Simmons SDS 7

— are the 8-bit 8-32KB EPROM sample player implemented with DAC0800 going into an analog 4-pole lowpass filter from Curtis, the legendary CEM3372 ‘synth on a chip’. Several synth manufactures depended on this chip at the time, among them the incredible Matrix 12 by Oberheim. SDS 7 also make use of this chip's 2 controllable mixer inputs and VCA circuit.

— the ability to store presets

— are the analog oscillator build on CEM3340, sharing the same CEM3372 filter. It can do frequency sweeps from 20 to 6000Hz. Only triangle waveform is utilized.

Other popular and well reputed synths using this oscillator chip are Roland SH-101, Jupiter 6, Memorymoog, OB-Xa, Prophet 5r3.

• click generator
• envelope generator
• LFO
• 3 pitch up generators

If you’re looking at the voice card wondering where all those 15 CV signals connects to, they actually share one common CV line on pin 12.

Address for the local multiplexer on each voice card are received on pin 5, 7, 9 and 11.

For each CV value CPU toggle pin 4 LOW for the receiving voice card through IC6/Bus-Board (74LS154). Independent if there is a card in the slot or not CPU board will burst out all 15 CV values to each of the 12 slots continues at speed around 7.3kHz, or every 7.3 millisecond if you prefer.

CV value in the range of 0-255 are read from the memory board in same order as layout on the panel and decoded in the central 8-bit DAC IC4/CPU board Ferranti ZN428E. Then send to pin10 non-inverting input on IC4/Bus-Board (TL084), from this op-amp output pin8 it comes as control voltage in the range 0-2.5V on bus pin 12.

Starting at decimal 15 each step here increases by +15. The 16. (or 1.) value seems to be a calibration value set at half the max CV voltage, here at 1.23V. It is not user programmable. Notice how pin 4 is held low for the duration of this step also.

With 16 controls (incl the empty one) on each of the 12 voice card the CPU board are updating every sample and hold circuit on the voice boards approxly 38 times a second.

On my unit master CV max at 2.46V, meaning each of the 256 CV controller steps differ by 9.61mVolt.

Located on the bus board between the Stereo Bus Summing Amp and the 4-Line to 16-Line Decoders we find the noise source common for all the voice boards on pin 6.

SDS 7 will play any EPROM you can fit in the 28pin socket. That is not to say it will play the sound correctly, although it might sound interestingly. To do so you need to tell the sample engine what size EPROM is inserted by solder two wire jumpers on the Analog Digital board.

Officially SDS 7 where released with 'locked' eprom sizes. Bass and Snare 8k, Toms 16k and Cymbals 32k (early boards have 2x16k).

For Dual Digital cards Simmons corrected their fault implementing dip switches, making it possible to sell the consumer single sounds at 76$ a pop, thats less than 1 second worth of noise.

Wire jumper closest to the EPROM socket hold pin 27 on the EPROM at 5V to ensure that EPROM type 2764 and 27128 doesn’t activate program mode.
For 27256 (and 27512) pin 27 function as address A14, wire jumper connects this pin to pin 13 (Q8) on binary counter IC10 (CD4040). Silk screen on pcb is confusing doe to an error, please see picture below for correction.

Wire jumper furthest away from the EPROM ensures the sample player stop after one complete sample has played. Signal is tappet from IC10 (CD4040) pin4 (Q7) 2764, pin13 (Q8) 27128, pin12 (Q9) 27256 (for your 27512 modification pin14 (Q10), wire jumper then connects anode side of D267 before going to the clock generator.

SDS 7 do not care if you use 27 series or 27C EPROMs. When buying go with 27C as it is more likely to play well with your cheap programmer. Reason is the old 27 series generally need much higher program voltage (21-25V) than its modern 27C counterparts (12.7V).

To confirm this general rule this picture show an exception labeled D27256 with program voltage 12.5V.

Simmons voice cards where from a time 27512 EPROM must have been few and scarce, if even availably. Today their cost is similar to 27256, but maybe more interestingly we can find them as eEPROM saving us the tedious UV light erasing of normal EPROMs.

