When thin sheet work goes wrong, most people start by adjusting parameters. But the real variable is usually whether the workpiece is actually held down. One concrete number makes the point: a 193×144mm 5-inch frame bottom plate on a regenerative-blower vacuum table gets an actual hold-down force of about 83 kg. That sounds like plenty. But once it is cut into several small parts, each only a few square inches in area, the hold-down force drops to a dozen or so pounds. That is why small parts need tabs.
The One-Minute Version
- Vacuum hold-down force = sealed area × effective pressure differential; in practice, don't count on more than 85% of the theoretical value. A regenerative blower produces only 4–5 psi; a rotary vane pump gets close to the full 14.7 psi.
- Hold-down force drops linearly with area. A full sheet holds; small parts cut from it don't. That is the fundamental reason for tabs, not just a precaution.
- Without vacuum, don't use onion skin. Practitioners report that onion skinning almost never works without vacuum hold-down.
- Tabs have a published spec to follow: about 4.76mm wide, half the material thickness deep, and the spacing between nodes on the part edge must be ≥ 10.16mm to fit one.
- Don't use foam tape. It's springy, which effectively leaves a compressible gap under the workpiece.
- Chatter suppression has measured results: a flexible support fixture cut dominant-mode vibration by a factor of 65, and mirror support reduced amplitude by more than 90%.
How to Calculate Vacuum Hold-Down Force
The formula is simple: hold-down force = sealed area × pressure differential. One atmosphere is about 14.7 psi, so in theory a 10-inch-square workpiece can be held with about 1,500 lb of force.
In practice there are two discounts. The first is what the vacuum source can deliver: a common regenerative blower produces only a 4–5 psi (8–10 inHg) differential, while a rotary vane or liquid ring pump can approach the full 14.7 psi. The second is a practical derating: because of imperfect seals, surface roughness, and dynamic cutting forces, actual hold-down force should not be taken as more than 85% of the theoretical value.
Vacuum Hold-Down Force vs. Part Area
Calculated at a 5 psi effective differential from a regenerative blower, derated to 85%
| Vacuum source | Achievable differential | Share of theoretical | Best suited for |
|---|---|---|---|
| Shop vacuum | 2 – 2.5 psi | 14–17% | Barely holds large sheets |
| Regenerative blower | 4 – 5 psi | 27–34% | Large areas, tolerates leakage |
| Rotary vane / liquid ring pump | Close to 14.7 psi | Close to 100% | Small areas, high hold-down requirements |
| For a typical CNC router, a working vacuum of 18–24 inHg is sufficient. Measured data: a regenerative blower produced a 250 mbar differential under a 50×50cm plywood panel; with a porous MDF bleeder board it dropped to about 150 mbar, a 50% loss. That shows how much the choice of bleeder board affects system performance. | |||
Spoilboards: More Than Something to Cut Into
A spoilboard has two jobs, and most people only use the first: protecting the table, and providing backing support on the exit side. The second matters especially for CFRP. Research shows that with backing support, exit delamination stayed at 1.00–1.02 across the full 1,403-hole tool life.
| Item | Recommended practice | Why |
|---|---|---|
| Material | 3/4″ MDF | Porous sawdust-and-resin substrate that lets air pass through the entire surface |
| Required preparation | Mill at least 0.5mm off each face | Removes the factory sealing finish; otherwise it isn't porous and the vacuum doesn't work at all |
| Surfacing tool | 3″–4″ fly cutter @ 12,000 rpm | A large diameter covers the surface in one sweep for a flat finish |
| Edges | Edge-seal or coat with epoxy | Prevents air leaking through the sides and moisture warping |
| Bleeder board | 6mm MDF milled down to 3mm | Neoprene cord set into the plenum groove as a perimeter seal |
| When to resurface | When the same toolpath cuts through noticeably differently at different spots | The surface is no longer flat because of grooves and warping |
| MDF warps in environments with large humidity swings; check it more often during humid seasons (in Taiwan, the spring plum-rain season). Once a spoilboard is surfaced, don't remove and reinstall it: as soon as it comes off, it is no longer coplanar with the machine's XY plane, and the surfacing was wasted. | ||
Double-Sided Tape: The Wrong Type Is as Good as None
Foam tapes, including VHB, are springy. It is like putting a compressible pad under the workpiece: the part sinks as the tool pushes down and springs back as the tool lifts. The result is inconsistent depth of cut, an invalid Z zero, and a complete defeat of the purpose of backing support.
