The unit powered up correctly, ran through the tuning cycle, and appeared to be perfectly fine – for about three minutes. Then the control panel would lock up and pressing buttons had no effect. If you turned it off and back on, the same thing happened. Obviously, unless you’re in a band playing REALLY short songs, this rendered it pretty much unusable.
The symptoms would generally lead me to think it was either a power or thermal issue. I dug up the schematics and checked all the voltages; they were all fine. The main chips – including the processor – are all socketed, so I popped them all out, checked for bent or mis-aligned pins, and reinstalled everything. No difference. It didn’t appear to be thermal – I got the same few minutes of normal operation whether it had been on for a minute or an hour.
In my day job I work with computers so at this point I’m thinking it’s a bad chip. The consistency of the failure, though, is unusual. In talking to a friend about it at some point in the conversation I was explaining that this unit is microprocessor based…which got me thinking (not as dangerous as it sounds). Once the unit finishes tuning, it has to monitor the status of all the front panel controls, not just the switches but also the rotary pots. Suppose one was dirty? It was conceivable that the CPU could read that as constantly changing, and hang up attempting to deal with it.
I removed all the knobs, removed the two control panel PCBs from the case, and cleaned & lubed all of the pots & switches. Plugged it back in, waited three minutes and then…it kept working. The first time I let it on for a couple of hours, randomly changing patches or moving controls every time I walked by. The second time, it was on for five hours – still no lockups. Re-loaded the factory patches, made sure the oscillators were properly scaled, and it was back to operating like it just rolled off the line.
First time I’ve run into this, but it’s conceivable that any programmable knobs-and-switches synth could develop this issue.
Given that there are plenty of resources available for most of the sound-generating gadgets discussed on this blog, one may wonder why I even bother. Well, there are a few reasons. I tend to run across various oddball models and think it’s worth getting the pictures and demos out there as a resource. Since you can pick many of these up for a few dollars, a quick overview and assessment by someone who’s actually got one sitting in their lap might be of use. Finally, since I’m a hardware geek I like to relate my own experiences and perspective with some of this gear.
This particular piece is one of my all-time favorites.
I mentioned my experience with the IBM PC Music Feature in the K1000 post. Basically, it was a Yamaha FB-01 on a chip. Early soundcards were primarily focused on gamers, and the IBM card was one of the few that you could actually make real music on.
In 1987 the actual FB-01 module, along with the Roland MT-32, were among the first dedicated stand-alone MIDI modules. With limited editing capability, they provided the user with a large assortment of good-quality sounds. But…
Although the concept was the same, these boxes were not interchangeable. Sounds were arranged differently, banks were numbered in different orders, and voice allocation was handled as each manufacturer decided. The initial MIDI standard described how data would be transmitted and received between devices. There were, however, no standards that described what should be done with that data. In 1991, the General MIDI standard changed that.
The MIDI Manufacturer’s association and the Japan MIDI Standards committee decided upon a specific set of voices that would be mapped to specific program locations on any keyboard or module that met the GM standard. There were also specifications for number of voices, polyphony, and many other items. The Roland SC-55 Sound Canvas was the first GM compatible device, released within hours of the official adoption of the GM standard. Not surprisingly, as Roland was one of the more vocal proponents of GM.
The SC-55 module generates sound using both PCM and a variation of Roland’s own LA (linear/additive) synthesis. For compatibility with existing games and other applications, there are also a bank of MT-32 voices. There are 317 instrument patches and 9 drum kits. It is 16-part multitimbral with 24-voice polyphony, and built-in reverb and chorus, individually selectable for each MIDI channel.
Along with the SC-55, Roland released the SB-55 Sound Brush, a basic MIDI sequencer/playback unit the same size and style as Sound Canvas. The two units could be rack mounted together as a complete composition and playback unit.
In addition to the various part controls, the front panel includes a headphone jack with volume control and a very handy front-panel MIDI-In jack. Rear interface consists of RCA L/R audio in & out, MIDI in, out, and thru, and the power jack.
I have yet to run across anyone who actual used it. Regardless of the intended function, the primary purpose of the remote seems to have been to get lost. I do still have mine, not through any particular care on my part but because I duct-taped it to the bottom of the unit.
There are limitations. Since this is in fact the first GM device, the specifications are spot-on. Many of the voices use more than one partial, cutting the 24-note polyphony down.
A common complaint is that the sounds seem “dull”. There is definitely a certain lack of high-end sparkle to many of the voices, but (as always) in a performance situation this is not that noticeable. Listening critically to each voice in isolation, there are other noticeable flaws. The decay times may seem unnatural, and there is some noise associated with several patches.
It should be kept in mind, however, that the unit was design to operate multitimbrally, and this is where it really shines. All of the voices blend together extremely well. The Yamaha TG-100, a relative GM contemporary of the Sound Canvas, has better drums, but files played back don’t seem as polished; some instruments seem out of place in the final mix.
I’ve found that when working with MIDI files in Sonar, I frequently have to reset the Sound Canvas, particularly when using any of the alternate sound banks (power on the unit while pressing Instrument Up and Down, and then Yes to initialize). If I switch songs, it doesn’t always respond properly to program change and the voices won’t change. It’s not a big deal.
There’s a little trick I’ve found with the guitar voices, admittedly some of the weaker ones present. The Steel Guitar has a nice attack and tone, but no bottom end. The Jazz Guitar is fuller, but the attack is mushy. Layering the two, with the Steel at about one-third higher volume, gives a very credible acoustic guitar tone. Of course, in doing so you lose a voice and a MIDI channel.
The 24-note polyphony has been cited as a limitation, but in 20+ years of using it I can recall exactly one time when I had to deal with audible drop-outs due to exceeding this limit. Nowadays, since you would most likely be using a DAW to record the parts it wouldn’t be a problem.
Even with all the higher-end virtual instruments and a plethora of physical hardware to choose from, this is still my go-to device for arranging new material and orchestrating existing pieces. The sounds all feel like home to me.
In part 2, we’ll take a look at some of the other versions of this well-loved box, and hear some demo songs.