In Michael’s great article on how to convert our Analog/Digital board to hi-hat he also show how to modify the card to accept 27512 EPROM. Google translation from German has a lot to be desired so I ended up reading datasheets to understand what’s going on. Two short wires is all it takes:

27C512 pin1 [A15] to IC10 (CD4040) pin 12[Q9], this enables one more address pin thereby doubling max sample capacity.
IC10 (CD4040) pin 14[Q10] to wire jumper furthest away from the EPROM. Think of it as fourth option for 4k, 16k, 32k, and now 64k, or solder directly to anode side of D267 and remove this wire strap as Michael demonstrate in his article. This pin enables correct stop position for the counters, letting the sample play to the end.

Early SDS 7 are infamous for its high noise floor. On my first SDS 7 in the ’80 mono/stereo outputs were practically useless due to high clock noise. Simmons tried to fix this with Noise Canceling which actually are very efficient. Older units can be modified similar way as Simmons did.

I did an observation while doing finishing tests on my SDS 7. Mono and Stereo out still have some digital clock noise. This noise is not possible to adjust with the noise cancellation VR on the buss board. While testing if the noise where coming from a defective voice card I noticed the noise are actually coming from empty slots. While direct out for slots with voice card inserted have normal hiss, noise signature are identical for all three direct out without voice card inserted as noise summed and amplified on the mono/stereo bus.

This is somewhat surprising as every direct out are terminated to ground with a 100k resistor on the bus board and a series resistor on the connector board. As far as my notes show stereo out from voice cards comes from the 4k7 pan pot (center tap to ground) over two 820ohm series resistors located on the voice card (R19/R20) function as mixer resistors for the stereo bus. Without card inserted there are no summing resistors connected and the PCB traces for the stereo bus are floating.

I’m clearing my work bench for another project at the moment, this issue are going in the pile of 'things to do'.

I decided to remove a cap to take measurement of a tactile switch to search for replacement. To my surprise I found they can be taken apart and cleaned. Now be warned. I didn’t know this and spend best part of the day on the floor searching for the all the small part inside the cap that I dropped.

Inside the cap there is a spring plate, a contact disk and a tiny spring. Contact plate and contact surface soldered to the pcb have a thin contact coating of some kind. If studied under a magnifying glass one can see this alloy is worn with blank spots where contact points meet.

Using my preferred method I applied Deoxid to both the disk and switch surface and let it soak for 10 minutes. Just to find out Deoxit will dissolve the alloy used in these switches.

For the rest of the switches I used Lotoxane cleaner. Inside of the plastic caps and plastic part mounted to pcb where also cleaned.

To make sure the switch doesn’t land on the same contact points I very gentle squeezed the long sides of spring plate between my fingers. When writing this I'm thinking I rather should have taken apart more switches and mixed parts between them in attempt to accomplish this.

Big improvement in performance. Too early to tell if it will last long enough compared to invested time doing this fiddly job.

Often seen in adds for second hand SDS 7 voice cards are reference to LED’s as a way to tell if the voce card are in working condition. But can this be done safely? Short answer is ‘NO!’

Prog (Program) is used to indicate that CPU will allocate all voice parameter adjusted on the control panel to the memory location representing the slot number where Prog LED is lid up. Both cathode and anode have dedicated pins on the voice card going to a 7407 LED driver on CPU board. Meaning this LED can’t tell anything about condition of your voice card. It can however tell if some parts of the CPU board is functional if you can turn the wheel and select any of your installed voice cards.

Trig (Trigger). Although giving a good indication if the card is been triggering the name is somewhat misleading. This LED is driven by the envelope circuit on the voice card. Value of Decay parameter should change how long this LED is illuminated when the envelope is triggered.

Yellow Backlight. I got one not lightning up, I’ll come back when I figured this one out. I suspect it just goes to the +5V power line through a resistor.

Turned out to be a contact problem for the 5V power line on the edge connector. When board are powered you can measure 4.6 to 4.8V from center pin and the pin above it on the header connector for the LED PCB.

I wanted to better understand the relationship between sample rate and pitch settings on my SDS 7. Table below show measured clock speed going to the sample engine in relation to pitch setting on panel.

I wasn't quite able to reach 44100Hz, adjusting value of R222 could fix that if needed.

* This is the slowest oscillation I was able to measure with my gear. Oscillation seems to stop at display value 50 (+/-3).

