Skip to content

Show the active deck on the LC6000 jog display (upload a per-deck image over the WD port) #165

Open

Description

@Skeyelab

Goal

The LC6000's jog display shows which Syndicate deck the controller is on right now. That means the deck number on the deck's color, and later the track's artwork or title. Today the display shows the unit's idle "pie", or whatever djay left there.

The protocol was found and verified on hardware in #121: our own PNG shows on the jog.

Behavior

  • On connect, the bridge says hello as a host on the Controller port, then uploads the active layer's image to the WD port and shows it.
  • When the active layer changes (note 51, LC6000 deck layers: MIDI channel N drives Syndicate deck N #79), or the deck color changes (Deck colors are a user setting (app badges and jog ring) #123), the bridge uploads that deck's image and shows it.
  • On quit, the bridge sends the Serato goodbye, so the unit returns to standalone.
  • Settings gets an on/off switch, "Show the deck on the jog display". Default to be decided; see the open questions.
  • If the unit refuses (an ack of 06, or no ack), the bridge logs it and doesn't retry in a loop. The LEDs and controls are unaffected.

The protocol (from #121; see its recipe comment)

  1. Hello, on the Controller port: F0 00 02 0B 7F 10 60 00 04 04 01 01 00 F7, then F0 00 20 7F 01 F7.
  2. Upload, on WD: 90 01 00, then 0A 01 00 06 0D, then 0C (size), then 0D chunks (21-bit offset plus an MSB-first 7-bit bitstream, 14332 bytes per chunk), then 0E (slot 01 plus the PNG's byte-sum as 8 nibbles). Each step is acknowledged with 01 10 0F 00 01 00.
  3. Commit, on WD: 7E 00 06 01, then 10. The unit answers 01 10 10.
  4. Show, on WD: 0A 01 00 01 0D, which displays image index 01.
  5. Goodbye, on the Controller port at quit: F0 00 20 7F 02 F7.
  • Image: a 198×198 RGBA PNG.
  • Upload time: about 5 s per 16 KB chunk on the wire. A 17–30 KB image takes about 7–11 s. While an upload is in flight, a newer request replaces the queued one, so only the latest deck's image is sent.

Plan: test-first, slice by slice

  1. Kit, pure:
    • JogImageUpload: from PNG bytes to the exact message list. Tested against djay's captured upload, byte for byte; the encoder already reproduces it. The test needs a small checked-in synthetic PNG, not djay's.
    • The checksum rule.
    • The ack parser: 00, 06, commit.
  2. Kit, Coordinator: a JogDisplay port (fake in tests) that receives "show deck N in color C". Cases to cover:
    • It fires on connect, on a layer handover and on a deck-color change.
    • It is debounced while an upload is in flight.
    • A refusal is logged, with no retry storm.
    • The goodbye is sent on stop.
  3. App: render the deck badge as a 198×198 PNG (CoreGraphics), plus a CoreMIDI SysEx sender with completion and ack listening on the WD source.
  4. Settings switch, with the help text.

Bisect first (hardware, each send OK'd by the CoS)

The working run replayed djay's whole first second. Before building, find the minimal handshake by dropping one part at a time:

  1. The CTL LED paint and identity requests: probably not needed, since they went out after the upload.
  2. Denon cmd 60, keeping only the Serato hello.
  3. 90 01 00, and the pre-upload 0A.
  4. Whether the goodbye clears the stored image or keeps it, and what a power cycle does.
  5. Whether the first byte of 0E, slot 01, can be 02, so two images can be stored for a fast deck switch.

Safety

Open questions

  • Default on or off? On risks surprising a DJ who also runs djay. Off hides the feature.
  • Artwork and track title: does Syndicate expose them over OSC? Separate issue.

Activity

  1. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Rendering note from #121. Orientation (user observation): our icon's notch, drawn pointing up-right, shows pointing down-left on the jog, and the image doesn't turn with the platter. So the display shows uploads transformed:

    • A single horizontal or vertical mirror is ruled out; those would put the notch up-left or down-right.
    • A 180° rotation (the same as flipping both axes) fits.
    • So does a mirror across the anti-diagonal.

    Rule for now: pre-rotate the PNG 180° before upload.

    Pending confirm (needs an OK to send): an upright F with a red square in the top-right corner, 198×198, 2.1 KB, which uploads in about 1 s. The plan is ready.