The right choice is thin, high-strength double-sided tape, commonly described in the industry as being about as thin as paper tape.
| Item | What to consider |
|---|---|
| Cutting force | Estimate the maximum side load and make sure adhesion has margin |
| Part mass | Light parts can rely on adhesion alone; heavy parts need additional clamping |
| Toolpath | Whether there are long stretches pushing in one direction |
| Bonded area | The smallest part area after cut-off is what matters, not the area of the whole sheet |
| Surface condition | Release agent residue on carbon sheet greatly reduces adhesion |
| Temperature | Heat from machining reduces adhesion |
| Backup protection | If the tape fails, where will the part fly? |
Tabs: There's a Published Spec You Can Copy
Leaving tabs is the most reliable way to protect small parts, and it doesn't require a vacuum system. SendCutSend publishes an explicit tab spec for CNC-routed parts:
Tab Dimensions and Placement
Based on SendCutSend's published tab spec for CNC routing
Onion skinning (leaving 0.1–0.2mm uncut all the way around) can be very stable with vacuum hold-down. But practitioners put it bluntly: without vacuum hold-down, onion skinning almost never works, because as soon as the sheet shifts, the tool cuts straight through and flings the part.
Another practical observation: the main causes of parts flying off are usually a dull tool raising cutting forces and an undersized vacuum pump, not a wrong choice of workholding method.
Chatter: Suppression Methods with Measured Results
Chatter in thin sheet is self-excited vibration driven by the regenerative effect, compounded by the low stiffness and poor damping of thin-walled parts. An academic review compiled the measured effectiveness of several suppression methods:
| Method | Effect | Feasibility for desktop CFRP machining |
|---|---|---|
| Multi-head support | Dominant-mode vibration reduced by up to 98% | Needs a dedicated fixture; suited to repeat production |
| Mirror support | Amplitude reduced by more than 90% | Needs twin spindles or a synchronized support head |
| Flexible support fixture | Dominant-mode vibration reduced by a factor of 65 | Requires fixture design |
| Paraffin wax filling | Amplitude reduced 46–69% | Not suitable for CFRP; residue contaminates bonding surfaces |
| Air-jet support | Response amplitude reduced 50% | Stirs up conductive carbon dust |
| Full-surface vacuum table hold-down | — | The most practical approach for desktop machining |
| The figures above come from a review paper on chatter suppression in thin-wall milling. Most of the studies covered metal thin-walled parts rather than CFRP sheet, so the mechanisms carry over but the numbers cannot be applied directly. For desktop CFRP sheet machining, bonding the workpiece fully to a flat spoilboard already solves the vast majority of chatter problems, because the sheet has no unsupported spans. | ||
Choosing Among Three Workholding Strategies
| Sheet thickness | Large parts (> 100mm) | Medium parts (30–100mm) | Small parts (< 30mm) |
|---|---|---|---|
| 0.5 – 1 mm | Vacuum + spoilboard | Vacuum + tabs | Vacuum + tabs (required) |
| 1.5 – 2 mm | Vacuum or thin double-sided tape | Vacuum + tabs | Tabs (required) |
| 2.5 – 3 mm | Vacuum, double-sided tape, or clamps | Vacuum or double-sided tape | Tabs |
| 4 – 6 mm | Clamps are enough | Vacuum or double-sided tape | Tabs |
| Clamps are simplest for large, thick parts, but the downside is that they get in the way of the toolpath, so clamp positions have to be reserved when you lay out the sheet. A vacuum table doesn't have this problem, which gives it an extra advantage in nesting density. | |||
When results aren't right, check in this order: 1. Did the workpiece move? (look for shift marks on the spoilboard) → 2. Is the tool dull? (test cut with a new tool) → 3. Is Z zero correct? (re-run tool setting) → 4. Only then adjust parameters. Most people go in the reverse order and spend a lot of time on parameters without any improvement.
FAQ
I don't have a vacuum table. Can I still cut 1mm thin sheet?
Yes, but you have to rely on full-surface bonding. Use thin, high-strength double-sided tape to stick the whole sheet to a surfaced spoilboard so that nothing underneath the sheet is unsupported. That both solves chatter and provides backing support on the exit side. The downside is that thin sheet deforms easily when you peel it off, so use heat or a dedicated solvent to help. Small parts still need tabs.
How do I remove the tabs after cutting?
Lightly sand along the edge with fine sandpaper or a diamond file; SendCutSend's official guidance is also that tabs can be sanded off. Do not snap them off with pliers. Carbon fiber has low interlaminar strength, and forcing a tab to break causes local delamination at the tab location that can run into the part itself. Also, dust from tab removal is generated very close to your breathing zone, so wear P100 protection.
