Skip to content

Latest commit

 

History

History
229 lines (168 loc) · 15.7 KB

machine-notes.md

File metadata and controls

229 lines (168 loc) · 15.7 KB

The machine: how a GLM installation is wired

Not the command set – the configuration around it. Channels and outgoing lines, what reaches the ERP and by which path, the memory card, unique data, and the per-package flags that say what went wrong.

Citation rules are in gxnet-notes.md, How to read these notes. Originals: GxNet-de.md for the subfunctions, the GLM product manual for the machine side.


1. Where the device may send, and what it sends there

GWBF GGW_SENDKANAL_A_ENABLE, and then GWC0–GWC3 for channels B–E; GW74/GW75 (from 13.40) for J and K. The wording is "Bereitschaft des GX über Kanal x initiativ zu senden" – readiness of the device to send on its own initiative over that channel – with 0 spelled out as "packungssynchrone Auszeichnungsdaten werden nicht versendet".

A per-channel gate on what goes out unasked. It does not decide where dialog answers and key presses go: see section 3.

  • A channel belongs to the device. One may be the system bus to a slave labeller, another the memory card. Writing a switch to 0 can silence a second printer.
  • The A-> prefix in the server log is the server's label for the connection, not channel A.
  • GGV_CHANNEL (GV0E, 14.0) and GGV_CHANNEL_MULTI_LABELER (GV0F) map channels to system-bus addresses. GGL_SBUS_ADR (GL38) is the device's own address, 0 meaning the bus is off. GGW_SBUS_KONFIG (GW38): 0 Profibus, 1 Ethernet, 2 both.
  • The switch exists for A–E and J, K only. The memory-card channel is set through the send editor.

And what goes out on a channel is a third, separate list

SRV_NET_KONF_CONTENT_SEND_BY_CHANGE (SV5A), ..._ADDON_PSV_PCK (SV5B) and ..._CONTENT_PSV_DATA (SV5C) read it. Each answers with SRV_UFKENN_CHANNEL_INFO blocks, and each block is a pair: LGW_UFKENN (LW02) and SRL_NET_CHANNEL_BITMAP (SL8C) – bit 0 internal, bit 1 = A ... bit 11 = K.

LGW_UFKENN is the token's 16-bit code: class in the high byte, index in the low, with class = (group << 4) | type. So GT63 is 0x0763 = 1891 and WZV_SDD_RESULT is 0x9663 = 38499. Group G is 0, not 7.

In text these values are signed words, so WZV_SDD_RESULT shows as -27037, not 38499.

The lists carry data, not events. They hold groups A, G and P. Dialog answers and key presses are not configured here.

SV5E is that nested block, not a command of its own. Asking for it directly gets fremdes Kommando. The readable ones are SV5A, SV5B, SV5C, SV5D, SV5F, each containing SV5E blocks.

Its writing counterparts are SV6A–SV6F, and those have error codes of their own – 30149–30153, "falscher Kanal bzw. falsche Kennung" – which lock labelling. Read freely, write only deliberately.

2. The outgoing-lines configuration is readable and writable over the wire

The <Add.data to PSV_PCK> setting – the one that decides whether a value written on the master reaches a secondary printer – does not need a trip to the terminal and an authorisation level. It is a telegram, from 14.00:

Read Write What
SV5A SV6A "send on change"
SV5B SV6B "Add.data to PSV_PCK"
SV5C SV6C "content of PSV_DATA"
SV5D SV6D "add.data to statistics"
SV5F SV6F contents of statistics reports 1–10

The write replaces the list, it does not add to it – and getting the channel or the identifier wrong does not merely fail. There are dedicated locking errors for it: 30150 PAE_LOCK_NET_KONF_ADDON_PSV_PCK, and 30149 / 30151 / 30152 / 30153 for the neighbouring commands. A locked device is a stopped line. Read first, add one entry, write the whole list back, and not on a running line.

