Squat detection

4/18/2026  Update- the leg bow moves up/down about 1/4" at the top midpoint.  An inexpensive "coin detector" circuit will mounted in the box structure forward of the leg bow.  Something semi-rigid (like a spring) will be mounted to the bow that will break the IR beam when the wheels are off the ground.  Since hard landings may cause extreme movement, care will be taken to allow for the bow movement.   The circuit will need to be weather proofed if possible, but may be difficult since the transmitter and reciever are mounted to the board.

I asked Grok about integrating wheels on ground and possible CG determination using bend or strain gauges.

 Using a strain gauge or bend sensor on the fiberglass main gear bow of your Cozy MK IV is a promising approach for a squat switch or weight detection system. Both options have merits, but the choice depends on your requirements for simplicity, accuracy, durability, and ease of implementation. Let’s break down the options and consider their suitability for your application, including the potential for gross weight and center of gravity (CG) determination.

### Strain Gauge **How it works**: A strain gauge measures the deformation (strain) of the gear bow by detecting changes in electrical resistance as the material flexes under load. For a squat switch, you’d detect the presence of strain (indicating weight on wheels), and for weight/CG estimation, you’d quantify the strain magnitude. **Pros**: - **High Sensitivity**: Strain gauges can detect small deformations, making them suitable for both binary squat switch detection and quantitative weight measurement. - **Quantitative Output**: With proper calibration, strain gauges can provide data for estimating gross weight and CG by placing sensors on the main gear bow and nose gear strut. - **Compact and Lightweight**: Strain gauges are small, lightweight, and can be bonded directly to the fiberglass gear, minimizing aerodynamic impact. - **Established Technology**: Widely used in aviation for load monitoring, with well-documented installation and calibration methods. **Cons**: - **Temperature Sensitivity**: Fiberglass has a relatively high coefficient of thermal expansion compared to metals, and strain gauges are sensitive to temperature changes. Without compensation (e.g., a Wheatstone bridge with a dummy gauge or temperature sensor), readings may drift, especially in varying environmental conditions. - **Complex Installation**: Requires careful surface preparation, precise bonding, and protection from moisture and vibration. Fiberglass’s composite nature can complicate strain transfer and bonding reliability. - **Signal Conditioning**: Needs amplification and conditioning circuits to convert resistance changes into usable signals, adding complexity. - **Calibration**: For weight/CG estimation, calibration is critical and may need periodic rechecking due to fiberglass creep or aging. ### Bend Sensor (Flex Sensor) **How it works**: A bend sensor (e.g., a resistive flex sensor) changes resistance based on the degree of bending. It could be mounted on the gear bow to detect flexing when the aircraft’s weight is on the gear. **Pros**: - **Simpler Installation**: Bend sensors are typically easier to mount, often requiring only adhesive or mechanical attachment, and are less sensitive to surface preparation than strain gauges. - **Less Temperature Sensitivity**: While not immune, flex sensors are generally less affected by temperature-induced errors in the context of binary detection (squat switch). - **Lower Cost and Complexity**: For a simple squat switch, a bend sensor requires minimal signal conditioning, making it a plug-and-play option for detecting gear deflection. - **Rugged**: Some flex sensors are designed for harsh environments, potentially offering good durability on a landing gear. **Cons**: - **Lower Precision**: Bend sensors are less accurate for quantitative measurements like weight or CG. Their output is typically nonlinear and less repeatable than strain gauges. - **Limited Range**: Flex sensors have a finite bending range, and excessive deflection (e.g., during hard landings) could damage them or yield unreliable readings. - **Durability Concerns**: Fiberglass gear undergoes cyclic loading, and some flex sensors may degrade over time under repeated bending or UV exposure. - **Not Ideal for CG/Weight**: Due to lower accuracy and repeatability, bend sensors are less suitable for precise weight or CG calculations. ### Alternative Solutions 1. **Microswitch or Limit Switch**: - **Description**: A mechanical microswitch mounted to detect gear deflection or contact with the ground. - **Pros**: Extremely simple, reliable, and binary (on/off) for squat switch use. No temperature compensation needed. - **Cons**: Not suitable for weight/CG measurement. May require mechanical modification to the gear or airframe, and moving parts can wear or jam. - **Use Case**: Best for a basic squat switch if weight/CG isn’t a priority. 