SSM2031 HF oscillators drive TOS and tone generator chips (TMS3631). Then synth channels perform waveshaping and it runs as a totally analog synth with the typical ADC comparator cycling through via 4051 analog mux/demux chips to read knobs and compare with previous RAM value from DAC. CPU then steps until comparator flip flops when a change occurs to create an ADC. If a knob has changed from it's memory value, then enter light is set to blink and The ADC comparator tells the processor where to stop correcting the new value and then this value is also being simultaneously sent to that sample/hold cell through the open 4051 once this value is stabilized.
There are six synthesizer channels that each contain a pair of saw converter op amps, SSM2044 VCF, SSM2024 vca, SSM2056 ADSR generator and there are three LFO's with two tied to the same speed and depth settings but a switch to enable/disable their effect on DCO1 and DCO2. Dco2 may also be independently detuned in the programming but also independently affected by pitch wheel and the mod wheel also is configuable to either affect LFO1/2 or LFO3. LFO3 affects either or both the VCF (with two waveshapes; square or triangle) and PWM for again either DCO. A very configurable machine.
The Expander has the same analog hardware but no knobs except for DCO B detune and global volume and tune. Patches much be loaded from tape or one at a time from a dk600 midi connection. And no pitch/wheel data is received!! Very sad that they couldn't complete that aspect on the expander. Yet one of the real strengths of these instruments are the keyboard sounds and having a dual stack or split is a very useful feature so the expander is still to be desired since these are a unique analog creation and 12 voices of analog is AWESOME even w/out pitch and modulation control.
Pots tend to go bad and make the thing think you are changing values all the time! Battery spills cause a lot of havok often. I saw the CPU even go down in one. Use great care when cleaning and keep yourself static guarded! I still don't know how I killed that CPU.
The mysterious problem I had on and off for some time; thinking I had cured it at various times...with garbled noise emerging particularly on high velocity hits appears to be solved! Sadly...the effect emerged about the same time a pot took a dive and some other things. What a cruel hoax it was as I look back on all the things that led me astray on this one. But anyway, the 5V supply sagged to 4.8V WHILE in the MANUAL it says this is ok, it may not be. The 4051 chips I was getting were maybe a hair more picky than the ones Siel used about voltages being passed that are right in the area of the power supply for the chip! I installed a trim pot in series with R6 on the power supply board. Trimmed to 5.05V on the plus side and it retrimmed the others and it works great. USE A HIGH precision meter and really trying to get those voltages all to .005V or so. If they are off you'll have to be retuning patches and so on. Beautiful synths when they work right. Never seen a problem on the analog board. Bruce Forat (formerly with Linn Drum) said he's found that the 4051 analog switch VARIES from brand to brand and he recommends the CD4051 or the original vintage motorola I'm sure is also fine that was in the siel obviously. I put in a Phillips recently and it had so much leakage you could see 40mV sawtooths on S+H outputs. Which makes an audible modulation of the VCF for instance.
5/06- In my latest episode with an expander, I've seen some very strange things. This machine went down after working perfectly for a while after a battery replacement/cleanup proceedure. I found that the 74LS154 address decoder that drives all the enable lines to ROM, DAC, etc. was only putting out a signal for EPROM1 (Expander only has one EPROM whereas there are two in the DK600 keyboard) Addresses were all moving so I went ahead and just swapped a chip in and it worked!!..for...a few minutes. Then started cracking up just like before. Err..well almost like before. Before I actually could change patches for a while but no sounds would come out! Then it degraded to not being able to do anything and then finally only a '.' in the display instead of 00. 06/06- IC22 was dragging a line low it seemed after cutting traces. However it went right back to it's old trick after trying to play a song! The chip now socketed still measured good...and in the process of checking things I found again that taking out the trace patch restored data entry! Bizarre. It leads me to think maybe it's IC17 this time so socket that chip which measures fine. And..wow. Now the thing just keeps working. I also scratched some debris under IC22's socket that could have caused the problem when I had IC17 out. The boards are ok but not great. The holes are a little tight and it's hard to get solder out. A nicer removal tool than solder sucker would be good. I've never gotten wicks to work on these boards.
