← All projects

Finding a dead channel in a 40-finger downhole caliper

During my internship at GOWell, one sensing finger on an MFC40C-G imaging caliper was reading pinned at full scale. This is how I traced it through the signal chain to a single IC.

Where
GOWell Oilfield Technology, Dubai
When
June–July 2026, about 5 weeks
My role
Diagnosis, repair support, calibration

What the instrument does

An imaging caliper goes down an oil or gas well and measures the shape of the borehole as it travels. The MFC40C-G does this with 40 spring-loaded fingers spread around the tool. Each finger presses against the well wall, and its position is measured by an LVDT, a transformer whose output changes as a core moves inside it. Read all 40 fingers together, many times per second, and you get a picture of the inside of the well: corrosion, deformation, buildup.

MFC40 imaging caliper on a workbench with its sensing fingers spread open
An MFC40 caliper on the bench in Dubai, its sensing fingers spread open.

GOWell builds and services these tools for oilfield service companies, including SLB, Halliburton, Baker Hughes and Weatherford. Over the internship I worked with 12+ electromechanical tools on the bench. This caliper was the one I got to know best.

The signal chain

To find a fault, you first need to know what a healthy signal looks like at every stage. Here's my simplified version of the measurement chain. Click a stage to see what it does and what I checked there.

Click or tab through the stages. The red outline marks where the fault turned out to be.

drives the primary coil Excitation≈ 4 kHz AC LVDT fingers× 40 Analog muxselect channel Conditioningsignal-proc. IC ADCraw counts Acquisitionfinger positions

Tracing the fault

The symptom. One finger channel read at full scale no matter where the finger was. Every other channel behaved normally.

The approach. Rather than swapping parts and hoping, I worked through the chain one stage at a time:

  1. Used the wiring diagrams to map exactly which connectors, traces and components that one channel passes through.
  2. Measured the signal with an oscilloscope at points along that path and compared it with the same points on a healthy channel.
  3. Found where the healthy and faulty paths stopped matching, and used what I knew about each stage to work out which part could cause what I was seeing.
  4. Narrowed it to a signal-processing IC in the conditioning stage.

The result. The IC was replaced, and the channel came back with normal readings.

A downhole tool on the bench with its housing open and the electronics board exposed
A tool opened up on the bench so the electronics can be reached for testing.
Close-up of a GOWell electronics board and wiring harness inside a tool housing
Close-up of the electronics and wiring harness inside a tool's housing.

What I did

  • Scope measurements and signal tracing
  • Reading the wiring diagrams and mapping the channel
  • Isolating the fault to the IC
  • Supporting the repair and retest

What I didn't do

  • The SMD replacement of the IC itself
  • Design any part of the tool. I diagnosed an existing commercial design.
ADC full scale
What the symptom looks like: one channel pinned at the top of the range while its neighbors read normally. This is an illustration, not the actual measured data.

Calibration and verification

Electronics drift with temperature, and downhole tools get very hot. I helped collect the data used to correct for that:

  • Placed the tool in a hot-air oven and stepped the temperature in increments of about 10 °C.
  • At each step, set the fingers to known positions with fixtures and recorded the raw sensor counts.
  • Ran continuity checks before and after each run, so a wiring fault couldn't masquerade as sensor drift.

It's slow, careful work, and it showed me how much of real engineering is verification.

What I learned

Know the healthy signal first. The fastest way to find a fault was knowing what every stage should look like. Debugging became comparison, not guesswork.
Read the system before the scope. Half the work was understanding the architecture from the wiring diagrams before taking a single measurement.
Explaining it is part of the job. I ended the internship with a technical presentation on the MFC40 system, followed by an assessment.

I also sat in on other investigations, including an EPDT tool with a suspected secondary-coil fault, and observed MCS equipment being tested. I was there mostly to learn on those, so I don't count them as my work.

← Back to all projects Next: Skribe, ESP32 audio streaming