IMDTouchKnowledge BaseShop main store

Troubleshooting

Fixing HMI macro error C70 after EasyBuilder migration

HMI and touchscreens Published 5 min read

Short answer

Error C70 appears because EasyBuilder Pro checks macro array bounds at compile time and EasyBuilder 8000 did not: this macro declares short buffRd[10] while its final SetData writes 12 elements from buffRd[0]. Decide the output contract at RW 20-31, then either declare short buffRd[12] and assign the two added elements, or cut the transfer count to 10.

On this page
  1. What C70 is telling you
  2. Audit every array with the arithmetic
  3. Decide the output contract at RW 20-31
  4. Apply the correction
  5. What a larger declaration does not fix
  6. Verify the corrected macro
  7. Questions asked

Error C70 is EasyBuilder Pro refusing to compile a macro whose transfer count exceeds a declared array. In the migrated project the macro declares short buffRd[10] and ends with SetData(buffRd[0], "Local HMI", RW, 20, 12) — a request for 12 elements from a 10-element array. EasyBuilder 8000 compiled the same code because it did not enforce this check. Correct the bounds before looking at drivers or screens: either the destination at RW 20 through RW 31 genuinely needs 12 words, in which case declare short buffRd[12] and assign the two added elements, or the payload is the 10 words the macro actually processes, in which case change the count to 10.

What C70 is telling you #

C70 fires during macro compilation. The compilation result names the macro, and compiling that macro on its own reproduces the array sizing failure. Because the project stops at compile, the controller link, the register addresses and the screen bindings are not involved yet; those branches matter only after the project builds and runs.

EasyBuilder Pro verifies array correctness where EasyBuilder 8000 did not. A clean compile in the old editor therefore proves nothing about a migrated macro's bounds, so audit every declaration in every macro brought across, not only the one the error points at.

Audit every array with the arithmetic #

The rule the compiler applies: the highest index a transfer touches equals its start index plus its element count minus one, and that must stay at or below the declared length minus one.

Applied to the failing call:

  • Start index 0, count 12: last index touched = 0 + 12 − 1 = 11.
  • Declared length 10: highest valid index = 10 − 1 = 9.
  • 11 > 9, so the call reads two elements, buffRd[10] and buffRd[11], that were never allocated. That is C70.
short buffRd[10]
...
SetData(buffRd[0], "Local HMI", RW, 20, 12)

The same rule clears the other three declarations:

Array Declared Highest index reached Where Result
buffRd 10 11 final SetData of 12 elements from buffRd[0] exceeds by 2
buffChk 20 19 unpack loop: at n = 9, ptr = 9 × 2 = 18, then ptr + 1 = 19 fits exactly
chePunz 41 40 41-element read from index 0; search runs p = 1 to 40 fits
buffPosMatrix 20 19 lookup loop n = 0 to 19; 20-element write fits

Only the final write from buffRd breaks the rule. The byte-unpacking loop reaches index 18 at n = 9 and index 19 after the increment — the last element of a 20-element array — which is legal and is not the failure.

Decide the output contract at RW 20-31 #

The macro reads 10 words from RW 0, unpacks them into 20 bytes, looks up 20 position codes, repacks 10 words and writes them from RW 20. A 12-element write covers RW 20 through RW 31; a 10-element write covers RW 20 through RW 29. Both silence C70, but they implement different outputs:

Required output Correction Extra obligation
12 words at RW 20-31 change the declaration to short buffRd[12] assign buffRd[10] and buffRd[11] before the write
10 words at RW 20-29 change the final count from 12 to 10 none; RW 30 and RW 31 stay untouched by this call

The decision rests on what consumes RW 30 and RW 31: screen objects, other macros, controller logic reading the panel's local words. The macro itself only ever defines 10 output words — the read at RW 0 asks for 10 and the repack loop fills indexes 0 through 9 — so unless a consumer needs those two addresses, the 10-word transfer is the correction that matches what the code does, and a 12-word write would push two never-assigned values into RW 30 and RW 31. Commit to short buffRd[12] only when the receiving layout specifies 12 words.

Apply the correction #

  1. Save a separate migration copy of the project before editing the macro.
  2. Open the macro named in the compile result and confirm the pairing: short buffRd[10] against SetData(buffRd[0], "Local HMI", RW, 20, 12).
  3. Search the project for consumers of RW 30 and RW 31, including screen objects and other macros, and choose the output contract from that evidence.
  4. For the 10-word contract, change the final call to SetData(buffRd[0], "Local HMI", RW, 20, 10).
  5. For the 12-word contract, change the declaration to short buffRd[12] and assign the two added elements from the receiving layout before the write.
  6. Compile the macro, then compile the complete project for the replacement terminal, clearing any further diagnostics one at a time.
  7. Leave the position-code transfer SetData(buffPosMatrix[0], "Drivers", 4x, 1#8194, 20) untouched; 20 elements out of a 20-element array is within bounds.

