Debugging synthesizer. Odd voltage rail pulled from 5v to 4v.

[katastic_voyage]

The background:

So I have a Korg DW-6000 synthesizer that does not work.

- Analog synth (~6 polyphony) digitally controlled by a CPU and memory for programming the PM/FM chips.

- The internal PSU, so far as I can tell, works fine. (Fuse is fine!) It has 3x 5-volt rails, -5, 7, and 11-15v rail. Powered by 7805, 7905, etc IC's all mounted to the same thick heatsink electrically separated with a thin paper layer.

This thing is old. It's early 80's. Through-hole, fairly sparsely populated, modular boards. It shares my components with the DW-8000 synth (simply more polyphony.) So if you google images of either of those, it'll show up.

Now the problem:

- It does not turn on. It flashes '88' on all the LEDs and then immediately goes dark. I've read somewhere that means the CPU has turned on and successfully cleared a bit from the LED drivers. However, nothing happens after that.

- AND, "sometimes" it doesn't even do that. In fact, more often then not, if I flick the power on, it does nothing at all. (There is a pop at the headphone jack as it turns on.)

- The PSU rails come on. But THIS is where I'm really confused. I started plugging only a single board (of ~5 different ones) in at a time. When I got to the CPU board >>the voltage goes from 5-volts to 4-volts<<. Sometimes even 3.9v.

- The resistance between ground and power plane with just the board (no PSU connected) is ~400 ohms. That doesn't seem too low. But I don't know if that's a real diagnostic method.

- The capacitors are all 1980's old. Many ceramic, and some aluminum/can ones. But I don't see any bulges on them.

- I haven't seen any cracked solder joints but I don't have a lot of experience looking for them.

---> ALSO, I have a cheap soldering station with integrated PSU. So I put 5 volts into the board myself (to see if the Korg PSU is faulty). I had to raise it to 5.6 volts to get 5.0 volts at the connector. And that's really confusing to me. I measured with both a multimeter and the solder-PSU's analog current meter, and both say it's about ~0.6 amps. 0.6 amps doesn't seem like a lot. So why would I get 5.0 volts at a connector where I feed the voltage in when running 5.6 volts from a supply? Is 0.6 volts being eating in the wires themselves? Something doesn't add up here. How can I lose 0.6 volts.

---> ALSO, I have a cheap'ish thermal camera which has very poor lens for close up work but from the globs, I can tell 5 chips are warming up. But, they're not warming up much at all. After ~10 minutes they're maybe 90-100 F. The CPU, and 4 SNxxxxxx chips. But I can't tell if that's normal because the board isn't fully powered up, or, if those 5 chips actually indicate a failure mode.

From what I can tell, before I can use this board again, I have to reload the patches into RAM by using a "tape" port (audio in jack). And I believe I found those. However, I don't believe that's the error here because, I believe, the display should power up with an error code that indicates the memory/patches are missing, and respond to button presses.

I know this synth (and synths in general) are an esoteric niche to ask about but the one thing that I know falls into general ECE field, is that strange 0.6v voltage drop.

Thank you in advance, any advice into figuring out this thing would be greatly appreciated.

 

[BradtheRad]

the voltage goes from 5-volts to 4-volts<<. Sometimes even 3.9v.

Could be a failing power supply. Test all diodes. Test all filter capacitors. Old electrolytics are prone to become dry, or leaky, etc.


The synth circuitry could have a short or partial short circuit. Either problem is difficult to diagnose.

 

[betwixt]

Firstly, it is quite normal for a voltage drop in wiring, particularly on power supply wires. It is nothing more than Ohms law, V=IxR where the drop is 'V' and caused by current 'I' flowing through the resistance 'R' of the wires. However, I would caution you against relying on external power sources unless you know they have a clean and regulated output and importantly, that they are turned on in the correct sequence with other supplies in the unit.

The '88' may not be what you think, it isn't safe to assume that it is some kind of internal test or that when it clears, the CPU has initialized the display. '88' means all segments are in the on state so it could just be the way the LED driver powers up sometimes.

These are the basic checks to carry out:
1. Ideally with an oscilloscope but a DVM can be used, check all the outputs of the PSU. A DVM may not be able to tell you if there is ripple on the supply but an oscilloscope will.

2. I'm not sure what CPU is used but most of them, especially of that vintage, need an external clock signal. This almost certainly comes from a quartz crystal based oscillator so you need to check it is running. Again, an oscilloscope is ideal for this, a DVM will not be able to measure the waveform and if you try measuring voltages, the DVM itself can upset it's operation.

