Roland SDE-2500 Modulation LFO VCF Broken

SOLVED?

I think I found the problem. The problem is that there is no problem. That wiggle in the modulation is actually normal.

This type of digital delay adjusts the clock to just read audio straight out of ram faster or slower which makes it sound like the pitch changes. But it's using the same clock to write the data as well. Meaning the data is being written faster or slower which changes the pitch of the tone being recorded.

Fortunately I'm not really interested in playing tones through it.

UPDATE:

Here is a video but with sound so that you can actually hear the warble as it descends pitch:

Roland SDE-2500 Modulation Broken

and another, again a pure fixed tone, this time from the scope but looking at DOUT of the master controller (yellow) and the DAC output (blue):

Roland SDE-2500 Modulation Broken 2

So am I correct that the pitch modulation and clock should move exactly together. Maybe it's just hard to see on the scope? Is there some clever way to see the rate of change on the scope. Like XY plot with a fixed square or something?

--8<--

I got a very clean Roland SDE-2500 but the modulation is totally broken.

Here are "waterfall" FFT plots rotated 90 degress of a 440Hz tone through the SDE-2500:



So the internal modulation is very broken. And even the external sine wave is pretty bad. The external triangle wave is worse.

Note: External Fn gen is an old Protek but makes very good LF waveforms.

Has anyone worked on this before? Does anyone have a manual?

I found service manuals for the SDE-2000 and SDE-3000. Here are the interesting bits:

SDE-2000:



SDE-3000:



Just for posterity, here are closeups of the two main PCBs:

Board ASSY74121080

Warning: 4MB


Board ASSY74121100

Warning: 4MB


The SIP packages are BA662 OTAs which is a big clue as to where the relevant circuits are.

Any suggestions as to what the problem is and how to fix it?

Here are "waterfall" FFT plots rotated 90 degress of a 440Hz tone through the SDE-2500

Click to expand...

This definitely does not tell me much - can you explain, either in words, or using simple scope pics?

Jakob E.

gyraf said:

This definitely does not tell me much - can you explain, either in words, or using simple scope pics?

Click to expand...

The SDE-2500 is a digital delay but it has rate and depth modulation controls on the front that can sweep the delay forward and back. If you put a pure tone into it, you'll hear the pitch go up and down with the rate and depth. Nice for chorus effect. Probably nice for synth.

But I can hear the sweep is not right. To visualize this, I used the "spectrogram" plot on the computer to plot frequency spectra over time. So those plots are just spectrograms but rotated so that X is time and Y is frequency. The first one is supposed to be a triangle pattern I believe. The second two are using an external CV for which there is a jack on the back that bypasses the LFO. Clearly it has an effect but not completely which means the problem is partially but not completely the LFO (or the CPU is feeding bad control currents to the OTAs).

I will use the scope today and poke around but getting plots off of the scope is painful.

It does look like it related to the SDE-3000 circuit. Here is a little higher res shot of that circuit:



I can see the IC47 and IC48 are the BA662 OTAs (those two SIP packages in the "jack board" pic) and the 1uF cap with the two transistors controlling the bias currents. I suppose I could reverse the schem a little. I really need a schem.

I have made lot of progress but I haven't nailed the problem. I've traced the circuit this far:



So apparently the OTA used to control modulation depth runs through a JFET that switches off modulation entirely, a buffer w/ bias control and resistor to a connector (pin 20). The output is a very good triangle wave running between 1.5 and 3V (as the depth is reduced amplitude decreases and reaches zero at 1.5V bias). Then it goes over a wire to the main board into the FC (frequency control) pin of the VCO IC SN74LS628. So the ~0.5 Hz triangle running 1.5V-3V is converted to a poor excuse of a square wave at ~5MHz which is completely consistent with the SN74LS628 datasheet (available everywhere). Then it goes through an RC filter and into the MSCK pin of the main controller RDD63H101. At this point the square wave is quite round (almost sine).

This topo is very similar to the SDE-1000 circuit which is here with overlays:



Unfortunately it's difficult to tell if the frequency sweeping of the MSCK signal is incorrect. Meaning as the CV voltage on the VCO FC pin goes up and down, the VCO is putting out a square wave that varies in frequency (~5MHz) in reaction to the CV. But it is difficult to tell from watching the scope if the rate of change in frequency is varying in a triangle fashion.

However, test point T2 is interesting. That shows a PWM signal although it's not just PWM, its also frequency modulated. Meaning it's a low duty PWM that looks like it gets stretched and shrunken as the CV goes up an down. And that does NOT move in a "triangle fashion". It is very smooth in one direction and then as soon as it reaches the crest, it moves too quickly back in the other direction.

So unfortunately it seems the problem may have reached the main controller.

I'm going to recap the electrolytic, tant and ceramic next to the main controller because the electrolytic looks like it has lots of black goo around it's base (you can sor of see it in the upper right of the pic).

Any advice is appreciated. At least maybe someone else will find this info useful if I don't.

Solved: See very top.