What a larger declaration does not fix #

Two gaps survive the compiler, and C70 catches neither:

  • The added elements have no source. The GetData at RW 0 reads 10 elements and the repack loop writes indexes 0 through 9, so nothing in the macro ever sets buffRd[10] or buffRd[11]. Raising the input read to 12 would change the input contract at RW 0-11 and still would not define the outputs; their values have to come from the application's data layout.
  • An unmatched byte leaves a hole. buffPosMatrix[n] receives a value only when src == tst for some p between 1 and 40; a byte absent from chePunz[1] through chePunz[40] leaves that position code unassigned on that pass. That is a run-time data question, so specify the behavior — reject the input, flag it, or assign an agreed fallback — and test it deliberately.

Verify the corrected macro #

  • Confirm the SYNC_TRIG_MACRO(2) sequence completes and the macro executes once per trigger under controlled conditions.
  • Load known words at RW 0-9. The repack reconstructs each word as low byte + 256 × high byte, so RW 20-29 must read back identical to RW 0-9; a mismatch means the unpack or repack path was disturbed.
  • On the 12-word contract, check RW 30 and RW 31 against their specified values, not against leftovers.
  • Read the 20 position codes at the Drivers destination 1#8194 and compare them with the positions expected for the bytes loaded at RW 0-9.
  • Keep compile-time and run-time failures separate from here: a project that compiles but reports no response from the controller is a driver and device question, the kind behind an EasyBuilder Pro upgrade that shows HMI no response, not a macro bounds problem.
  • Carry the same discipline into the rest of the conversion by validating HMI variables during a software migration, because the compiler accepts addressing and naming differences the old project hid.

Questions asked

What does error C70 mean in an EasyBuilder Pro macro?

It is a compile-time array bounds error: a transfer or index reaches past the declared length. Here the final SetData requests 12 elements from buffRd[10], whose valid indexes end at 9. The compilation result names the macro, and the project cannot be built for the replacement terminal until the mismatch is removed.

Why did the same macro compile in EasyBuilder 8000?

EasyBuilder 8000 did not enforce this bounds check, so its clean compile never proved the transfer stayed inside the array. Treat every macro leaving the old editor as unverified and run the start-index-plus-count-minus-one check against each declaration before compiling in EasyBuilder Pro.

Should the declaration become buffRd[12] or the count become 10?

The destination decides. If screens or logic read RW 30 and RW 31, the application needs 12 words, so declare short buffRd[12] and assign the two added elements before the write. If nothing consumes those addresses, change the count to 10, because the macro only ever defines 10 words.

Is enlarging the array declaration enough on its own?

No. The GetData at RW 0 reads 10 elements and the repack loop fills indexes 0 through 9, so buffRd[10] and buffRd[11] would reach RW 30 and RW 31 without ever receiving a value. Define their contents from the receiving layout, or take the 10-word branch instead.

Do the other arrays in this macro also overrun?

No. buffChk peaks at index 19 of 20 when the unpack loop reaches n = 9, chePunz is read across all 41 declared elements while the search uses 1 through 40, and buffPosMatrix is written across its full 20. Only the 12-element write from buffRd is out of bounds.

Will C70 also catch values that were never assigned?

No. C70 compares counts against declarations at compile time; it does not track whether an element holds a defined value at run time. The unmatched-byte case in the lookup leaves buffPosMatrix elements unassigned without any diagnostic, so that behavior must be specified and tested separately.

Keep reading

  • Troubleshooting

    Fixing TwinCAT HMI publish failures after a NuGet update

    The publish task fails with every symbol unresolved because TF2000, TE2000 and the HMI NuGet packages are at mixed versions after a partial TcPkg upgrade. Update all three groups together on the engineering PC and the IPC, redeploy an empty project to prove the path, then republish: the failure sits in the environment, not the project.

    HMI and touchscreens

  • Troubleshooting

    Fixing an HMI Project Upload That Reports Communication Failed

    Treat a failed HMI project upload as three separate branches: the installer's parallel-port dongle driver message is a licensing fault, a "communication failed" report before any progress is a transfer-path mismatch, and a "file system error" after the panel answers is a storage or destination fault. Identify the layer from the message and the moment it appears, fix that layer only, and retest. A dongle fault never blocks the cable, and a successful ping never proves the transfer service is up.

    HMI and touchscreens

  • Troubleshooting

    Fixing an HMI Project That Fails After a Git Clone

    The clone is broken because a referenced bitmap .PLK was never committed, not because of an ignore rule. Read the HMI Output tab for the exact filename, prove the file is untracked with git check-ignore followed by git ls-files, then restore and commit it and re-test from a clean clone. Track the .tgt and .def files for the HMI BIOS version as well.

    HMI and touchscreens

Fixing EasyBuilder 8000 error 0xc0000005 at startup