3. Again on the CPU, it almost certainly has a reset pin. Typically this holds the CPU in a 'pause' condition until something somewhere lets it proceed. Often this is a signal from the PSU to say the voltages are all OK and stable. you need to find the reset pin and ensure the CPU is allowed to run.

Only when the power, clock and CPU are confirmed running (or at least able to run) you can start checking for logic faults.

Brian.

 

[ZASto]

Just in case, if you did not eant to use any search engine: https://www.manualslib.com/manual/1246821/Korg-Dw-6000.html
Took me only a second.

 

[katastic_voyage]

I checked the clocks. There's like, 3, and they all run at the correct 12 MHz. I dunked everything in an ultrasonic cleaner for the heck of it--to clean some grime off--but there's no changes.

OH, the PSU's regulators had cold/bad solder joints so I reflowed them.

But nothing is changing. I tried getting a MIDI cable and attaching a USB/MIDI keyboard to it and playing it. Nothing happened.

I'd replace the entire CPU/control circuitry with an Arduino (as the analog synth module is a completely separate board). However, from what I can tell, the analog synth board gets digital data that comes from a lookup table for each "step" in playback, so I'd have to completely figure out the communications protocol, and emulate it. I think I read at some point that it uses each of the 5th modules, with time domain multiplexing at twice the playback rate (22.5*2 = 55, IIRC), because each channel has two operators added and I guess they re-use a channel for that. That's unfortunate.

The amount of difficulty in figuring out these seemingly easy, sparsely populated boards, is daunting. It'd probably be as much or more work to replace the control circuitry. But I can't seem to make headway by isolating the problem to a specific board. (With the exception of the standard "make sure the PSU works" route.) Hmm.



---> I'll check the CPU reset pin and re-check the clocks. Before, I recall putting the scope on one and seeing a fairly noisy clock (it repeated exactly, but had lots of over/undershoot kind of noise that didn't resemble a square wave.) I'll have to confirm that since I fixed the PSU.

 

[betwixt]

A 'dirty trick' that might help if it uses the 7810 processor is to remove the EPROM (U2) and replace it with a plug that links all the 'D' pins to GND.
Obviously it won't work as a synth but what it should do is make the CPU think every instruction is 0x00 which translated to 'No OPeration'. As it always sees there is nothing to do, it will keep advancing to the next memory address so you can trace the address and data buses for wrong levels or count sequences.

Brian.

 

[BradtheRad]

It has 3x 5-volt rails, -5, 7, and 11-15v rail. Powered by 7805, 7905, etc

I started plugging only a single board (of ~5 different ones) in at a time. When I got to the CPU board >>the voltage goes from 5-volts to 4-volts<<. Sometimes even 3.9v.

To be thorough you ought to test each regulator IC with a light load (say 1k ohm). It requires disconnecting their outputs. The aim is to find out whether a voltage regulator is failing, or whether the CPU board draws excessive current.

You may need to unsolder a wire from the regulator output, or slice through a copper trace. Later you'll reconnect it. These steps are still in the realm of being easy and simple diagnostics which can be done without a schematic.

 

[katastic_voyage]

Okay, so I just realized I've been doing one thing terribly wrong. Any measurement with my oscilloscope has been wrong, because I haven't used any kind of differential probes. So that may be why the clock and data waveforms looked very messy.

(My DMM measurements have been fine, however, and show 5V, -5V, etc.)

- I have a scope with two channels and standard ground clips.

- The PSU is not grounded. Which is strange to me because the entire bottom case is metal, and as far as I can tell, it's electrically isolated from the circuit boards. So IF that were true, would there be any harm in connecting a ground wire to it?

- Since I've got an isolated powersupply (right?), if I have two probes, I should measure by DIFFERENCE (ala Channel1 - Channel2) with one probe on "common" and one probe on my measurement point, right?

- And if a scope is always connected directly to ground (and I just checked with my DMM between the coax shields and the ground pin), then what are the "ground clips" for? And do they have to be attached? Are they just to force your circuit to reference to ground?

Either way, I tried the SUM method and I still can't get a proper reading. I just see plenty of large sine wave.

I should sleep. I'm not thinking clearly enough for mains AC work!

 

[katastic_voyage]

Wait! I think the 0.6V drop I saw is simple! It's not really there! If I set my "ground" reference point for the scope after the transformer but BEFORE the diode bridge, it's going to have an addition 0.6V drop before it does anything. Right? Pretty sure... I'd measure the diode drop but I just flipped it off and it's still showing voltage from the mains caps so I better wait a moment.