Related settings that are ordinary reads:

  • WZL_WDOUTFLG_A ... _K (WL25–WL29, WL99, WL9A, WLA9–WLAC): which label kinds (single, totals) go out on each channel as PSV_PCK.
  • GGW_PACKUNGEN_PSV_DATA (GWFC): 0 = only PSV_PCK for valid packages, 1 = PSV_DATA for valid and PSV_FEHLDATA for invalid ones. If the ERP only parses PSV_DATA, defective packages are arriving somewhere it is not looking.
  • GGW_TEL_QUIT_MODE (GW58, 11.00): 0 = single acknowledgement, 1 = double. On "double" one command produces two acknowledgements, which a receive loop that stops after the first will leave in the queue for the next request to trip over.

3. Getting the operator's answer back

A dialog answer (WV63) and a key press (WV05) arrive by themselves. They cannot be requested: A?WV63 is refused with LGW_RETURN 2 and A?WV05 with 4.

They go where the "Level 9" row of the connection parameters points, a row separate from channels A–K:

<Configuration> / <Communication configuration> / <Outgoing lines> / <Connection parameters>
Field Value
Device ID 1
Bus type TCP/IP
Bus address, connection process 0
IP the server running _connect.BRAIN
Port the local port set for this device in _connectConfig

Restart the machine afterwards. Once connected, the machine fills in the row's socket column, so that shows whether it worked.

If the row is AEDV (serial port) or empty, the dialog opens, is acknowledged, and closes when pressed, and not one frame reaches the network. Nothing reports an error. A server log showing no inbound frame after the press points here. Keys program and appear either way.

Then the client side: read the right queue. See bcs-notes.md, Who gets a spontaneous telegram. If a BRAIN2 queue claims WV63, disable that module for the device (Oee).

Ports. In the line's connections, 1045 is the device's port (the server connects to it), and the port in the row above is the server's (the device connects to it). Both directions are in use, each side acknowledging with BIZERBA_OK.

The memory card is not event-driven. Package data reaches BRAIN2 by polling: MDW_GET_BUFF (MW06) out, MDV_SEQUENZ_END (MV08) back, every five seconds.

4. Per-package error flags: PSL_PCK_ERR_FLAGS (PL13)

A 32-bit field in every package record, so it arrives with the data the ERP is already collecting – no extra polling, and in particular no need to go anywhere near the memory-card buffer.

Bit Meaning
0 internal error
1 label missing or faulty (printer or applicator)
2 package could not be weighed
3 weight too heavy or too light
4 package contains metal
5 a slave labeller acknowledged negatively
6 package data could not be sent
7 a user-defined formula returned an error
8 data changed at an impermissible moment
9 switched to transport mode
10 ejection triggered by the user
11 package too long or too short
12 separation error
13 code read-back returned an error
14 statistics report running
15 faulty but not cancelled, because the total was already sent
16 marked faulty by an external signal on the I/O unit
17 scanner could not supply data fast enough
18 RFID write error
19 TTI error
20 logo missing or faulty
21 could not be weighed during teach
23 empty package
24 no unique data available
25 test package in scale-check mode

Bit 24 is the "the marking codes have run out" signal, and it is per package. Bit 13 is the verification that the printed code scanned back correctly.

Cumulative counters for the same causes exist as ordinary reads: SRL_PCKE_* (SL40–SL5D, 14.20) – among them SRL_PCKE_CONSISTENCY_COUNT (SL48), which counts packages spoiled by an unfinished sequence or a PLU change. That is a direct measurement of how badly changeovers go.