    • Under a 180° rotation it shows as an upside-down F with the square bottom-left.
    • Under the anti-diagonal mirror, the F lies on its side.
  2. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    User decisions (2026-11-16):

    • On by default.
    • The long-term goal is real-time feedback from the app on the jog display, for example track title, BPM, position or progress, and deck state, not only a static per-deck image.

    What that implies for the design:

    • Uploads take about 1 s (2 KB) to 11 s (30 KB) over MIDI, so real-time can't mean re-sending the full image each frame. Options to investigate:
      1. Small, fast PNGs: palette-indexed and heavily compressed, a few KB, updated at ~1 Hz for slow-changing info (title, BPM).
      2. The display's own elements: on the Prime 4, 0A (visible elements), 0B (element color) and 7C (brightness) drive built-in overlays without an upload. Test whether the LC6000 has a position ring or text elements we can drive cheaply.
      3. Several slots: pre-upload a few images (slots 01, 02…) and switch between them with 0A's index, so switching is instant.
    • So v1.1 ships the static per-deck image (deck number and color). Real-time comes as follow-up tickets once the cheap update paths are measured.
  3. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Bisect and probe plan (hardware; every send OK'd by the CoS)

    Measured so far: upload speed is fixed by the wire. Every run moved about 3.1 KB/s of SysEx: 16,393 B in ~5.3 s, 2,458 B in ~0.78 s, and djay's 29.7 KB PNG in ~11 s. Wire bytes are the PNG size × 8/7, plus about 12 B of framing per chunk.

    • A 2–4 KB palette PNG takes ~0.8–1.5 s: fine for a deck or layer change.
    • Real-time feedback can't be image uploads. Position, BPM and state changes have to use small messages that are already on the unit. An image upload is only for what changes rarely: the deck, the artwork, the title.

    Steps

    1. (a) Upload time against PNG size. Measure deck-badge PNGs: a palette PNG (color type 3) at 1–2 KB, gray, and RGBA, from send to commit ack.
      • Also check that the unit accepts color type 3 and type 0/2 at all, since djay only ever sent RGBA.
    2. (b) A second slot. Upload into slot 02 (0E 02 …), then show it with 0A 01 00 02 0D and switch back with index 01.
      • If both slots stick, deck changes become instant: pre-upload each deck's badge, then send a one-message 0A per layer change.
      • Also find how many slots exist, and whether they survive the goodbye and a power cycle.
    3. (c) Element messages (Prime 4 style). Probe on the LC6000 class 10: the 0B element color, the 7C backlight, and 0A text codes (1/64…64, A–D, "--").
      • That's cheap text and color over the stored image, the base for real-time state such as loop size or a beat-jump readout.
    4. (d) Ring and position. The Prime 4 drives the platter position and slip indicators with pitch bend (E0/E1, 14-bit). Probe on the WD port for a live track-position indicator, then compare with what the LED ring on the Controller port already shows.
    5. Minimal handshake (from the issue body): drop the CTL paint and identity requests, cmd 60, 90 01 00, and the pre-upload 0A, one at a time.
    6. Orientation: the F test result. So far a 180° pre-rotation fits.

    Defaults (user's decision): the feature is on by default. The long-term goal is real-time feedback on the jog, so the design keeps a fast path (element messages and pre-stored slots) separate from the slow path (image uploads).

  4. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Orientation confirmed by the F test: the user saw the F upside down, with the red square in the lower left. That is a 180° rotation, not a mirror.

    Rule: rotate the image 180° before encoding the PNG. The checksum in 0E is then computed over the rotated PNG's bytes.

  5. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    What Denon's own software shows on the LC6000 platter display (from the Denon manual, as quoted in search results):

    • playhead position (a moving marker or needle)
    • auto loop size (a number)
    • album artwork (an uploaded image)

    A Native Instruments forum post adds that the display can show artwork, track position, jog position and numbers 00–99 without modification.

    So the display likely has built-in elements for position and a two-digit number, which are exactly the cheap real-time paths to probe after the image work: 0A visible-element mask, text and number codes, and a position value (maybe pitch bend on WD, as on the Prime 4). Candidates to drive from Syndicate feedback: Progress → playhead marker; loop size or deck number → the 00–99 number; Loaded/TrackInfo → an artwork or deck image upload.

