Synth pages
   Module summary
Synth photo
Synth story
DIY modules
   A110 sine shape
A117 clock in
A127 LP/BP switch
A135 indiv outs
A140 self-cycle
A150 switch voltage
A151 2/3 steps
A155 16-steps
A161 clock in
A163 DC couple
IEC power inlet
Plan B summary
M9 bleed-through
M10 envelope
M11 filter
M14 crossfade
M15 sine adjust
M15 others
M17 timer
M28 upgrade
M38A Expander
Diode Ladder Filters
Korg MS10/20 Filters
Moog paper refs
Filter pole animations
Serge VCS
EDP Gnat schematics
Old brochures
PE Minisonic
About me
Photo links
Purple Mountain
Contact details
What's new
Cheltenham Circular Footpath
The Trouble with Christmas
this night wounds time
Behold the Computer
Solartron scans

Site map

I'm giving a new way of using Facebook a try...

Tim Stinchcombe's Synth Stuff


A-117 'Sloppy Clock' Modification

On page 6 of the A-117 Owner's Manual there is a section entitled '"Playable" noise', which suggests driving the 117 with, for instance, an A-111 High End VCO, which will then result in a sort of 'pitched noise', which changes as the VCO output pitch does. My attempts at trying this were frustrating, because it was clear that for some reason the 111 output couldn't clock the digital noise properly, and the shift-register internal to the 117 would simply stop, resulting in no output! The problem could be overcome by buffering the 111 output, but this was just plain inconvenient. At the time I came up with the mod I wasn't entirely sure what was causing it, but it seemed to be that if the rising/falling edges of the external input weren't fast enough, then the shift-register wouldn't clock properly: since there was a redundant Schmitt trigger in the internal clock circuit, I figured that simply moving this to immediately before the shift-register clock input would ensure sufficiently fast edges to drive the register properly - this is indeed what the mod does.

Very soon after I posted this mod to the Yahoo! group, Dieter Doepfer posted that the root cause had been found, and that there was a much simpler modification involving just the addition of a single resistor: see message #4134 at the group (requires group membership!). I did experiment briefly with that suggestion, and concluded that indeed it probably would cope with most scenarios likely to be seen, so the mod that follows is somewhat overkill! (A description of the cause of the problem is at the bottom of the page.)

The basis of the mod: in the circuit there is an astable Schmitt trigger oscillator, followed by 2 Schmitt triggers, which are normalled to the ‘Ext. Clc’ input socket to provide the internal clock signal when nothing is connected. From the input socket the clock signal is switched through a transistor and then on to the shift-register producing the digital noise. The modification moves one of the Schmitt triggers to be between the transistor and the shift-register, thus having the effect of ‘squaring up’ the incoming signal before it goes to the shift-register.

Orientation note: "Top, bottom, left, right"; for the component side, when viewed with the face-plate held downwards; for the solder side, face-plate held upwards.

The above picture highlights the components: locate the leftmost hex Schmitt inverter chip, 40106, and the shift-register, 4006, on the left side of the board (both close to the potentiometer – note there is a second 40106 on the right of the board, this is not affected). Find the 10k resistor (brown, black, orange, gold) running vertically immediately to the left of the 4006 chip. Locate the 1k resistor (brown, black, red, gold) running horizontally below the 4006. Also note the position of transistor Q1. Flip the module over. This is an overall view of the bottom:

And below is closer detail...

...showing where four tracks need to be cut:

1. There is a short track connecting pins 4 and 5 on the 40106 – cut this track.
2. Find the track running between pin 6 of the 40106 and the left end of the 1k resistor. Cut this track.
3.Cut the track running from the 10k resistor to pin 3 of the 4006 (top end of the resistor viewed on the solder side).
4. Pin 3 of the 4006 also connects to the collector of Q1 (the topmost pin as seen from solder side). Cut this track (so cuts 3 and 4 'isolate' pin 3 on the 4006).
(Holding the PCB up to a light should help to determine that a track has been cut.)

Four straps now need to be soldered on:

1.Solder a wire from pin 4 of the 40106 to the left end of the 1k resistor.
2.Solder two wires to the (topmost) end of the 10k resistor (the end which used to be connected to pin 3): one of the wires should be soldered to the collector of Q1 (this re-establishes this connection severed by cuts 3 & 4); the other wire is soldered to pin 5 of the 40106.
3.Solder a wire from pin 6 of the 40106 to pin 3 of the 4006.

Double check that the soldering has not caused any 'bridges' between neighbouring tracks that should not be connected, and that the ends of the wire straps are not touching any nearby pins or tracks (again, holding up to a light may help with this). If all checks out OK, it is safe to switch on and verify that the module will now clock from just about any 'sloppy' input (which should include relatively low frequency sinewaves!).

The problem itself is caused if the module clocking the 117 is AC coupled, as is the case with the A-111. The 'Ext Clc' input is taken through a resistor to the base of an NPN transistor, the collector of which feeds the clock input of the shift-register (and is also pulled up to the +12V rail via another resistor); thus as the input goes from ground to 'high', the transistor switches and the collector goes from 'high' to (nearly) ground. Unfortunately with the AC-coupling capacitor in the A-111 output being connected directly to the resistor on the base, along with the base-emitter junction of the transistor it essentially acts as a 'DC restorer', causing the waveform at the base to be pushed down in a negative direction, so much so in fact that it is only just managing to switch the transistor at all. The resulting rise/fall times of the collector are so tardy that the shift-register isn't able to see the edge! The Doepfer solution of tying a resistor from ground to the input adds a DC path when the transistor is off, allowing the capacitor to discharge, which prevents the whole waveform from being pushed negative.

[Page last updated: 14 Dec 2009]