Why the Patch Always Breaks at Midnight

Start With the Obvious Before You Start With the Interesting

Something changed, and the patch you had working two hours ago is now a mess. Before reaching for an oscilloscope, work through the candidates in order of how often they actually cause the problem.

Signal loading is the first suspect. Every module has an output impedance and every downstream input has an input impedance; when you split one output across three or four inputs, you are forming a voltage divider, and the driven signal drops. A VCO pitch output that looked fine driving one oscillator will sag measurably when you parallel it to three VCOs without a buffered multiple in the chain. The symptom is usually pitch that is flat across the whole keyboard range rather than drifting at one end — 1V/oct calibration is doing the right thing, but it is working with a reduced voltage swing. Add a buffered multiple and recheck.

Ground loops are second. They appear when your case ground and some external piece of equipment — an effects pedal, an audio interface running on its own wall-wart — share a ground path through your audio cables that differs in potential from the shared chassis ground. The result is usually a fifty- or sixty-hertz hum sitting underneath the patch, most audible in quiet passages. Unplug external gear one piece at a time. If the hum disappears when you remove a specific cable, you have found the loop. The fix is a galvanic isolation transformer on the audio output, not a frantic rerouting of your patch.

The Slower Failures

Once you have ruled out loading and ground issues, consider thermal drift. Analog VCOs and filters built around matched transistor pairs or tempco resistors take time to reach thermal equilibrium. If your patch has been running for only ten minutes and you have good tracking through the first three octaves but the fourth and fifth are sliding sharp, the oscillator has not finished warming up. This is especially pronounced in dense cases where the power supply and neighbouring modules are raising the ambient temperature slowly across a session. Letting the case run for thirty minutes before you commit to tuning is not superstition; it is thermodynamics.

Cable resistance matters less than the other factors in most patches, but it becomes relevant in long modulation chains where a small voltage offset accumulates. Cheap patch cables with thin core wire and oxidised jack connectors can introduce a few ohms of resistance in series — trivial for audio, but a meaningful error when you are passing a precision CV that determines pitch. If swapping a cable for a fresh one of known quality changes the tuning, the old cable is failing. Wiggle the jack at both ends; intermittent contact is more common than uniform resistance, and it produces the maddening symptom of a patch that works for twenty seconds and then goes wrong.

Power supply sag is easy to overlook. As a session lengthens and modules warm up, their current draw can creep above what they draw when cold, particularly op-amp-heavy designs. If your power supply is already near its rated capacity — a common situation in a half-populated case being pushed toward full — you may get rail sag that upsets reference voltages throughout the rack. The symptom is a general instability that is hard to localise because it is hitting every module simultaneously. Measure your twelve-volt and five-volt rails with a multimeter at the bus board. If you see more than a few percent below nominal, the supply is the problem, not the patch.

The Discipline

Work the list in order. Loading is common; ground loops are common; drift is common; bad cables are common; supply sag is less common but catastrophic when it occurs. Exotic explanations — parasitic capacitance between module front panels, RF interference from a phone — do exist but account for perhaps five percent of midnight failures. Solve the boring problems first. The patch will still be there when you have done so.