6/06- Working on yet another mainboard, find that R18 went intermittently open which can cause the IC16's front end to float high all the time so as to ignore interrupt signals from DAV1 (front panel switch held) and DAV2 (Display related switch held). These should be triggering the cpu interrupt input pin 17. Generally it's a good idea to just replace all those parts and verify all traces in the area due to battery leakage. But this machine has yet another problem. All S+H's are high even though DAC out looks fairly normal. Thus we get slow envelopes and max noise. Similar to what can happen if the dac buffer just fails w/ high output. It's so hard to monitor all the digital stuff simultaneously with my setup that I just replaced the op amp and comparator to no avail after trying a different dac. Let's see, input from the knobs is rock solid. But DAC out has this jittering which obviously correlates to something telling it 'no we need to go higher higher. It's not the CPU as we swapped that. Not the SRAM "". Hmm.
Kbr>2/07- And another problem mainboard. I started out cleaning up the battery area and rigorously insuring every connection and impedence in the blast zone of course. Then found it has a weak dac buffer output and the TL082 is split at the seam..couldn't see until desoldering :-). But wait..there's more! The ADC comparator was out too which restores knob response. Not enough? IC20 was eliminatiting many LED's and replacing that 74LS377 restored those. And now..we have a synth that doesn't store! It responds it seems to the 95[enter] command to turn off protect. However hit record and you see data coming to the main board but the LED never flashes and nothing gets saved! Oh I do find the *fun* ones. Well, I'd think possibly this is an SRAM problem since it has to set a flag somewhere in memory. And that's the only memory I know of on these. I had a 6116-4 handy. Don't do that :-). Unless you like an additional lfo that never goes away and very segmented envelopes. Hehe. Sonically beautiful with the original SRAM. Try another SRAM that's the same. ANd same result. Try another cpu. Same. Try the new eprom. Same. It actually acts like it's saving a patch but then you reboot and it's really gone. I don't get it. (12/14..did I ever solve this one? seems I did since that unit isn't around anymore. sorry)
12/14 Delving into DK600 land again. I've got three mainboards I'm going at it with. After resolving multiple issues as usual I had one all working perfect and it was ready for sale. Got a down payment and got home to prep it and it's dead. Just white noise coming out. Well I see the DAC signal looks fine and it's reading the from panel knobs in the stream I can see. Holding them just fine. But the S+H cells are NOT getting refreshed! Odd because all of the address lines seem to move fine and the inhibit lines are getting their 20usec few tenth volt pulses that should enable the chips to open the gates and pile charge ont he capacitors.
After comparing some signals I see that the VWR signal is not correct. We can see knobs moving which means the DAC and ADC (ie. the cpu and comparator) are working. However I see that the DK600 works this way: It gets the voltage from the knobs and stores in memory when it determines the values. AFTER ALL of them are accumulated THEN it makes another pass and retrieves each the 16 knob voltages in the order top to bottom that they are shown on the schematic diagram, storing them to the S+H capacitors as it goes through them. Thus is solved a great mystery. :-) Now we know that when you get DAC output and knobs are having an effect on the signal but no values are getting updated, it's most likely going to be because the memory chip is not getting read for some reason. Chip enable? Bad chip? etc.
Interestingly I had tapped on it and it all just started working for a few seconds..then failed again and I haven't seen it work since. It eventually turned out to be T3 the transistor that switches the Enable SRAM line active when the power good signal is achieved at power up. But in the process I prepped up another board someone wanted to have repaired, thinking I could check some things on it and find a difference. However, sadly it wound up with exactly the same problem it appears after getting it where it would behave stably on the panel switches at least. I finally resolved that this one had a bad trace which caused the SRAM chip not to enable.