5. Device status without touching anything

PSL_RECEIVE_GXSTATUS (PL03) is a plain read and answers with a number from a long list. The ones worth knowing:

Value Meaning
0 ok / XON
1 XOFF, not ready to receive
2 label roll empty
5 no label on the carrier
7 memory card full
8 memory card not writable
14 label supply running out
16 data volume too large for the storage medium
105 no print image / print job
107 layout missing
111 signature in PSV_PCK missing or wrong
114 / 115 printer 1 / 2 not reachable
120 code build rule missing
122 code data wrong
125 code data too long
128 maximum code string length already reached
129 data error, e.g. code substring missing
130 error in the code structure
131 variable longer than the specified number of digits
132 code build rule empty
133 impermissible code content
137 invalid number of digits
138 faulty AI length
141 missing data in PSV_PCK

Other cheap reads that change nothing:

  • SRW_LABELER_STATE (SW98, 14.00): 1 active, 2 passive, 3 standby, 4 error.
  • SRL_LABEL_ROLL_DIAMETER (SL85): current roll diameter in mm – a label shortage predicted rather than reported.
  • SRL_AM_OPERATING_TIME (SL82): seconds in START.
  • SRV_LAST_PCK_INFO (SV70): a full record of the last labelling process – article, weight, tare, belt speed, package length, print and application times. Does not touch the unique buffer.
  • SRV_MULTIPLE_DEVICE_ADDR (SV25): the addresses of the devices making up a multiple labeller.
  • SRW_UNICODE_DEVICE (SW85, 13.00): 0 codepage device, 1 unicode device. This is what the BCS IsUnicodeDevice call wraps.
  • LGV_XL_DATA / LGV_XL_INFO (LV11 / LV10, 13.60) with WZW_HDL = 536 reads the device's message history in pages. Note the handle here is a fixed constant meaning what to read, not a session identifier – worth knowing before assuming WZW_HDL is always a caller's choice.

6. Unique data

The file goes over FTP to uniquePckData/<name>.txt, relative to the FTP home (/bizstorecard/bizerba/public/ on the device). The .txt extension is mandatory: any other name is never read and stays in the folder.

A slave labeller does not get unique data. It prints the field's value from the machine instead. If a value must also appear on the slave, write it to that field on the master by command.

GGW_UNIQUE_DATEN (GW7D, 15.20) is documented, contrary to what the English edition suggests: range 0/1, 0 = without, 1 = with.

XCX_DELETE_UNIQUE_DATA (XX13, 15.20) carries the ordering rule in its own description: "delete all unique data, only possible when unique data is deactivated (GGW_UNIQUE_DATEN = 0)".

SRW_UNIQUE_PCK_DATA_READY (SW9B, 16.40) is settled as a flag, not a count: 0 = unique data not ready, 1 = ready. As a low-stock warning it is useless; the "ran out" event is PSL_PCK_ERR_FLAGS bit 24.

Errors specific to the feature

Code Symbol Meaning
22551 PAE_UNIQUEDATA_NO_VALID_PUBLIC_FILE no valid file in /public/uniquePckData/
22552 PAE_UNIQUE_BUF_EMTPY buffer empty on read → switch buffer; error only when all are empty
22553 PAE_UNIQUE_EMTPY_LINE blank line read in the file
22554 PAE_UNIQUE_LINE_LENGTH maximum line length exceeded
22555 PAE_UNIQUEDATA_DELETE_FAILED deletion failed
19768 PAE_ABLX_UNIQUE_DATA error while setting the unique data
30183 PAE_LOCK_UNIQUE_PCK_DATA "no unique data available" – locks labelling
30185 PAE_LOCK_WRONG_UNIQUE_PCK_DATA "wrong structure of the unique data file. The file has been deleted!"

Three consequences worth stating plainly:

  1. A malformed file is deleted by the device and the line locks. So "the file disappeared" does not mean "the file was accepted" – it means one of two opposite things, and only the log tells you which.
  2. There is a maximum line length. The number is not printed; it is a real limit and a generator has to respect it.
  3. There are several buffers, and the "empty" condition is only an error once all of them are. Remaining-code counting is not a single number.