2. **Optical or Laser-Based Deflection Sensor**: - **Description**: A non-contact sensor (e.g., laser displacement sensor) measures gear bow deflection by detecting changes in position or distance. - **Pros**: High precision, no direct contact with the gear, and potentially easier temperature compensation. Suitable for both squat detection and weight/CG estimation. - **Cons**: Expensive, complex to integrate, and sensitive to environmental factors like dirt or vibration. May require significant modification to mount. - **Use Case**: Viable for a high-end system but likely overkill for a Cozy MK IV unless precision is paramount. 3. **Load Cells**: - **Description**: Install load cells at the gear attachment points or axles to directly measure weight. - **Pros**: Highly accurate for weight and CG calculations, with established use in aircraft weighing systems. Temperature compensation is standard in many load cells. - **Cons**: Requires significant structural modification to integrate into the gear system, adding weight and complexity. Not ideal for a simple squat switch. - **Use Case**: Best for precise weight/CG measurement but impractical for retrofit on a fiberglass gear bow. ### Analysis for Your Application - **Squat Switch**: For a simple weight-on-wheels detection, a **bend sensor** is likely the most elegant and cost-effective solution. Its simplicity and sufficient sensitivity to detect gear deflection make it ideal for a binary output. A microswitch could also work but may require more mechanical integration. - **Weight and CG Estimation**: If you want to extend the system to measure gross weight and CG, **strain gauges** are the better choice due to their higher precision and ability to provide quantitative data. By placing gauges on the main gear bow and nose gear strut, you can correlate strain to load and calculate CG based on the distribution of weight. However, this requires: - Temperature compensation (e.g., using a Wheatstone bridge with a dummy gauge on an unloaded fiberglass sample). - Calibration with known weights to establish a strain-to-weight relationship. - Signal conditioning (amplifier and analog-to-digital converter) for accurate readings. - Software to process data and compute CG based on gear geometry. - **Fiberglass Considerations**: Fiberglass gear bows are anisotropic and may exhibit creep or fatigue over time, which could affect long-term calibration. Strain gauges must be bonded carefully to ensure consistent strain transfer, and protective coatings (e.g., epoxy or silicone) are needed to shield them from moisture and abrasion. Bend sensors are less sensitive to these issues but may not withstand repeated high-deflection cycles. ### Recommendation - **For Squat Switch Only**: Use a **bend sensor** mounted on the main gear bow. It’s simple, lightweight, and sufficient for detecting deflection without the complexity of strain gauges. Ensure the sensor is rated for the expected deflection range and environmental conditions (e.g., FlexiForce or Spectra Symbol flex sensors). - **For Squat Switch + Weight/CG**: Use **strain gauges** on the main gear bow and nose gear strut. This setup offers the precision needed for weight and CG estimation. Implement a Wheatstone bridge with temperature compensation (e.g., a dummy gauge on unloaded fiberglass) and a microcontroller-based data acquisition system (e.g., Arduino or similar with a strain gauge amplifier). Calibration will be critical, and you’ll need to account for fiberglass’s thermal expansion (coefficient ~10-20 µm/m/°C) and potential creep. - **Hybrid Approach**: If budget and complexity allow, consider a bend sensor for the squat switch (simplicity) and strain gauges for weight/CG (accuracy). This separates the systems, reducing the squat switch’s reliance on complex electronics while enabling precise weight measurements. ### Practical Considerations - **Installation**: For strain gauges, follow aerospace-grade bonding procedures (e.g., using Vishay or Omega gauges with M-Bond adhesive). For bend sensors, ensure secure mounting without constraining gear flex. - **Electronics**: A simple squat switch needs minimal processing (e.g., a comparator circuit). Weight/CG measurement requires a data acquisition system with at least 12-bit resolution and temperature compensation. - **FAA Compliance**: Since the Cozy MK IV is experimental, you have flexibility, but ensure any modifications meet airworthiness standards (e.g., FAR Part 23 for guidance). Consult with an A&P mechanic or DER if integrating into a certified aircraft or if safety-critical systems (e.g., gear warning) rely on the sensor. - **Testing**: Bench-test the system with simulated loads and temperature cycles. For weight/CG, calibrate with known weights on the aircraft and verify CG calculations against manual measurements. ### Conclusion For a squat switch alone, a **bend sensor** is the simplest and most elegant solution due to its ease of use and adequate sensitivity. For combined squat switch and weight/CG measurement, **strain gauges** are superior despite their complexity, offering the precision needed for quantitative data. If budget and expertise allow, strain gauges with proper temperature compensation and calibration are the best path for a dual-purpose system. Avoid load cells or optical sensors unless you’re prepared for significant modifications. Let me know if you need help with specific sensor models, circuit design, or calibration procedures!

