If you are not collecting data from your deep hole drilling machines, you are flying blind. I have implemented data collection systems in a dozen shops, and every time the data revealed an opportunity that nobody expected. A machine that everyone thought was running at 85% efficiency was actually at 62% when you counted the coolant system downtime that nobody was tracking.
What to Collect
The starting point is spindle runtime, cutting time, and cycle count. These three numbers tell you utilization. Most machine controls already track these. The trick is pulling them out and storing them where you can analyze them.
Beyond the basics, the parameters that matter specifically for deep hole drilling are coolant pressure at the tool, coolant flow rate, spindle load, and feed force. These four parameters correlate directly with tool condition and hole quality.
| Data Point | Sensor Type | Collection Frequency | What It Teaches You |
|---|---|---|---|
| Spindle runtime | PLC register | Per cycle | Machine utilization, scheduling accuracy |
| Cutting time | PLC register | Per cycle | Cycle time trends, efficiency changes |
| Coolant pressure | Pressure transducer | Continuous | Tool condition, pump health, filter status |
| Coolant flow rate | Flow meter | Continuous | Chip evacuation, pump performance |
| Spindle power | VFD feedback | 1-second interval | Tool wear, material variation |
| Feed force | Load cell or VFD | 1-second interval | Drill condition, material hardness change |
| Spindle vibration | Accelerometer | 100 Hz sample | Bearing wear, imbalance, chatter |
| Coolant temperature | Thermocouple | 1-minute interval | Thermal stability, chiller performance |
I installed a simple pressure transducer and flow meter on a machine that was having intermittent chip evacuation problems. The data showed the pressure dropping from 1,000 psi to 650 psi for 30 seconds every time the chip box emptied. The operator had never noticed the pattern because the pressure gauge was on the back of the machine.
Data Collection Methods
PLC Data via MTConnect
MTConnect is an open standard for machine tool data interchange. Most modern CNC controls support it natively. The implementation cost is basically zero beyond a network connection and a data storage platform. I have pulled spindle load, feed rate, and axis position data from Fanuc, Siemens, and Heidenhain controls using MTConnect.
Retrofit Sensor Kits
For older machines without data collection capability, retrofit sensor kits are available. A basic kit with pressure, flow, temperature, and vibration sensors plus a data acquisition module costs $2,000 to $5,000 per machine. Installation takes one to two days.
| Data Collection Method | Cost per Machine | Installation Effort | Data Granularity |
|---|---|---|---|
| MTConnect (native) | $0–$500 | 2–4 hours | 100 ms to 1 sec |
| PLC direct read | $500–$2,000 | 4–8 hours | 100 ms (limited by scan) |
| Retrofit sensor kit | $2,000–$5,000 | 1–2 days | 1 ms to 100 ms |
| Manual data logging | $0 (labor cost) | Per cycle | As good as the operator |
I do not recommend manual data logging for anything other than a temporary study. The data quality degrades after the first week as operators get busy and skip entries. Automated collection is essential for long-term analysis.
Edge Computing vs Cloud
The decision between edge processing and cloud storage depends on the shop infrastructure. I prefer edge computing on a local server with cloud sync for backup. If the internet connection drops, you do not lose data. A Raspberry Pi or industrial IoT gateway can collect data from up to 20 machines and store it locally for $500.
Analyzing the Data
Trend Monitoring
The most valuable analysis is simple trend monitoring. Plot coolant pressure at the tool versus time for the last month. A gradual downward trend means something is wearing. A sudden drop means a failure. I set automated alerts at 10% deviation from the rolling 30-day average for each parameter.
| Parameter | Alert Threshold | Typical Action |
|---|---|---|
| Coolant pressure | 10% below 30-day avg | Inspect filter, check pump |
| Spindle load | 15% above baseline | Check tool condition |
| Cycle time | 20% above baseline | Investigate process change |
| Coolant temperature | 5°F above set point | Check chiller performance |
| Vibration (overall) | 0.1 in/sec above baseline | Schedule bearing inspection |
OEE Calculation
Overall Equipment Effectiveness (OEE) for deep hole drilling needs to account for coolant system downtime, which is often excluded from standard OEE calculations. I define availability as time the machine is ready to cut, including coolant system ready. Performance is actual cycle time divided by ideal cycle time. Quality is good parts divided by total parts.
I tracked OEE at one shop and found that the most expensive machine had the lowest OEE at 54%. The cause was 90 minutes of unplanned downtime per shift for coolant system issues that nobody had been recording because the downtime was attributed to “tool change” in the production log.
Data-Driven Maintenance Scheduling
The data enables condition-based maintenance instead of time-based maintenance. Instead of changing coolant filters every month, you change them when the delta-P trend crosses the threshold. Instead of rebuilding the rotating union annually, you rebuild it when the vibration signature changes.
I moved one of my shops from time-based to condition-based maintenance for coolant pumps and rotating unions. The pump rebuild interval went from 12 months fixed to 14–18 months on average, with one pump going 22 months. The machine monitoring systems article covers the sensors and software in more detail.
Key Takeaways
- Collect coolant pressure, flow, spindle load, and feed force at a minimum for deep hole drilling.
- MTConnect provides zero-cost data collection on modern CNC controls.
- Retrofit sensor kits cost $2,000–$5,000 for older machines without data capability.
- Automated trend monitoring with 10% deviation alerts catches developing failures.
- OEE that excludes coolant system uptime overstates machine productivity.
- Condition-based maintenance extends component life by 30–50% compared to fixed intervals.