Also: PAE_LOCK_MEM_QUOTA (30204) – space in the public directory running low or full. And PAE_LOCK_BANDETECTED (30211): repeated bad credentials temporarily ban a service (SFTP...), so an upload client must not retry a failed login blindly.

7. Where a barcode's content actually comes from

Nothing sends a finished barcode to the device. The label carries a code build rule (Codeaufbauvorschrift, GGT_CAB / GT20, selected per code field by GGL_CABNR1..7), and an interpreter on the device assembles the string from that rule plus the current package data. So "which field does the marking code travel in" is a question about the rule, not about the protocol.

The pieces the rule can insert, for the parts that matter here:

Identifier Value What it inserts
CODE_PARAM 0001 code family, ratio, module width, height, flags
CODE_KONST 0002 a constant, max 30 characters (} must be doubled)
CODE_TEILSTR1..7 0005, 0034–0037, 0230, 0231 code substrings GGT_CTS1..7
CODE_TEXT1..20 0118–0137 the full text fields
CODE_ETX1..30 0192–0201, 0364–0383 the simple texts GGT_SIMPLE_TXT1..30 – ETX11–30 from 15.60
CODE_ANUM1..20 0105–0107, 0202, 0203, 0232–0246 the free numbers GGL_ANUM1..20
CODE_AI01, CODE_AI21, CODE_AI93... various application identifiers
CODE_GS 0308 a GS separator character
CODE_FNC1 0047 the Code128 FNC1
PA_CODE_ENCODING 0351 character encoding: 0000 ASCII, 0001 ISO8859-1, 0002 UTF-8, 0003 CP936. Goes second, right after CODE_PARAM.

Formats sit in braces after the identifier. The one that matters for text: CODE_FORM_STELLENANZ (class J), a four-digit count where 0000 means the whole text, and any other value takes that many characters from the start and pads short texts with spaces.

Code flag bit 15, for CODE_GS1DATAMATRIX: 0 = FNC1 as separator, 1 = GS.

Which means a GS1 element string is not one field. (01) GTIN + (21) serial + GS + (93) tail is assembled from an AI, a constant or substring, a simple text, CODE_GS, another AI and another simple text. To find out what a given label expects, read GGL_CABNR1 (GL4A) for the rule number, then GGT_CAB (GT20) for the rule itself. That answers it without asking anyone.

Relevant failures: PAE_ETI_CODE_DAT_LEN (15781) "code data too long for the code family", PAE_ETI_CODE_OVF (15769) "code field too small", PAE_ETI_CODE_DRU_RESOLUTION (15770) "printer resolution unsuitable for the code family".

8. Package-synchronous odds and ends

  • PSV_CTRL (PV02) still exists but is marked "do not use for new developments", and so is its handle PSW_CTRLHDL (PW0A). Use LGV_SEQUENZ.
  • PSV_DATA (PV04) and PSV_STSTK_INFO (PV10) both carry an explicit "this command must not be logically acknowledged". Answering them is an error, and there is a code for noticing it: PAE_NET_WD_QUIT_TOO_MUCH (17285), "more acknowledgements for PSV_PCK received than were sent".
  • PSW_PCK_VAL_REQ (PW12, 6.60) is the device asking a real-time host to supply data for one specific package, sent when the package's leading edge passes trigger 3. The host answers with PSV_PCK_VAL (PV08) carrying the same handle. This is the mechanism for per-package data without a file at all – and it needs the spontaneous channel and hard real time.
  • PSW_CHANNEL_DISABLE (PW13), inside PSV_PCK_VAL, suppresses printing per channel for a single package.
  • AMW_TRIG2_MSG_AKTIV (AW75): 0 off, 1 released by a PLU change or a command, 2 (from 8.62) released only by XCW_PCK_SYNC. Value 2 is the one that stops a PLU change from implicitly releasing a held package.
  • Footnote 47: on fatal errors ("send failed") the GX puts itself into the locked state. A controlling program that drops its connection at the wrong moment stops the line.