Why does the bleeder board on a vacuum table cut suction in half?
Because the bleeder board itself is a flow restriction. Measured data shows a regenerative blower producing a 250 mbar pressure differential under a 50×50cm plywood panel, dropping to about 150 mbar with a porous MDF bleeder board: a 50% loss. That is a necessary trade-off. The bleeder spreads the vacuum evenly; without it, only the area directly above the grooves holds. The fix is a bigger vacuum source, not removing the bleeder.
Can hold-down clamps damage carbon fiber sheet?
Yes, if you tighten metal clamps directly onto it too hard. Carbon fiber sheet is far weaker in compression through its thickness than in-plane, and concentrated local pressure can leave dents or even cause delamination. In practice, put a layer of soft material (such as cardboard or thin rubber) between the clamp and the workpiece to spread the load, and tighten only until the workpiece won't move. There's no need to crank it down.
Why do my parts fly off halfway through the cut?
The three most common causes, in order, are: a dull tool raising cutting forces (practitioners report this is the number one cause), an undersized vacuum source, and the wrong cut-off order. The outer profile should come last for each tool so the part separates from the sheet as late as possible. Also check how much area the part has left once it is cut free: if it is only a few square inches, the hold-down force may be only a few kilograms, which was never going to hold it.
Sources
- CNCCookbook, Router Vacuum Table CNC DIY (hold-down force formula, regenerative blowers at 4–5 psi, pump selection). Accessed September 3, 2026. https://www.cnccookbook.com/router-vacuum-table-cnc-diy/
- Rivcut, Vacuum Fixturing Thin Wall Parts (practical derating: no more than 85% of theoretical). Accessed September 3, 2026. https://www.rivcut.com/blog/vacuum-fixturing-thin-wall-parts
- Mekanika, How to Design a Vacuum Table for Your CNC Machine (regenerative blower at 250 mbar; drops to 150 mbar with a bleeder board). Accessed September 3, 2026. https://www.mekanika.io/en/blog/cnc-milling/how-to-design-a-vacuum-table-for-your-cnc-machine
- CanCam, Spoilboard Surfacing and Improving Suction on a Vacuum Table (3/4″ MDF, 0.5mm removed from each face, fly cutter at 12,000 rpm). Accessed September 3, 2026. https://www.cancam.ca/learning-centre/spoilboard-surfacing-and-improving-suction-on-a-vacuum-table/
- SendCutSend, Nesting Guidelines (CNC routing tabs: 0.1875″ wide, half the material thickness deep, node spacing ≥ 0.400″). Accessed September 3, 2026. https://sendcutsend.com/guidelines/nesting/
- WoodWeb Knowledge Base, Tabs vs Onion Skinning (onion skinning rarely succeeds without vacuum hold-down; parts fly off mainly because of dull tools and undersized pumps). Accessed September 3, 2026. https://woodweb.com/knowledge_base/Tabs_Vs_Onion_Skinning_To_Hold_Pieces__818617.html
- Bantam Tools Support, Double-Sided Tape (thin, high-strength tape recommended; foam tape is a poor choice for machining). Accessed September 3, 2026. https://support.bantamtools.com/hc/en-us/articles/115001658494-Double-Sided-Tape
- Sun, Yan, Sun et al. (2025), A review of chatter suppression in thin-wall milling, International Journal of Extreme Manufacturing 7(6):062003 (measured effectiveness of each suppression method; most studies cover metal thin-walled parts). https://doi.org/10.1088/2631-7990/adec24
- Gutiérrez, Meseguer, Muñoz-Sánchez & Feito (2023), Coatings 13(8):1440 (a backing plate kept exit delamination at 1.00–1.02 over a full 1,403-hole tool life). https://doi.org/10.3390/coatings13081440
- ToolsToday, Cutting Carbon Fiber on a CNC: Tips, Safety, Bit Selection (vacuum table as the preferred workholding for thin carbon fiber sheet). Accessed September 3, 2026. https://toolstoday.com/learn/cutting-carbon-fiber-on-a-cnc-tips-safety-bit-selection
±0.05mm Accuracy Only Means Something When the Part Is Held Securely
Carbon-CNC offers an optional vacuum table, and Z travel of 100mm (Pro) / 80mm (Lite) leaves room for a spoilboard and fixtures. To discuss which workholding method fits your part sizes, contact us.
Get a Quote