Why Do Beam Cutting Robots Cut the Wrong Holes and Copes From Correct DSTV Files?
What file formats does an H beam plasma cutting robot read - DSTV, NC1 or DXF?
The working format is DSTV NC, almost always saved with the .nc1 extension (you will also see .nc and .dstv). It is plain ASCII, one file per part, and it is a data standard set by the German steel construction association, not a proprietary format - so it is the same thing your detailer exports from Tekla Structures or Advance Steel. Inside, the file is split into coded blocks: ST for the header (piece mark, quantity, profile, grade, length), BO for holes with face, position and diameter, AK and IK for outer and inner contours such as copes, notches, birdsmouths and web openings, SC for saw cuts and bevels, KO and SI for marking and fit-up lines, and EN to close the part. The YOMI beam robot reads that structure and drives up to ±45 degree beveling, 12 m of effective length and 5000 kg of profile weight from it. DXF is for flat plates and does not describe a 3D member, so do not expect a plate nesting file to run a beam.
The drawing is right, so why are my web holes landing 2-4 mm off after cutting?
Because DSTV coordinates are measured from a nominal profile, and rolled sections are not nominal. The standard puts the origin at one end of the member and dimensions holes from a reference edge, and each hole carries a reference flag - bottom edge, top edge, or axis. If your exporter defaults every web hole to the bottom edge, the whole hole pattern inherits the rolling tolerance of the section depth, which EN 10034 allows to be a couple of millimetres and more on deep sections. Detailers normally reference web holes from the top flange and flange holes from the flange centre, so the tolerance lands where it does no harm. The second half of the fix is on the machine: this robot measures the actual web height and flange width with a laser before it cuts and compensates the toolpath, which is why it can hold ±1.5 mm over a full 12 m member. So: set the reference flags correctly in the model, keep the measuring system clean and calibrated, and stop adjusting the drawing to chase a tolerance that belongs to the mill.
Which data mistakes actually stop the machine or scrap a beam?
In practice five, and none of them are machine faults. Profile not in the controller library - the ST block names a section the control does not recognise and the part is rejected, so confirm your YM-600/1000/1250/1500 web range and keep the library updated. Units - DSTV is millimetres; an imperial job exported without conversion quietly scales or misplaces everything. Mirrored pairs - left and right members copied instead of mirrored, which produces two identical handed beams and a fit-up problem on site. Cope on the wrong flange, or cope depth taken from the drawing rather than the connection, so the machine reproduces the error perfectly. Marking text longer than the marking window, which truncates the piece mark and leaves fitters unable to match members to the erection drawing. A sixth, quieter one: quantity in the header not matching the bundle actually released, which is how you get 3 beams instead of 4.
What should I check in an NC1 file before releasing the package to the shop floor?
Run a five-minute pre-production check on the file, not on the drawing. Header: piece mark matches the erection mark, grade is what the job specifies, profile name and its stated dimensions agree - an HEA200 that claims 300 mm depth is a red flag - quantity and length match the cut list. Holes: count per face against the connection detail, and verify gauge and pitch on at least one bolt group per face; bolt holes should be 2-3 mm over bolt diameter in metric practice (18 mm for M16), 1/16 in imperial. Contours: cope depth and length at both ends from the AK block, and confirm the cope is on the correct flange. Openings: web penetrations in the IK block match the service-opening schedule. Marking: piece mark present and on a visible face. The cheapest tool is a browser or desktop NC1 viewer that renders the part as dimensioned 2D faces and a 3D model - plotting coordinates by eye is where mistakes survive.
How does the robot know where the beam actually starts on a 12 m member?
It does not trust the nominal end. The profile is fed on a roller frame, the machine references the member end and then measures the real section - web height, flange width, and where the flanges actually sit - before generating the cut path. That is what allows a welded built-up beam with camber and weld distortion, or an imported section with 2-3 mm of depth variation, to still come out inside tolerance. Cutting accuracy is quoted to ISO 9013 / ISO 8206 with ±1.5 mm over the length, cutting speed 10-2000 mm/min and traverse up to 6000 mm/min. Two practical consequences: rust, mill scale and spatter on the measuring reference will bias the measurement, so clean the flange face the sensor reads; and if you feed a member shorter than the program expects, or two members butted together, the end reference is wrong and every hole shifts. Keep the cut list sequence matched to the feed order.
How are piece marks and fit-up lines applied so fitters can match parts to the drawing?
Through the KO and SI blocks - marking runs as part of the same cycle, driven by the same file, so the mark cannot get separated from the part. That sounds trivial until you have a yard full of identical 12 m columns: an erection mark on a visible flange face is the difference between a fitter picking the right member and a re-work. Practical limits to respect: marking adds cycle time, so mark the minimum (piece mark, and fit-up lines only where the drawing calls for them); the marking window has a maximum text length and a truncated mark is worse than no mark; and on galvanized or painted members you need a marking method that survives the process, or you fall back to durable labels plus a paint mark. Many shops run the robot mark on bare steel and then apply the piece mark in paint after galvanizing, keeping the robot mark as the internal traceability reference for the EN 1090 / AISC paperwork.
