While chasing my full Doepfer system dreams, i picked up an A-160 for rhythm generation. It's good for four-on-the-floor action, since it's a binary divider. But also for subharmonics - the A-160-2 is no good for this, since it uses a microcontroller and will undersample, while this thing uses a humble CD4024 binary counter to do its job, and will never have the nasty clock limitations. Since, y'know, the incoming signal is the clock. I got the module, and then stuff happened, and i moved, and then my system spent some time dusting up in a corner.
Upon reassembling and servicing, though, i faced a horrific discovery not explicitly listed on any official Doepfer resource from what i can see: this module, as the Doepfer know-howser Raul Penna puts it, is a "mathematical" divider, not a "musical" one. In more technical terms, that is an upwards binary counter, not a downwards. This is handy in computers: when you have a simple counter, usually you want it to count up. This is bad in music, because it means initially it is all zeroes, and upon a clock pulse, the fastest division turns one. Another clock pulse, it turns zero and the next division turns one. You see the pattern? This thing produces downbeats! And i'm not saying downbeats aren't musical or fun, but when the entire set of divisions is downbeats, the rhythmic pattens that emerge are... confusing to say the least, and even irritating to my mind.
I guess they saved on inverting buffers and just rammed the CD4024 output pins right to the sockets, didn't they?
, i thought - since the 4024 is meant for computers, it is an upwards counter and would need an inverter after every output to produce pulses that a musician expects. But no! There is a transistor-based inverter for each of the six outputs, just wired as common-emitter (follower) instead of common-collector (inverter) for some weird reason! But then a second thought came in... i could inverse the buffers myself.
This page contains instructions on how to make your A-160 into a "musical" (a.k.a. functional) clock divider. This mod took me about 2 hours to figure out and implement, so it's not any kind of insane overkill stuff. With all my love to and fangirlism over Doepfer, i'm not sure how they made such a strange device. But hey, we've got soldering irons and bravery! So let's get to it.
Here lies an untouched A-160, unaware of what's about to happen to it. Read the manual while you're at it, wouldya? | |
Carefully undo the eight screws. 11mm hex bit works fine for me, but mine is metal, so i have to make sure NOT to scratch it against the panel and leave circular marks. Use a paper towel if clumsy. Put the panel and screws somehwere safe. | |
Now locate the two groups of transistors. These are BC547s. I'm a 2N3904/06 girl for NPN/PNP transistors, but it's the same stuff with an inverted pinout in this application. All of their collectors point to the edge of the board and are shorted to 12V, while the emitter points to the panel and has three resistors coming from it: one 10K load resistor to ground, one 2.2K current-limiter for the LED, and one 1K for output impedance. That's why there are six distinct groups of three resistors. Anyways... pop the six transistors out and service the holes so that they have no solder blobs over them. You can use wick or a pump. I do the good old "heat it up and blow real hard on it". It's dangerous, but it's free! | |
Now we take each pulldown load resistor and raise one of its legs - the one that goes to groundfill. Do not desolder the other leg. Stand all the resistors and bend all the free legs 90 degrees, so that they all align into an imaginary straight line. We're turning these pulldowns into pullups now. | |
Grab the desoldered transistors and solder them back, backwards! Yes, go against the silkscreen. We're swapping collector and emiiter places. Then, take some bare wire, and make the imaginary straight line from the previous step real. This one has to be connected to 12V - nicely exposed for us on the rightmost two pins of the A-161 expander connector. I don't have and am not planning on having one, so i felt free to solder directly to the pins. If you want to preserve the header, only connect the resistors with bare wire, and use some insulated wire to run it around and below the board to the same pins' solder pads. Bare wire can be found in Ethernet cables, by the way - feel free to salvage the ones that are out of order anyways. When done, push the resistor chain towards the jacks so that it leans 45 degrees from the board instead of a right angle; this will keep the build within the module panel's limits. | |
Whip out an exacto knife and cut exactly these two traces. This disconnects the collectors-to-be-emitters' solder pads row (and, unfortunately, the CD4024's VCC pin) from +12V. | |
Take some insulated wire and make a ~40mm bit with stripped and solder-coated ends. I used a torn-up IDE HDD cable. Solder one end to the resistor chain's end that is closer to the power pin header at the bottom of the module. There is a machining hole punched nicely adjacent to it, which we will use to bring the wire to the other side of the module. | |
Solder the other side of that wire to +12V. You can solder directly to the pin header, or to the capacitor leg that goes to the +12V trace like i did. This connects our pullup resistor ends to +12V, and since they connect to the expansion pin header's +12V on the other side, the rest of the circuit now receives +12V as before. On the same note, take some more wire one of the transistors' emitters to the now-unused pulldown resistor ground pin. This will effectively ground all the six transistors' emitters. |
|
The CD4024 chip's pin 14 (VCC, or "power +") is now shorted to ground, too! Cut it off from the trace as shown, then run a wire over to +12V next door. We're almost there, but the next part is finnicky - grab some tea if tired. | |
So... for some reason, the mad genius of Doepfer GmbH wired the chip's outputs directly to the transistors' bases. I didn't measure the heat of the chip on the unmodded device, but it seems to have worked fine?? I guess they're protected from becoming accidental current sinks. But now that they're pumping juice to transistors' bases and directly to ground from the emitters (~= shorted), it's gonna become a nice heatpad unless we take action. The six traces going from the IC to the transistor bases have to be cut, and 10K resistors installed inbetween each IC pin/T base pair instead. | |
Bend each resistor's leg 90 degrees firmly and cut it so that there's about 2mm of the leg left. There should not be any length of leg NOT perpendicular to the resistor's body, or it's going to short to the adjacent pin. Now, solder them into each pad. Make sure the resistor is pointing about 20 degrees up from the board while being installed. Cut the free legs so they are about 5-6mm each. | |
Prepare six lengths of insulated wire, about 40mm each. Strip and solder-coat each end, then clip to about 2mm on one side and about 5mm on the other. The long end goes on the resistor leg that's in the air. The short end is to be soldered into the IC pin pad. Double check which transistor base went to which IC pin, or stuff may get shuffled around! Some heatshrink on each resistor end is necessary to protect from shorting to the adjacent module. | |
And there we have it!~ Musically functional A-160, yours forever, warranty voided successfully. There probably is some shortcoming to this approach - now the output impedance is basically 11K instead of 1K, since +12 is coming from the pullup.. but as long as you don't drive, like, 5 clock inputs from a single output, you'll be fine. I had no issues clocking and gating any gateable Doepfer and Make Noise module in my system, but as with anything DIY in Eurorack, i cannot guarantee 100% compatibility with every single module. Bigger output impedance basically means modules with small input-to-ground impedance will combine with it and create a voltage divider, and divide our precious 12V to something closer to six-ish. Will it still work? Probably. Is there a module with which it won't? Certainly! But it's still better than having six outputs of.. err.. mathematical clock divisions. |