I have a one-owner Hammond XB-2 organ - the first of their digitals. Its problem is “odd output” - definitely not Hammond-like. I have a copy of the service manual with schematic and I’ve been trying to run the problem down. I’ve seen in several places on the Internet that the MUSE and DRB chips on the main board - the ones that are actually computing the audio - are bespoke to the XB-2 and they often fail; there are no replacements other than working pulls.
I’m hoping that’s not the case here and I’m therefore trying to rule everything else out. I’ve been poking around with a scope but not coming up with anything yet, so I hoped if I described its symptoms maybe someone could give me insights.
I’ve done the factory reset and replaced the internal battery, which was down from 3V to about 1.5V. When you press a key, what comes out sounds square-wavy with some spiky distortion blended in. I’ve confirmed that it’s coming out of the DAC on the main board like and everything downstream of that (including vibrato/chorus and leslie effects) seems to work. The CPU is getting good drawbar input and displaying the drawbar positions on the LCD display, but they have absolutely no effect on the sound. Polyphony works (although the sound is even more horrid). Turning on percussion does appear to add something to the note attack. All of the pushbutton controls work as expected. But the most specific thing it’s doing is that the octave from middle C on the keyboard upward is a repeat of the octave below and so is every octave up from there all the way to the far right of the keyboard. I feel like that’s got to be a clue of some sort but I have no idea what.
its almost certainly the electrolytic capacitors have time expired. Simply replace them.
If there are any tantalum caps replace them too with new electrolytic.
Here’s why I’m not suspecting electrolytic or tantalum caps. First, tantalum caps generally fail catastrophically and that hasn’t happened. I checked the power supply first thing when I opened it up and whereas the +5V rail was down to +4.8V, I adjusted it via trimpot back up to 5.0V and it didn’t make any difference. I also checked the +5V rail for ripple and found none. That was also true of the +15V and -15V rails.
There does seem to be a 1000uF electrolytic between Vcc and ground near a handful of pins on the DRB chip but like I was saying, if that had problems bad enough to do this it should be visibly damaged and affecting the whole bus.
I’m not saying the electrolytics and tantalums shouldn’t be replaced for reliability’s sake but I really have trouble accepting them as the cause of the symptom I described. In fact, if I shotgun-replaced every one of those caps on the main board and found no joy I will have thrown good time and money after bad.
One thing I have thought to do is to go through all the various data and address busses looking for stuck bits. The final D/A conversion is fed by a single serial data line coming right out of the MUSE chip. I can perhaps try to scope that and see if the C keys that repeat are in fact sending the same pattern to the D/A; if not, then the D/A could be the problem.
Yes Tantalum caps explode in flames, but they can also internally fail leaving no exterior or visible trace. That’s why they’re so hated.
The bus can be checked with a logic probe, you don’t need a scope, but if you have one all your looking for is signs of operation. So long as all cpu bus pins are operating its not likely a bus issue.
If it is, as you suspect those chips in question, then all you can do is find another to salvage from.
You can also try asking at MUSE if they have or know of extra chips around.
I have been poking around the Hammond XB-2 main board with schematic in hand trying to simultaneously figure out how it works and see if I can make any determinations about what is and is not working.
I spent some time on the DAC, which is a common LC7881. I’ve been able to see the serial bit stream going into it but I don’t have a remotely easy way to decode them; if I could show that each C from middle C up created the same bit stream then I would know for sure that the DAC isn’t the problem; ultimately, though, I have to conclude that that is the case anyway simply because the drawbars have no effect. There can’t be any internal flaws in the DAC that explain that. However, one thing that is strange is that the LC7881 is a 16-bit DAC and yet the number of active bits going into the DAC from the MUSE chip looks like maybe 11 or 12 (yellow trace - BTW there are single-trigger snapshots that I’m uploading here):
The mainboard schematic is three pages but here’s the one that shows the MUSE and DRB chips. There is a ROM chip that has 18 address lines and 16 data lines that services the DRB chip; I suspect it’s some sort of wavetable ROM in part because the lower several bits of address lines are performing a binary count; the upper bits are doing a more inscrutable dance. There may be several tonewheel recordings being used and there are also key click and key percussion sounds in play so it wouldn’t surprise me if it’s jumping around from wavetable to wavetable. In any case, whereas the address line signals look just like I’d expect, what’s coming out of the ROM chip’s data lines looks diseased (blue trace):
There’s nothing going on between the ROM and the DRB chips but board traces and a pair of pull-up resistor arrays so if I’m right that these ROM data signals are pathological, it’s got to be either the ROM or the DRB chips or perhaps the common end of one or both of those resistor arrays has somehow come unmoored from Vcc.
Sorry, here’s the schematic page I meant to include:
Just to update, I ran the Hammond’s MIDI output into the MIDI Monitor app on macOS and was able to determine that the repeating-middle-octave thing was not reflected in the MIDI output; you’ll get C4, C5, C6…if you press those keys. Speaking of which, this is a single-trace of what comes out of the audio if you press C4:
The “spikies” are always in motion but the distorted squarewave shape remains constant. As I’m sure you folks realize, Hammond output is a sum of up to nine quasi-sinusoidal overtones; it shouldn’t look like this at all
. I am just about at the end of my rope here; the inscrutable MUSE and DRB chips and the associated mask ROM are basically unobtainium and the first two of those as well as the microcontroller are SMD flatpacks. The audio output seems like it’s loaded with clues as to what’s wrong but working backward from that to the mainboard’s operation is about to stump me. The behavior of the buttons, button LEDs, the two-row LCD display, and the MIDI output are all like nothing is wrong. Drawbar movement shows up in the LCD display; the pitch and mod wheels work but there’s just this weird audio. Oh there was something strange I noticed in the MIDI Monitor app: even with OMNI off, it’s sending on three channels: set output to channel 1 and you get note output in channels 1, 3, and 4. Set output to channel 16 and you get note output in channels 16, 3, and 4. I know Rev. 2 ROM (which this one has - that’s a story unto itself) had enhanced split-keyboard features but with this thing having one foot in the grave I’m unlikely to put much brain into trying to figure out why it acts like this.