Return of the Gremlins

Return of the GremlinsI think the coffee spilling inside the lunch box cooler is what released them. They're probably offspring of the ones that ruined my Oshkosh trip a few years ago. That year the gremlins had been especially active — a flat tire, a plug wire coming loose, the Dynon screen shorting out internally and a string of other daily anomalies. It got bad enough that my wife convinced me to ground the plane and I drove to Oshkosh instead. That forced the little pests elsewhere for a while, but this spring's weather must have called them back.I'd broken the Condition Inspection into stages and flown several post-maintenance flights to verify everything and isolate maintenance-induced issues. Minor drag reductions included countersinking the cowl screws deeper so the heads sat flush. I was also keeping an eye on a prior oil cavitation event (momentary low oil pressure) linked to turbulence; we'd inspected the regulator spring and ball, installed a new crush gasket, checked the screen, changed the oil and filter, and verified the sender.  Oil sample taken.  The bottom cowl cooling ramps had been tweaked, steadily improving the CHT deltas to less than a 20 F spread.Camping gear went in Thursday night. Before dawn Friday, I packed clothes. Weather looked marginal but doable for an early departure. After breakfast and a final sky check, I texted LuAnn (at her brother's in Oklahoma) around 8 AM that I was heading out soon.As the hangar door rolled up, the sky had that "get going now" look. A student was in the pattern as I taxied. Broken clouds were already at pattern altitude, so instead of risking getting pinched between FTW Class D and DFW B, I departed Runway 14, turned crosswind, and climbed through a shallow canyon to get on top west of Bravo and head south then east. Takeoff felt great — little radio chatter while climbing to 11,500 feet. All parameters looked good, though oil temp was rising, so I enriched the mixture slightly and shallowed the climb.

Broken layers.  In the distance, another above 11K' south of DFW. Photo was taken NE of Denton looking south.

Clouds from the south were solidifying, with another broken layer above 9,500 feet so I opted to level there for now. It was clear to the north, but graphical METARs on the HDX and FlyQ showed airports under the Bravo flipping from VFR to IFR quickly. I could still see terrain below, and things looked more scattered east. It's fine… until it's not.

Then the panel lit up: "Oil Pressure!" — 90 PSI in the yellow, sometimes higher. Nearest VFR options were Ennis or Tyler farther east. I modulated throttle/RPM; the pressure responded but stayed erratic. All other temps and indications remained solid. I'd seen a similar "glitch" before that we'd traced to turbulence-induced cavitation, and recent post-maintenance checks (screen, regulator, filter, oil color still amber after 30+ hours) had looked clean. Still, with a long trip ahead and fresh in my mind how quickly things can go south after recent work, the mental checklist started: How many strikes? Abnormal indication… marginal weather… spilled coffee in the cooler…I found a stable spot below 2,400 RPM where pressure held around 80 PSI. Rusk and Tyler were still VFR. I was already monitoring ATC and nearby airport frequencies, and I reviewed divert options, transponder codes, and possible radio calls. Rather than press on, I turned toward Tyler at 9,500 feet as a precautionary measure. Clouds were lifting in spots, some METARs improving, and ADS-B traffic was increasing. It didn't look terrible below, but I could see why fields were toggling IFR as clouds passed over.At Tyler, I decided the Florida trip was off. I had planned to fly down for the Sport Air Racing League events on Saturday and Sunday, then continue on to Lakeland for Sun 'n Fun. Stopping at Tyler risked being stranded for several days as the weather system moved north. Things seemed to be trending better, so I diverted toward McKinney (where James Redmon is based), descending to 8,500 feet while trying to keep the RPM under 2,400 and CHTs near 300°F. I was still weighing it: sensor glitch? Alternator debris in the accessory case? New filter shedding media? Weather holding? McKinney sits on the edge of DFW Class B.  I didn't want to give up altitude so opted to skirt the Bravo on the North side.   Is that still two strikes? The airports aren't IFR here…