  6. added a commit that references this issue on Nov 16, 2026
  7. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    User check (2026-11-16): djay Pro does not draw a needle or position marker on the LC6000 display while a track plays, so djay can't teach us the position or number elements. Ways forward: (1) probe directly with Prime 4–style messages (0A element masks and the number/text codes, 0B colors, pitch bend on WD), with the user watching, one change at a time; (2) optionally capture VirtualDJ (free for home use, supports Denon displays) or Serato, if they draw position on the LC6000.

  8. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Position marker found (user capture, djay playing, 2026-11-16). djay draws the platter marker with Pitch Bend on the WD port, ch1:

    • about 50 Hz (every ~20 ms) while playing, a signed 14-bit value from −8192 to 8191 that wraps as it climbs (…7847, 8036, −8179, −7990…)
    • the step is about +190 every 20 ms, so one full wrap (16384) takes about 1.8 s, which is 33⅓ rpm. The value is the virtual record's rotation angle, not track progress.
    • while the platter is scratched or nudged, the values follow the jog (both directions)

    djay also mirrors the play/cue state to WD as note 1 on ch1 (vel 127/0, toggling about every 0.5 s while cued: the blink), at the same time as the Controller LEDs.

    So the cheap real-time marker is: send a pitch bend on WD ch1 with angle = (playback position in seconds × 33⅓/60 revolutions) mod 1 × 16384 − 8192, at ~30–50 Hz while playing. Syndicate's Progress feedback times the track length gives the position, interpolated between feedback messages with the tempo. No image upload needed.

  9. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    First real-time slice: the platter marker (pitch bend on WD)

    Rewritten after the second capture (comment below): the marker is a stateful angle, not a function of the track position.

    From the captures, djay keeps a virtual record angle and sends it as pitch bend on WD ch1:

    • While playing, the angle climbs at 33⅓ rpm with playback time, and is sent at about 50 Hz.
    • While the platter is touched or turned (paused or playing), the angle follows the hand both ways. It's sent only while it moves.
    • Play resumes the climb from the current angle. A seek doesn't reset the angle.
    • Note 1 on WD ch1 is the play/cue state: steady 127 while playing, blinking 127/0 every 0.5 s while paused or cued.

    Slice 1, Kit, pure, test-first

    PlatterAngle, a value type holding the angle in turns (0 ≤ a < 1):

    • advance(seconds:rate:): a += seconds × (5/9) × rate, where 33⅓ rpm is 5/9 turn per second and rate is the tempo ratio, 1.0 at 0%.
    • turn(jogCounts:): a += counts / 3600, using the jog's 14-bit position change and the profile's counts_per_rev (3600).
    • pitchBend: ⌊a × 16384⌋, on the wire 0…16383. 8192 is center, so the values read signed −8192…+8191, which gives the capture's …8036, −8179… wrap.

    Tests, with literal values:

    From Step Result
    0 advance 0.45 s 4096
    0 advance 0.9 s 8192
    0 advance 1.8 s 0 (one full turn wraps)
    0 advance 0.02 s 182 (the capture's step was ≈ +190)
    0 advance 0.45 s at rate 2.0 8192
    0 turn(jogCounts: 900) 4096
    0 turn(jogCounts: -900) 12288
    4096 seek 4096 (nothing moves)

    Slice 2, Coordinator, with a fake WD port

    Slice 3, App

    The WD port opens next to the Controller port, under the same "show the deck on the jog display" switch (on by default).

    Hardware probe first (needs an OK)

    1. A slow pitch-bend sweep on WD ch1, with no host hello: 0 → 16383 in steps of 512 at about 4 Hz. Does the marker draw without the Serato handshake? If not, the hello comes first, as for images.
    2. Note 1 on WD: does it change the display, for example a play or cue state?

    jogseq allows both messages on WD now (PR #167).

  10. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Second paste (djay: paused, hand on the platter, then play): this confirms and adds detail.

    • While paused, WD pitch bend is only sent while the platter moves. It follows the hand both ways (e.g. 1995 → −8105 → 7934 … back), so the marker tracks a scratch on a paused deck. With the platter still, nothing is sent and the marker holds.
    • Play resumes the steady climb of about +189 every 20 ms (33⅓ rpm) from wherever the angle was; there's no jump back.
    • WD note 1 (ch1): toggles 127/0 every 0.5 s while paused or cued (the blink), stays at 127 while playing, plus a quick off/on burst on play/stop. That's djay mirroring the Play/Cue LED state onto the display, probably a play-state indicator drawn on the screen.
    • A seek doesn't reset the angle; it keeps advancing from the current value.