A revision comes through after release - how do I avoid re-cutting the whole batch?
Re-export only the affected piece and re-run that single file, and make the revision visible in the filename rather than trusting memory - piece mark plus revision letter plus date. The machine is obedient, not clever: it cuts exactly what the current file says, so the only real control is which file is loaded. Keep the released NC package archived per revision, because the first question after a fit-up problem is always what file was actually cut. Also check what the revision touches - a hole moved 10 mm is a single-file re-run, but a change in flange plate thickness on a built-up member changes the web height the machine measures, which can invalidate a cope that was previously fine. One habit pays for itself: run a first-article beam with a few holes, one end cope and one bevel before releasing a full package, and measure it. One test member costs a fraction of a scrapped bundle.
What is the fastest way to get from Tekla to a good first beam on the robot?
Set the interface once, properly, then test. Tell your detailer the machine make and model, that you want DSTV in millimetres with marking included and where, your default grade, and the reference-edge convention you enforce - web holes from the top flange, flange holes from the flange centre. Export one simple member: a straight beam with one bolt group per face, one end cope and one 45 degree bevel, and check the file in a viewer before it reaches the machine. Cut it, then measure hole positions from the reference edge, cope depth and length, bevel angle and overall length against the drawing. If that member is good, the package will be good; if it is 3 mm out, you have found a convention problem before it cost you a bundle. This is also the moment to confirm that the profile you use is in the machine library and that the bevel and kerf compensation values match the plasma or oxy-fuel process you will actually run - 170-300 A plasma with a 1.5-3 mm kerf behaves very differently from oxy-fuel on 6-32 mm vertical cuts.
Technical Specifications of H Beam Plasma Profile Cutting Robot Machine
| Parameter | Value |
|---|---|
| Share | |
| Categories | H Beam Cutting Machine |
| Brand | YOMI CNC Cutting&Welding Machinery |
| Product Specifications | YM-600,YM-1000, YM-1250, YM-1500 |
| Cutting method | Plasma/Flame |
| Plasma power source | Herpertherm XPR300A |
| Plasma cutting thickness | Pierce cutting thickness 1-45mm |
| Oxy fuel cutting thickness | Vertical cutting thickness 6-32mm |
| Effective cutting length | 12m |
| Cutting precision in length | ±1.5mm |
| Beam Web Height(YM-600) | 100-600mm |
| YM-1000 | 100-1000mm |
| YM-1200 | 200-1250mm |
| YM-1200 | 200-1250mm |
| Beam flange width | 100-600mm |
| H beam/I beam/Channel steel/angle steel | YM-XH-600, 100-600mm |
| YM-XH-1000 | 100-1000mm |
| YM-XH-1200 | 200-1250mm |
| YM-XH-1500 | 600-1500mm |
| Cutting method | Plasma/Flame |
| Plasma power source | Herpertherm XPR300A |
| Plasma cutting thickness | Pierce cutting thickness 1-45mm |
| Oxy fuel cutting thickness | Vertical cutting thickness 6-32mm |
| Cutting speed | 10~2000mm/min |
| Moving speed | 10~6000 mm/min |
| Maximum profile weight to be cut | 5000Kg Note:Can customized according to offered max weight |
| Beveling cutting | ±45 degree |
| Profile cutting form | Fixed length straight cut, fixed length oblique cut and end socket function |
| Cutting accuracy execution standard | ISO9013-2002\ISO8206-1991\ JB/T10045.4-1999JB |
Product Gallery







Applications
The H Beam Plasma Profile Cutting Robot Machine is widely applied in steel structure prefabricated buildings, bridge and infrastructure construction, shipbuilding and marine engineering, heavy machinery manufacturing, power plant and industrial facilities, and offshore wind power structures.
Learn more about the H Beam Plasma Profile Cutting Robot Machine:
https://www.steelstructurer.com/pid18376141/H-Beam-Plasma-Profile-Cutting-Robot-Machine.htm
For more information about H Beam Plasma Profile Cutting Robot Machine and other steel structure intelligent equipment, contact YOMI CNC Cutting & Welding Machinery. Our team of experts is ready to help you find the best solution for your manufacturing needs.

- Full Name:
- YOMI WhatsAPP
- Tel:
+86 19053498672
- Email:
- sales@yomi-china.com
- WhatsApp:
- 8619053498672
- Address:
- Factory Adress: 9 Changjie Road, Ningjin,Dezhou, P.R.China

