For fifty years, Gerber has moved boards from design to fabrication, and it does that job well. The trouble starts when the same files get reused for something they were never built for: reviewing the design. A Gerber set is the right way to build a board and the wrong way to understand one — and the gap between those two is the whole ODB++ vs Gerber story.
What Gerber contains
Gerber is, at heart, a stack of images. Each layer — copper, solder mask, silkscreen — is a 2D drawing of shapes, paired with a drill file telling the fab where to make holes. It describes, with total precision, the geometry a fabricator needs to manufacture the board: where copper goes, how wide, where the holes are.
That precision is why it endures as a pcb manufacturing data format. A fab’s CAM engineer opens a Gerber set and has exactly what they need to make the board. For that purpose, nothing is missing.
What Gerber does not contain
What’s missing only matters when you try to review from it. Gerber knows shapes, not meaning.
- No netlist. Gerber has no concept of a “net.” Two pads connected by a trace are just two shapes and a third shape between them. Nothing says they’re the same signal, and nothing lets you follow that signal anywhere.
- No component identity. There’s no
U12, no refdes, no part number — just a cluster of pads. The file doesn’t know a chip sits there, let alone which pin does what. - No structure. No stack-up meaning, no hierarchy, no link between a pad and the device it belongs to. It’s a picture of the outcome with the reasoning removed.
Modern Gerber (X2/X3) can attach some netlist and component attributes, but it’s optional, frequently stripped in export, and still image-first — a bolt-on, not the native model. For a dependable gerber review of connectivity, you can’t count on it being there.
What ODB++ adds
ODB++ is a database, not a drawing. It carries the same geometry Gerber does — and then the things Gerber threw away.
- Netlist. Connectivity is first-class. Every net exists as a net, so you can follow a signal through connectors, series parts and planes to where it actually lands.
- Component intelligence. Refdes, package, and pin mapping travel with the design. A pad isn’t just copper — it’s
U12pin 14, with a part behind it. - Structure. Layer stack-up, hierarchy, and the relationships between pads, nets and components — the design as a connected model, not a flat image.
| Carries | Gerber | ODB++ |
|---|---|---|
| Copper / mask / silk geometry | Yes | Yes |
| Drill data | Yes | Yes |
| Netlist (connectivity) | No* | Yes |
| Component identity (refdes, pins) | No* | Yes |
| Stack-up + structure | Partial | Yes |
Why reviewers care
A review asks questions of connectivity and function: where does this signal terminate, what loads this rail, which inputs are unconnected, is this pull-up on the right net. Every one of those questions is about nets and components — the two things Gerber doesn’t hold. From a Gerber set, a reviewer can confirm the copper looks sane and little else. From ODB++, they can interrogate the actual design. That’s why an odb++ review is a different activity from staring at Gerber layers, not just a nicer viewer.
Real examples
The gap is concrete:
- Trace a net. “Where does this reset line go after the series resistor?” In ODB++ you follow it. In Gerber you eyeball touching shapes across layers and hope.
- Identify a component. “Is this the 1.8 V regulator?” ODB++ names the part. Gerber shows a footprint and leaves you to guess from the silkscreen.
- Check a supply. “Does this device’s supply pin reach a real rail?” That needs the pin’s function and the net it’s on — connectivity and component data together. Gerber has neither.
When Gerber is enough
None of this makes Gerber obsolete. It’s still the right tool when the question is manufacturing, not intent: releasing a proven board to fab, a CAM engineer preparing panelization, or a quick visual sanity check that the copper matches what you expect. For handing off pcb manufacturing data, Gerber is doing exactly what it was designed to do.
When ODB++ becomes necessary
ODB++ becomes necessary the moment the question shifts from “can this be built?” to “is this the design we meant?” That’s every real review: functional correctness, connectivity, parts used within their limits. It’s also what makes review possible across a mixed toolchain and out to an EMS partner. ODB++ is neutral — one connected model anyone can open, whatever CAD drew the board. That neutrality is what lets a team review without a CAD license at all.
See what a review looks like on the connected model instead of the images.
See ProbeHow Probe uses ODB++
Probe reviews from the ODB++ you already export for fabrication — or from a schematic and BOM before layout exists. It reads the connectivity and the components, not just the shapes, so every reviewer can follow a net to where it lands, see the function a component belongs to, and search the whole board. The same neutral file that goes to your fab becomes the thing your whole team reviews. No CAD seat — and findings that carry their proof and persist to the next board.
Right format, right job
Gerber isn’t the loser in the ODB++ vs Gerber comparison — it’s the winner at manufacturing, and it should stay there. The mistake is asking it to do review, a job it was never built for. Reach for the format that carries the connectivity and the components, and the review stops being a guessing game over images and starts being an interrogation of the real design.
Probe is in development, launching later in 2026. Bring your own ODB++ file and see it live: in 30 minutes we load your design and run a review on the connected model — nets, components and all. Register your interest, or book a session.