McKinney TX

 Denton, Gainesville, I-35, and Hwy 287 were all within easy reach. I descended further to 6,500 feet, planning lower near Denton to avoid arrivals into DFW, but not too close with high density student traffic. Then the phone lit up with texts from my wife: "You're not on FlightAware. Everything ok?"  The transponder showed it was receiving and I had ADS-B traffic on my screen.  I'd moved the antenna to the copilot side during the CI, so I power-cycled the transponder anyway. I was now literally just minutes from home.THUMP! Like a bird strike, not mechanical. Then another. Nothing else changed. Copeland, Propwash, Alliance passed below. Approaching at 2,000 feet, T67 had no traffic and higher clouds. I announced five-mile final for Runway 14 as a helo crossed in front of me going to Copeland.  I noted conflicting windsocks (not unusual due to the hangars, trees and RR tracks), got a bit slow, and touched down firmer than ideal — but safe.Back in the hangar, I texted LuAnn that I was down. Post-flight inspection told the story: A screw holding the temporary fence (lip) on the #3 cooling ramp had come loose, allowing the fence to flap around and cause the thump. This was from my recent tweaks to improve CHT deltas. I also found the cowl screws were not as tight as when installed (countersinking had disturbed the fiberglass surface where the washers seat, so it will need some cycles to restore its hard polish).EFIS data was uploaded to Savvy's site.  This was flight #531.  The #1 & #3 temp graphs indicated when the ramp extension had started coming loose.  Early in the flight, they were within 4 degrees and later went to a 20 F delta.The oil pressure sender seems to be the main culprit behind the erratic oil pressure reading (At least for now). Note that I had already replaced the original pre-2018 Kavlico sensors per Dynon’s Technical Service Bulletin (issued May 2020, updated August 2020) regarding potential issues with the older units. I swapped them for the newer Kavlico v2 sensors (P/N 103757-000 / 105492-002 style), which now have over 350 tach hours on them with no prior problems. The suspect sender was replaced with one of the known-good extra spares from another builder, and oil pressure is now rock-solid at 70 PSI with zero jitter. Finally, my avionics neighbor spotted a pushed-back BNC center pin on the transponder coax about 1/16", which we fixed by installing a new connector and moving the antenna back.I’d been distracted by neighbors during the earlier test run, and another gremlin had quietly arrived. After shutting the engine down and pushing the plane into the hangar, I turned the master and engine bus switches back on — and the relays didn’t pick up. My first thought was, “Did my EarthX battery just die?” The next morning everything was normal. Research suggests the BMS likely isolated due to multiple start/charge cycles until the cells cooled. Resting voltage measured 13.53 V, which is excellent for a three-year-old battery.The old sensor was bench-tested with an air pressure gauge, 5-volt power supply, and voltmeter. At 70 PSI, the green wire voltage fluctuated only a few microvolts (2.302–2.313 V), equivalent to just a few pounds — nowhere near the 100+ PSI swings seen in flight. I’ll still check grounds and connections closely. These piezoelectric sensors are notoriously difficult to diagnose with a simple ohm meter.No test flight was possible before the weather went IFR, so the trip to Florida — and the Sport Air Racing League events plus Sun ’n Fun  — is officially scrubbed. Time to sweep the gremlins out of the hangar. There must be a high concentration of them southeast of Dallas.
Update 4/17/2026 - I've added a redundant oil pressure sensor and added it to the Dynon EMS220 along with a widget labeled "OP2".  I've reinstalled the suspect sensor in this position and will monitor it against the new sensor and eventually test less expensive sensors.