On yet a third board I found another useful tidbit. After doing a big clean up on a KIWI version 293/186 serial, it still wouldn't run the program. DB3 line was struggling to produce signals over .3V in height! Something clamping it to ground but..what? All pull up resistors measure perfectly good! I replace IC3 the tranceiver and no difference! No variation in measurements to ground or 5V compared to nearby lines and no connection to anything else one might things was a solder bridge. Odd fail mode. I finally pulled IC15 which was a bit rusty anyway being near the battery. It measures bad! And now the CPU is trying to run the code. IC15's functions are still not working so good..buttons flickering; turning on wrong things etc. Entering numbers gives double number. 2=22 etc. The pull ups were dirty...those should always be done since they're right by the battery. However that made no difference but I found the bad trace finally and also it turned out that IC20 was also bad.
11/07- Mainboard for James LePera. This one took the prize I think. After cleaning up a bit more find the following problems:
* Bad pin 12 under IC16
*Bad IC11, IC8, IC9
*Bad sockets on IC13, IC14, and DAC! (Acid on this one just...ate stuff for quite a ways!)
*Not only trace problems around resistor pack nearest battery, but internal short in it!
*Bad IC30, IC34, IC17 intermittent, then IC5 heats up after replacing leaky C15 and D5!
*In the process of testing this thing finally found R85 open on one of my analog boards which caused a variation in the attack as I recall. What a gnarly repair that one was.
I have several scraps from Siel units and lots of the discrete parts.
Contact Sound Doctorin'
Also someone suggested this company for switch caps/buttons. Rapid Online
Manual summary (For those familiar with synths in general. ie. the stuff you won't easily figure out.)
1) 95 [enter] = 'protect off'. This must be performed when you boot the machine if you want to save patches. You can hit it at any time, for instance even after you've already gotten into editing then realize you can't save the sound until you turn protect off. Normally, when protect is off, you hit 'record' after editing a sound. The LED's flash and you simply enter the patch number where you want the edit to be stored. (Also note that after a patch is edited, there is a "." next to the number)
2) 96 [enter] = 'MIDI CH SELECT' (But only on later ROM versions!) After pressing that code enter desired channel (00 - 15), ENTER. This data is not saved when the machine is powered off.
2) 97 [enter] = 'split'. The DK600 may be split such that the upper keys of the split point play the local sounds in the DK600. The lower keys will produce midi messages that can be used by an expander or whatever else you might hook up I suppose. I've not tried this yet but a cool concept in that you can actually at least SORT of have a 12 voice DK600 this way! An awful lot better for certain kinds of sounds!! Anyway after you hit 97 [enter] it will display '00'. Then key in '01' and then briefly hit the key where you want the split to be. Hit [enter]. To restore to non-split mode simply hit 97 [enter] again then it will display '00' again and hit [enter] without doing anything else.
3) 98 [enter] = 'dynamic adjustment'. This function when selected will display the current dynamic 'curve' in use. You may punch in 01, 02, 03 or 04 to adjust to a different level then hit [enter]. '01' is the highest dynamic setting and '04' is the least dynamic.
The tape interface is pretty straightforward provided you have some way of knowing your signal level. There should be near 0dB signal going to and from tape or hard disk record. I've had good results using my PC to store sounds rather than a tape recorder.
Anyway make sure audio cables are attached appropriately to ins and outs of tape interface on DK600. Turn the 'tape' switch near them on. This disables the DK600 as many have found out the hard way :-). Select the function you want to perform (to tape, from tape or verify). You should always Verify before assuming that the information is valid so do that after you complete a save!