    So the bridge's marker model is: angle += Δplayback-time × 33⅓/60 × 16384 (wrap to ±8192), driven by our position/tempo while playing and by the jog position deltas (CC 17/49) while the platter is touched. Send ~50 Hz only when the angle changes.

  11. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Track-load capture (djay, 2026-11-16): cover art loads FAST.

    • On load djay sends two uploads back to back, both to slot 02:
      1. 0C size 02 15 6E = 35,566 B → 0D×3 (16393, 16393, 7902) → 0E 02 | 00 00 04 05 06 00 04 0A → 0A 0D 00 02 0D (show index 02)
      2. 0C size 02 2A 47 = 38,215 B → 0D×3 → 0E 02 | … → 0A 0D 00 02 0D
        (Note: 0C's 3 bytes are the 21-bit size; the leading byte isn't a slot. The slot is the first byte of 0E, and 0A's third byte picks which slot to show.)
    • Timing: djay hands all ~74 KB to CoreMIDI within ~40 ms, and the unit's ten 0F acks arrive over the next ~1.1 s. So the real throughput is roughly 60–70 KB/s, about 20× our measured 3.1 KB/s. Our jogseq paces chunk by chunk, waiting for acks. The fix: queue all chunks at once and collect the acks afterwards.
    • Slot 01 = the djay logo / idle image; slot 02 = the track artwork, swapped with 0A … index 02.
    • The marker resets to −8192 (angle 0) on load, then climbs while playing.

    Impact: per-track artwork and fast per-deck images are realistic (~0.5–1 s). A 2-slot design: slot 01 for the deck badge or idle image, slot 02 for the current track's cover (from Syndicate TrackInfo, if we can get artwork) or a rendered title card.

  12. added a commit that references this issue on Nov 16, 2026
  13. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Upload speed, corrected (from the djay track-load capture on #165): the ~3.1 KB/s I measured came from my sender's API, not the unit. MIDISendSysex paces SysEx at MIDI 1.0 speed (31,250 baud). djay hands about 74 KB to CoreMIDI in about 40 ms, and the unit acks it within about 1.1 s: roughly 60–70 KB/s. So a cover-art upload takes about 0.5 s, not 10 s.

    Slots: 0C's three bytes are only the 21-bit size. The slot is 0E's first byte and 0A's image index: the logo is in slot 01, and cover art goes into slot 02, shown with 0A 0D 00 02 0D. The tools now support both (jogseq burst, jogbuild plan … [slot]).

    This revises the plan's "uploads are only for what changes rarely": at ~65 KB/s, per-track artwork or a title image is practical, about 0.5 s per load. The marker stays pitch bend.

  14. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Confirmed on hardware: a burst upload of our deck-ring icon (rotated 180° by jogbuild) to slot 02, shown with 0A 01 00 02 0D. The user sees our icon on the jog.

    • Host hello first. The unit answered 9F 37 7F, 90 33 7F and 8F 37 00.
    • Acks: 00 ×4, at +0.097, +0.104, +0.106 and +0.561 s. Then the identity reply, and the commit acked 01 10 10 at +0.585 s.
    • Timing: 0.55 s from the first 0D to the last ack, for 20.3 KB on the wire. That's about 37 KB/s including the unit's processing; the last ack, about 0.45 s after the third, looks like the PNG decode. The paced send took about 7 s for the same image.
    • Sender: jogseq burst (UMP SysEx7 via MIDISendEventList), from PR tools/lc6000-display: burst sends at USB speed, upload slot in plans (#165) #168.

    #165 is unblocked for implementation planning. The protocol is confirmed end to end: the hello, upload at USB speed, the checksum, commit, show by slot, and the 180° rotation. Still open from the probe list, but none of it blocks the design: slot persistence across the goodbye and a power cycle, and the minimal handshake.

  15. Skeyelab commented on Nov 16, 2026

    @Skeyelab
    MemberAuthor

    Pre-rotation confirmed: with the 180° pre-rotation from jogbuild, the user sees our icon upright, with its notch up-right as drawn. The slot 02 switch (0A 01 00 02 0D) showed it.

    Next session's probes (each needs an OK):

    1. Preload two slots. Put one image in slot 01 and another in 02, then swap with a single 0A index message: instant images on a layer change.
    2. Sweep the pitch-bend marker on WD ch1, without the hello.
  16. added a commit that references this issue on Nov 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions