If you have seen the diyAudio schematic (currently #1 Google Images for
"discrete component flip flop"), you already noticed the notation at
every diode position: LED or 1N914. That notation has been there since
the original circuit was posted. Most people read it as a cosmetic option
-- LED if you want an indicator, signal diode if you don't.

That reading is wrong. The LED is the correct component. Not for
cosmetic reasons. For structural ones.

THE DIODE IS ALREADY THERE

In the DTL N-ary topology, the base path already requires a series diode.
This is not optional -- the series diode is what creates the immunity gap
between the active state and the inactive states. The contact potential of
the diode junction (Vf ~0.6V for 1N914, ~0.25V for BAT54 Schottky) is the
logic threshold. Remove the diode and the circuit loses its starvation
mechanism.

So the diode is structural. It is always there. It carries current every
time that position is active.

Now: an LED is a diode that emits light when current flows through it.
It has a forward voltage (Vf ~1.8-2.2V typical for visible red/green).
It has all the same electrical properties as a signal diode in this
circuit. It blocks in the reverse direction. It conducts in the forward
direction. It provides the series junction that creates the immunity gap.

When you substitute the LED for the 1N914, you get:

  Function 1 -- Series blocking diode: identical behavior, slightly
  higher Vf which WIDENS the immunity gap rather than narrowing it.
  The circuit works better, not worse.

  Function 2 -- State indicator: the LED lights when its transistor is
  active. One LED on at all times. Visual confirmation of which position
  holds the state. This was already in the original 1990s tube amplifier
  design -- one green LED indicating which channel was selected.

  Function 3 -- Optical bus emitter: the same current that flows through
  the LED as a series diode also drives photon emission. The LED fires
  the moment the transistor activates. No separate emitter needed. No
  separate trigger. The state change IS the optical emission.

Zero added components. Three simultaneous functions.

THE FLIP-FLOPTICOUPLER

When the LED output is received by a photoreceiver (photodiode or
phototransistor) at another cell's base input, the result is a
galvanically isolated logic coupling. The sender and receiver share
only light -- no electrical connection, no ground loop, no EMI coupling
between cells.

This is the Flip-Flopticoupler: a DTL cell whose LED emitter directly
drives the base of another cell through a photoreceiver. Multiple cells
can communicate without any copper connection between them.

Properties this enables:

  EMI immunity: light does not couple capacitively between traces.
  Hundreds of optical channels can run side-by-side with zero cross-talk.
  In high-voltage environments -- near motors, transformers, welding
  equipment -- where copper traces pick up interference, the optical
  path is completely immune.

  Galvanic isolation: no electrical path between sender and receiver.
  No ground loop. No voltage level mismatch problem between cells.

  3D stacking: optical vias through transparent substrate connect layers
  without drilling copper Through-Silicon Vias. Layer 1 can communicate
  with Layer 4 by shining light straight through.

  Phantom State optical: when no transistor is active, no LED emits.
  The optical bus is dark. Darkness IS the default state. Zero energy
  to maintain it. Zero energy to detect it.

THE SERIES DIODE SCALING LAW

The second property is independent of the optical extension and applies
to any version of the circuit -- LED or signal diode.

In the standard N-way circuit:

  Active collector:        0.2V (Vce saturation)
  Inactive node voltage:   0.2 + 0.6 = 0.8V (through one diode)
  Required to drive base:  Vf + Vbe = 1.2V...