Anyway, once the function is selected...for 'to tape' simply start the recording process and hit 'enter'. A lead in sound will go off for a bit and you may adjust your volume if need be to get 0dB or a hair less on the meter. Then the data will transfer and when it's over just shut off the tape, rewind, then select 'verify' (ie. both lights on) on the mode switch. Start the tape and hit enter on the DK600 AFTER you hear the first audio signal coming through (assuming you have a monitor of some kind..). If a recording is useable by the DK600, then the verify LEDS go out and the enter blinks. Otherwise you don't have a good recording or levels aren't right or something. Check cables. Check levels. Try again. (This advice cost me many hours the last few days. I finally gave up! Two good machines and I can't get either one to load sounds or verify what my studio one loads out! "From tape" is basically identical to verify except you use the 'from tape' mode obviously.
MIDI is crude on this early MIDI synth. Omni always on with early versions. Recommend using a midi filter if you have a network of midi machines so you can address this one a single channel. OR scoring the EPROM upgrade that allows you to turn off Omni after power up. The only useful thing you can do otherwise is dump a single patch to expander.
WIth all protects off (95 [enter]), set DK600 to "midi int" mode. Then select patch you want to xfer. Expander to "midi ext" mode. Hit 'record' on DK600 and expander display should blink. (If you have a midi cable between them obviously...) Now select the patch on the expander that you wish program to store to. Press [enter] ON THE DK600 and it should transfer.
NOTE: The above is from the manual. I think I may have tried it with expander once and it worked..or..maybe not. One thing for sure though. I could *not* make it work with two DK600's and I found out why. The quickest way to xfer patches is to do the initial steps up to where you put the expander in External mode. Ok we'll stop saying 'expander' and say 'target machine' now to generalize this. You take the 'source machine' and put on BOTH int and ext lights. With tape off of course. Then you can select the patch you want to transfer. The target machine will also change patches automatically. WHEN YOU HIT "record" on the source machine...the most interesting thing happens and I'll have to research when I get time. My midi analyzer shows only that a controller message gets sent of this format (hex) B0 57 7F and that's it! However...when the midi cable is hooked to another DK600, as soon as that record gets hit and that message emerges, something unseen happens on the midi cable and the entire patch information is transferred. The lights continue to flash on the source machine until you cycle through the interface options back to where they are both turned on again. THen simply select your next path to xfer and repeat hitting the record button. It takes a second or two for the patch to load! But it works. The instructions above did not work.
If you put the source machine in 'int' mode, then hit record so the lights flash there, then put it in 'ext' mode and his 'enter' the machine sends a sysex message. I tried EVERY which way..to get that blooming message to load on a target machine from computer. It would lock up the machine many of the ways I tried to where voices continually sounded! As in even when you slam all the keys down or change patches!! Boy..talk about quirky :-)
That is most of the information about the DK600 that I know that is not fairly intuitive to anyone familiar with synths. I really wouldn't waste time with the manual beyond this I don't think. Unless you need to be..told how to plug in the AC cord and stuff :-). ONE OTHER 'trick' though. If you start with a cold initialized patch after a dead battery for instance....I've noticed that the LFO's are way slower than they are supposed to be programmable to be :-). And..you can actually save that setting...it just isn't possible to dial it in! I've made some of my sounds using the super slow lfo afforded therein. -Bob
SERIAL Numbers: I've seen serial numbers starting with 330/ that were omni only here. 365/ are poly it appears. Of omni I have retired #216, and poly I own #11, 204,retired #271 and am selling #850. It therefore seems highly unlikely that they made more than a couple thousand total of these. Send your serial number!) I see Tycho owns another 330/221 and 292/280. So at least one more series exists it appears.And I got 369/40, and 292/171 (and /135)just appeared on ebay so there must be another series of even earlier ones than I've seen.6/09- I got 365/113 and 365/202. And there's 294/140.. Another series on ebay. And someone said they own 369/280. Hmm. Interesting. 280 is the biggest number we've now seen twice as the second number except for the 365 series. I'm thinking it was probably the only one that had a 1000 ish in it. 293/186 on a KIWI version in Croatia! Perhaps a few hundred of those also out there?