Please understand all values are approximations as it is impossible to lock the clock speed too exact sample rate with only 8-bit control voltage resolution and no calibration for referance voltage for the clock oscillator. Also the speed vary a lot. The are two reasons for this, first when the clock oscilator ramp up and second the Bend function is never completely off. Last could probebly be fixed with a trimmer for R65, but I think it would be tricky to calibrate and probably not stable.

For Windows 10 users Audacity 2.1.0 http://audacity.sourceforge.net/ seems to have the conversion tools needed.

To convert audio file to .bin compatible file for the EPROM programmer select File > Export Audio
from the files dialog select Other uncompressed files, then press the button Options. Set Header: RAW (header-less) Encoding: Unsigned 8-bit PCM

If you like to open a Simmons .bin file for editing, File > Import > Raw data. Select Unsigned 8-bit PCM and manually fill in the sample rate.
"But I don't know the sample rate" I hear you say. I use trial & error until it sound right in the editor, starting at 22050 and 24000Hz then check if sample count equals file size.

A way to add silence and correct length to samples

Manually edit files

Since I already use Notepad++ to create this web page I only had to add the HEX puggin from the plugin manager. Now I can read and manually edit .BIN files before burning.

Combining samples

Software for my EPROM programmer let me program the EPROM in portions. But I prefer combining all the sounds to a single file before transferring it to the programmer.

From console windows (cmd on Windows) we can combine .BIN files with the copy /binary command like this:

copy /b sound1.bin+sound2.bin MySound.bin

What's special about Unsigned PCM?

As nearly everything I talk about I’m no authority on this topic eider. But my understanding at this time are that it’s only the way we interpret all the zeros and ones we write or read to/from any type of memory or data storage.
With 8-bit Unsigned our data are referred to as numeric numbers between 0 and 255. Signed values are -128 to 127.

If you’re like me you mess around with unknown audio files in WAV editors and accidently find yourself with a visual representation of an audio file where there are no data below the Zero line. Opening a unsigned file as signed could give result like that.

RAW Headless just means the audio file is not an container where information about the content of the file like audio sample rate, bit depth or typical tags like name of song etc. are present. Which after all are pointless in an EPROM device where firmware usually are hard coded and know what to expect at any predefined address.

This 16pin IC socket used as cable connector interconnect sequencer inputs to the edge connectors for the voice cards. Pins are connected logical as a normal IC, pin 1 to channel 1, pin 2 to channel 2 and so on. Pin 13-16 are not used.

Looking at the PCB traces and where the connector is located one are fooled to believe triggers also connects to the CPU board. But they are actually running underneath the CPU edge socket. Easy way to hook up an internal sequencer or MIDI trigger if needed.Possible why Simmons took the trouble with the long ribbon cable.

This mystery module labeled Hybrid found on each voice board is responsible for decoding the single CV line coming from CPU board into 15 individual control voltages for the voice board. Burst refresh rate is approx 7300 times a second.

On the backside there are 8 VCA (voltage controlled amplifier).

There exist an EPROM adapter card for the Analog/Digital voice board called 2085 daughter board which can hold 4 EPROMs. Friendly members on the Yahoo Simmons list helped me piece together historical and new technical information for this board. Now posted with the EPROM modification

After marked memory cartridges has been made as a personal project where some units where produced and sold to community members.

The blue low profile Zero Insertion Force Socket shown in some of the pictures on this page where actually sold as accessory to simplify eprom replacement.

69 sounds on separate eproms where officially release according to an English flyer where SDS EPB (Simmons Digital Sampler E-Prom Blower), MTM and SDS 7 are on display.

These rubber 'feet' at the center rails for the back pane inside the case lid where retrofitted later in production. Mine is dried out and no longer flexing, I cut some springy packaging material I had laying around and glued in.

On my unit I've discovered that Pitch and Bend raises ever so slightly when entering edit mode.

Doesn’t matter which parameter I enter edit mode for. As all cards have the problem it seem like the CPU board do something funky and alter the reference voltage for the control voltage DAC when in edit mode.

Working with normal and shorter percussive sounds this isn’t intrusive, I’m guessing it’s simply part of the charm with the SDS 7 when I try to edit longer drop tones with low resolution samples.

One easy way to check for this is to set a long Decay time, lower the pitch of the digital voice until the sample nearly stop playing. Now try stepping out and in of Pitch adjustment several times and hear if the sample run faster when in edit mode. If so, try entering edit mode for any of the parameters and listen.

