Repository navigation
v3.13.0
The two NFC readers no longer talk over each other. Each one now has the air to
itself while it reads, which is the single biggest reliability gain this scale
has had: on the bench, the weaker of the two readers went from answering 60 % of
its attempts to 89 %, and the scale got faster at the same time.
Your scale's network name changes with this update. It is no longer
tigerscale-CE3F — see "Your scale's address changes" below. If you have the
old address bookmarked, you will need the new one.
Fixed
- The two readers were jamming each other. The scale asks one reader at a
time, but the one waiting its turn kept its radio switched on, and with the
two antennas facing each other 75 mm apart that is exactly what the reading
one had to hear your tag through. The idle reader is now silent while its
neighbour works. Measured on the bench over the useful power range: the weaker
reader answered 60 % of its attempts before and 89 % after, with more reads per
second rather than fewer, because attempts stop being wasted. - "Silence the readers when the platform is empty" silenced nothing. The two
functions that switch the radios off never sent the command — they set a flag
and returned. Both radios stayed on permanently. They now really switch off. - The readers were given too little time to answer. A tag that replied
slightly late was reported as "no tag". The window is now 100 ms with two
attempts instead of 30 ms with one, which also raised throughput. - The default reader power is now level 7 instead of 3. With the readers
taking turns, 7 reads every spool on both bench series while 3 missed about one
in six on the weaker side. This only affects a scale that has never had a power
level set; yours keeps whatever it is on, and Settings > RFID > Auto still
measures the best value for your own unit. - Every scale had the same network name. The name was supposed to be built
from the scale's own hardware address, but it was read a fraction too early in
the startup and came out as the same four characters on every unit ever
shipped. Two scales on the same network therefore fought over one name, and
which one won was decided afresh at each restart — so a bookmarked address
could quietly open the other scale. Each unit now takes its name from its own
hardware address.
Your scale's address changes
The first time your scale takes this update, it stops answering to
tigerscale-CE3F.local and starts answering to its own name, built from its
hardware address — for example tigerscale-FF94.local. There is deliberately no
redirect from the old name, so that one address means one scale from now on.
Three ways to find the new one:
- Work it out: the name is
tigerscale-followed by the last two bytes of
the scale's MAC address, in uppercase. The scale shows its MAC during WiFi
setup, and your router's device list shows it too. - Ask the scale:
http://<its-ip>/api/statusreturns amdnsfield with
the full name. - Skip the name: open the scale by IP address. It shows its IP when it joins
WiFi, and your router lists it.
If you have two scales, expect them to disagree until both are updated: an
updated scale uses the new name, one still on 3.12.0 or earlier keeps the old.
Known issues
- On one bench unit, the right-hand reader reads 7-byte tags (NTAG / TigerTag)
normally but cannot read 4-byte MIFARE Classic cards. The fault follows the
reader, not the card, and a second unit does not show it. This looks like a
hardware fault on that module rather than a firmware problem, and it is not
believed to affect TigerTag spools, which use 7-byte tags.
Updating
Over the air from Settings > Update, or over USB with
bash scripts/flash.sh --fs. Both keep your WiFi credentials, TigerTag session
and calibration. Only the scale's network name changes.