Pressure Transducers on the IS Machine: Worth the Spend?

It's 2am on a ten-section IS machine in August, and the shift supervisor is staring at a blank-side pressure trace that just dropped half a bar on section 6 for no reason anyone can point to. No alarm. No visible defect on the belt yet. Just a curve that doesn't match the master curve it matched four hours earlier.

But that's the honest pitch for digital pressure transducer rigs on the forming end. They can show you a problem before the cold end does. A transducer only tells you what's happening at the sensor, not what's happening in the mould, the glass, or the operator's head. Plants that install the hardware and stop there usually end up more confident and no more correct.

What a good pressure trace actually shows you

On a well-instrumented section, you're watching blow air pressure, plunger air, and often blank-mould cavity pressure, all timed against the section cycle to within 10ms. That resolution matters. A plunger-up event that drifts 15-20ms on one section while the other nine hold steady is your first sign of mechanical wear, long before it shows as a choked neck or a leaner on the packer.

Correlate that against gob weight. Most forming shops run to a target of ≤0.4% CV, and a transducer rig that's actually being read, not just logged, will show a pressure signature shift on the same section where weight starts to walk. I've had baffle marks blamed on baffle alignment for three shifts running when the real cause was a slow leak in the blank-side air line, a fault that never showed up on a manual check but sat plainly on the pressure curve the whole time.

A pressure transducer doesn't diagnose anything. It just refuses to let the floor guess.

What it doesn't tell you when the sensor's off, or lying

In 2016 I worked with a plant running an older Emhart 8-section machine on cam-timing control, with transducers bolted on as a retrofit two years earlier. Half of them hadn't been recalibrated since installation. The trend screens looked busy and useful. Nobody on shift could tell you which curves were trustworthy and which had drifted so far off zero they were just noise with a nice colour.

That's the failure mode nobody puts in the sales deck. A dead or drifted transducer doesn't fail loud. It keeps producing a trace, the trace just stops meaning anything, and the operator staring at the screen has no way to know that from the screen alone (you'd think a self-check alarm for sensor drift would be standard by now. On the newer rigs, some do have it, so check before you buy).

We hit the same thing on a 2019 audit of a Spanish wine-bottle plant running mixed vintage sections, newer Heye Smart H1 stations sitting next to older refurbished ones. The Heye sections had live drift compensation built into the transducer package. The older sections didn't, and three months of stable-looking pressure data on section 4 turned out to be a sensor holding a fixed offset the whole time. Stones off that section got chased as a batch-house problem for two weeks before anyone thought to pull the sensor.

The handover problem is bigger than the sensor problem

Good instrumentation only pays off if the data crosses shifts. On most lines I've audited, the 0600 handover misses the night shift's pressure trend notes on more than half the sections that had an event. Not because the data isn't there. Because nobody's job is explicitly to hand it over.

The hot-end superintendent should own the call on what a pressure deviation means and whether it triggers a mould change or a recipe hold. The operator shouldn't be adjusting set points off a pressure trace without that sign-off, full stop. I once watched a well-meaning operator chase a phantom pressure dip for forty minutes on a section that just needed the transducer re-zeroed, because nobody had told him that call wasn't his to make alone.

Three ownership questions clear most of this up fast:

• Who owns the transducer calibration schedule, and how often does it actually run
• Who reads the trend data at handover, by name, every shift
• Who has the authority to act on a deviation before it becomes a defect on the belt

Skip those three and the hardware is decoration.

Retrofit or replace: the real cost math

Bolting transducer rigs onto an old machine isn't cheap, and it isn't always the right call. On a machine already scheduled for rebuild inside eighteen months, I'd rather spend the money on a forming audit that shows you where the real variance sits, section by section, before you decide what to instrument.

Where it does pay off is on mixed-OEM fleets under real changeover pressure, particularly across container glass plant builds in the GCC region, where turn-key OEM packages get specified without much thought to whether the plant can actually read the data they generate. A vendor-neutral container glass consultant looking at that instrumentation, tied into proper digitalisation and reporting rather than a standalone trend screen nobody checks, is usually worth more than the sensor spec sheet itself.

On a US plant I audited in 2025, running a mix of legacy and rebuilt sections, we tied pressure trend data into the Job Change Tool's KPI tracking for the first time. Section-level changeover variance fell -22% in the first quarter, once operators could see in real time which section's curve hadn't settled yet before calling first ware good. Not a transducer win. A workflow win the transducer made visible.

The kit is good, better than it was ten years ago, and on a modern machine with proper calibration discipline it earns its keep fast. But it's an instrument, not a decision-maker. The plants that get value from it already know who's supposed to look at the screen, and what they're allowed to do about what they see. If you're not sure your plant clears that bar yet, that's what a forming audit is for.