When sharing presets or copying presets between cards there are no guaranty you can duplicate the sound 100% accurate. The nature of being analog and why so many love the sound of analog synthesis are the inconsistency that comes with it. Typical precision of components are +/-5% for resistors and +/-20% for the caps used throughout the SDS 7 circuits, as in most synths and commercial electronics of its time.

I found that my 5 Analog/Digital card differ so much between each voice card I have to take into account +/- 3 for many of the settings I want to copy from one card to another. Tuning two cards to play together in harmony with long decay can prove difficult.

One more for the backlog, modify R222 (82k) and R206 (150k) with trimmers for calibration of voice tuning.

This is how SDS 7 output a square waveform sample, 55Hz at 8000 sample rate generated with no aliasing in Audacity.
Waveform keep this shape independent of pitch. Filter are full open and Modulation speed and amount to zero.

Actually SDS 7 doesn't start the sample bang on when triggered, it fades in starting out as a sinus waveform.

This layered screen dump from my scope try to visualize two audio pulses from direct out with identical settings except decay. Sample is square waveform 1000Hz 441000 sample rate played with high pitch. Click is set to 255 seen as a single line in the beginning of the pulse. Left pulse Decay=0, Right pulse Decay=255. Sadly the Decay do not only change the length of the tone, long decay time also soften the attack. This is also true for the analog oscillator, clearly a function of the envelope generator.

Here are a few Simmons compatible .BIN files I’ve compiled with Audacity, ready to be burned with a standard EPROM programmer.
When original sample contributor is known credits are found in the ‘readme.txt’ found in the compressed ‘zip’ files.

Underneath the rubber surface we find the actual internals for the selector pad glued to the top of metal chassis. It's a 4x4 membrane switch consisting of two layer of plastic film with electric conducting surfaces which short as blade switches when pressure are applied.

Kudos to Ed Rose for right to use photo

Sadly they're nearly impossible to repair if/when oxidation start to damage contact surfaces. One possible repair route might be to use single thin film membrane switches to replace the damaged area.

Simmons made the MTM module to handle trigger to MIDI and MIDI to trigger. Below is my poor mans DIY attempt.

Work in progress, not tested yet. Note that quick and dirty test platform with Arduino on a Nano board will only give 5 Volt trigger outs. A proper MIDI interface with velocity will need DAC's and op-amps for level shifting.

My Arduino sketch so far DrumTrigger.ino

DO NOT BUY ONE!

TopWin6 website is long gone and development stopped.

TopWin6.exe only support Windows 32 bit. I keep an old XP machine around for occasions like this.

I’ve measured output from 34063AP1 step-up converter inside the programmer to 12.47V. Program voltage on EPROM chip is 12.4V. I consider replacing a resistor in the feedback loop with a trimmer to tune it closer to 12.75V.

Will only work with 27C cmos series EPROMs with max 12.7V programming voltage.

For EPROM STMicroelectronics M27C256B-12F1 / M27C256B-10F1 select generic 27C256 12.7V under unknown brand. Under Config set Delay to 1ms. Despite what you might think, longer delay do not work.

While researching pin configuration for EPROM sizes I come to think of Akai S700 12bit sampler I had back in the late ’80. More specific I thought of some great drum sound I sampled from pieces of my acoustic kit inside a big concrete car park.

This sampler used some rare 2.8” floppy disk called Quick Disk. I remember two packs of 10 empty disks set me back more than a week salary. With its whopping capacity of 128k bit that usually meant one disk/one very short sound. Think of it, in comparison a 512k bit EPROM with 8bit samples in SDS 7 can almost ’last forever’ without looping. If we only could find a way to change samples easily.

Another historical fact that popped up on the wiki, the invention and first commercially use of floppy disk and EPROM are both dated to 1971.

Demonstrating use of gate signal with SDS 7.

Gate are activated using a 9V battery connected to a micro switch into sequencer input. Pitch shifting is all done by SDS 7 bend/modulation preset without external CV control.

Be prepared for unpredictably behavior. I've not figured out how the envelope trigger with gate signal. With long deacay you can hear the sample clock generator runnings for lengt of sample without envelope opening. Sometimes I get click similar to Click generator even when its set to zero. This seems to differ between voice cards.