Feature: Statistics & Consumption Counters¶

Concept¶

Happy Hare tracks two independent kinds of numbers:

Swap and gate statistics - detailed timing for every phase of every toolchange (both lifetime totals and the current job), plus a per-gate quality assessment - all built in, always recorded, no setup required.
Consumption counters - a simple user-defined counting/warning framework for anything you want to track and be reminded about, like a filament cutter blade that dulls after a few thousand cuts.

Both live behind a single command, MMU_STATS.

Commands¶

Swap statistics¶

MMU_STATS                # Summary: swap timing table, pause time, toolchange count
MMU_STATS DETAIL=1       # Add per-gate load/unload timing and slippage/failure detail
MMU_STATS TOTAL=1        # Force the lifetime-totals table even mid-print
MMU_STATS RESET=1        # Reset swap and gate statistics back to zero

Full parameter reference: MMU_STATS. A plain MMU_STATS looks something like this (columns/rows shown are the shipped defaults - see Tuning below for customizing them):

+-----------+----------+-------------------+----------+
|  114(46)  |unloading |      loading       | complete |
|   swaps   |    -     |   -    |   post    |   swap   |
+-----------+----------+--------+-----------+----------+
|     total |    47:19 |  37:11 |     33:39 |  2:00:38 |
|       avg |     0:24 |   0:19 |      0:17 |     1:03 |
|  this job |    10:27 |   8:29 |      8:30 |    28:02 |
|       avg |     0:13 |   0:11 |      0:11 |     0:36 |
|      last |     0:12 |   0:10 |      0:14 |     0:39 |
+-----------+----------+--------+-----------+----------+

11:43:27 spent paused over 10 pauses (All time)
8:15:38 spent paused over 3 pauses (This job)
114 / 220 toolchanges
Number of swaps since last incident: 105 (Record: 1111)

The header's 114(46) is total swaps, with this job's count in brackets. The toolchange line shows <done> / <slicer total> once a slicer tool map with a known total is loaded, or just a plain count otherwise. Columns are grouped in two tiers: an unloading/loading/complete phase across the top, and the specific timing (- for the main move, post for what happens after it) underneath - useful for spotting which step of a swap is taking longer than expected. Exact spacing shifts with your data and console_stat_columns selection; treat the layout above as illustrative rather than a literal byte-for-byte transcript.

MMU_STATS RESET=1 clears swap totals and gate statistics back to zero - it does not touch consumption counters; reset those individually (see below).

Gate statistics¶

Every gate also gets a quality assessment, drawn from load/unload timing and (if an encoder is fitted) slippage tracking:

Gate Statistics:
0:😎, 1:😃, 2:😊, 3:😐, 4:😟, 5:😢, 6:😱, 7:-, 8:-

(- means the gate has no recorded activity yet.) DETAIL=1 adds the raw numbers behind that assessment, per gate:

Gate 0: Load: 1234.5mm (slippage: 0.3%); Unload: 1180.2mm (slippage: 0.4%); Failures: (load: 0 unload: 0 pauses: 0); Quality: 99.7%

The slippage/quality figures only appear for a gate with an encoder fitted; load/unload failure and pause counts are tracked for every gate regardless.

Note

Don't chase a perfect score on every gate - a summary trends back towards good on its own once whatever caused a rough patch is fixed, no reset needed. Treat it as a comparison between gates rather than an absolute target; a gate that's noticeably worse than its neighbours is the one worth investigating (calibration, friction, or a move speed that's too aggressive for that particular gate).

Consumption counters¶

A counter tracks any consumable you want reminders about. Worked example - a filament cutter blade rated for about 4000 cuts:

MMU_STATS COUNTER=cutter_blade LIMIT=4000 WARNING="You may need to replace your cutting blade"

Then, wherever the consumable actually gets used (typically a macro call):

MMU_STATS COUNTER=cutter_blade INCR=1

Important

Always run the LIMIT=/WARNING= setup command once before the first INCR=1 - creating a counter this way starts it at 0 and does not apply that same call's increment. It's harmless to re-run the setup command later (it doesn't reset the count), so a simple macro can just always call it before incrementing.

Crossing the limit logs a warning (and, if the counter was set up with PAUSE=1, pauses the print):

Warning: You may need to replace your cutting blade
Count cutter_blade (4001) above limit 4000
Use 'MMU_STATS COUNTER=cutter_blade RESET=1' to reset

Check current counts, reset, or delete a counter you no longer need:

MMU_STATS SHOWCOUNTS=1                       # List every counter
MMU_STATS COUNTER=cutter_blade RESET=1       # Reset just this one counter to 0
MMU_STATS COUNTER=cutter_blade LIMIT=-1      # Temporarily disable its limit check
MMU_STATS COUNTER=cutter_blade DELETE=1      # Remove it entirely

MMU_STATS SHOWCOUNTS=1

Consumption counters:
Count cutter_blade: 568 (limit 4000)

Counters persist across restarts until explicitly deleted.

Note

Happy Hare may add more built-in preset counters (varying by MMU type and which options you have enabled) in future versions - this page only describes the fully user-defined counter mechanism available today. Two of the pieces for a couple of obvious ones already exist as plain config values - mmu_macro_vars.cfg's _MMU_STATE_VARS section ships a servo_down_limit and a cutter_blade_limit (see Macro Variables: State change hooks) - but neither is wired to an actual counter automatically; you'd still set one up yourself with MMU_STATS COUNTER=... as above, just reusing Happy Hare's own suggested limit value.

Where statistics and counters are stored¶

Both swap/gate statistics and consumption counters are persisted the same way as everything else Happy Hare needs to remember between restarts: in its Klipper [save_variables] file, typically mmu_vars.cfg.

Warning

mmu_vars.cfg can be hand-edited directly, but be careful - corrupting it can leave Happy Hare unable to start.

Tuning¶

The swap-statistics table's columns and rows, and how gate quality is displayed, are configured in mmu.cfg's [mmu_parameters] section (not the file literally named mmu_parameters.cfg):

log_statistics             : 1        # 1 = log the table on every toolchange (default), 0 = still recorded, just not printed
console_stat_columns       : unload, load, post_load, total   # Any of: pre_unload, form_tip, unload, post_unload, pre_load, load, purge, post_load, total
console_stat_rows          : total, total_average, job, job_average, last
console_always_output_full : 1        # 1 = always show the full table, 0 = only outside a print
console_gate_stat          : emoticon # string | percentage | emoticon
console_show_colored_text  : 1        # 1 = color console output where supported, 0 = plain monochrome text
console_show_filament_color: 1        # 1 = show a colored "swatch" for filament, 0 = a plain asterisk instead

Trim console_stat_columns/console_stat_rows to shrink the table - useful if you're displaying it somewhere space-constrained, like a KlipperScreen popup. console_gate_stat swaps the gate-statistics line between the emoji version above, a plain string (poor/good/perfect/...), or a percentage. console_show_colored_text/console_show_filament_color sit in the same config section but aren't specific to this page - they affect color and swatch use across Happy Hare's console output generally, including gate-map listings elsewhere on this site.

Troubleshooting¶

MMU_STATS RESET=1 didn't clear a consumption counter - by design; it only resets swap/gate statistics. Reset a specific counter with MMU_STATS COUNTER=<name> RESET=1.
A brand new counter's first INCR=1 didn't seem to count - the setup command (LIMIT=/WARNING=) must run before the first increment; see the warning above.
Gate statistics look wrong for gates with no encoder - slippage and quality are encoder-only; a gate without one only tracks load/unload failures and pauses, which is expected, not a fault.