Shawn26 Posted March 5 Author Report Posted March 5 1 hour ago, Thedude said: Any update on this? I received good news yesterday from the Avionics Shop. Last week, we removed the magneto and sent it to the mag shop for another inspection. They replaced the contact and bearing, both of which were already replaced during the 500-hour inspection. I reached out to Savvy Aviation for advice, and they confirmed that the only likely source of the erratic RPM readings (based on the P-lead signal) was the magneto creating an arc from the point, bearing and conatct. The mag shop recommended reinstalling the magneto, and if the issue persisted, they offered to build a new magneto from scratch to help rule out potential causes during the troubleshooting process. After reinstalling the magneto, the problem remained. The Avionics Shop, frustrated after chasing the issue for so long, decided to install a new Garmin GEA24B. The erratic RPM readings disappeared and we have a good reading. According to Garmin, the GEA24B has better noise filtration than my previous GEA24A. I haven't picked up the airplane yet, but I trust their diagnosis. This process was extremely time consuming, labor-intensive, and expensive. My main concern is what will happen during the next 500-hour inspection. If I receive a noisier magneto, I’m not sure if the GEA24B will be able to filter out the distortion, potentially putting us back to square one. It's something to consider for the future.
EricJ Posted March 5 Report Posted March 5 I just had my dual mag out for 500-hour inspection. Prior to removal I had noticed that I could hear the ignition on my #2 radio when I was monitoring 121.5. After reinstalling the mag after the 500-hour service the ignition noise is gone. I did not replace the condensers (yet, although I ordered two new ones, long story), so it really was just the mag service that cured it. If you get continued problems after getting a fair amount of time on the mag, maybe try a different mag shop. My understanding is that the dual mags are a little tricky to set up compared to individual mags. 1
PT20J Posted March 6 Report Posted March 6 I wonder if installing a filter between the mag switch and the GEA 24 would help. https://lonestaraviation.com/product/faa-pma-magneto-filter/?srsltid=AfmBOor0hZYMZBEybv-zcpTVcfLp-qrsMUAFTVDflHqFtFddXQJYZnmJ&v=28886f13f578 1
Shawn26 Posted March 6 Author Report Posted March 6 2 hours ago, EricJ said: I just had my dual mag out for 500-hour inspection. Prior to removal I had noticed that I could hear the ignition on my #2 radio when I was monitoring 121.5. After reinstalling the mag after the 500-hour service the ignition noise is gone. I did not replace the condensers (yet, although I ordered two new ones, long story), so it really was just the mag service that cured it. If you get continued problems after getting a fair amount of time on the mag, maybe try a different mag shop. My understanding is that the dual mags are a little tricky to set up compared to individual mags. This is my third time using Arrow Accessories for the magneto repair. I never had issues with their service. I know QAA also repair Bendix Dual mag. Beside these two I don’t know any other shop! Who did you use for your magneto?
Shawn26 Posted March 6 Author Report Posted March 6 (edited) 26 minutes ago, PT20J said: I wonder if installing a filter between the mag switch and the GEA 24 would help. https://lonestaraviation.com/product/faa-pma-magneto-filter/?srsltid=AfmBOor0hZYMZBEybv-zcpTVcfLp-qrsMUAFTVDflHqFtFddXQJYZnmJ&v=28886f13f578 That is something that I can ask my avionic shop. Edited March 6 by Shawn26
EricJ Posted March 6 Report Posted March 6 11 hours ago, Shawn26 said: This is my third time using Arrow Accessories for the magneto repair. I never had issues with their service. I know QAA also repair Bendix Dual mag. Beside these two I don’t know any other shop! Who did you use for your magneto? There are a lot of shops that will service dual mags these days. I keep hearing that Aero Accessories in Van Nuys is good: http://aeroacc-vny.com/ Mine was just service by Air Power Accessories in Camp Verde, AZ: http://www.airpoweraccessories.com/ 1
Minivation Posted September 16 Report Posted September 16 (edited) I realize that this thread is over half a year old at this point but I am following up in case anyone else runs into this issue in the future. After nearly a year, we have finally isolated the root cause and have managed to resolve it. The bottom line is, the RPM signal being sent into the GEA24 sensor adapter is a pulse signal, and the duty cycle of said pulse signal must be greater than 50% in order for the GEA24 to process it properly at higher frequencies (i.e. RPMs). In other words, for every high-low period in the signal, the signal must spend more time in the "high" state than in "low." Since my last comment on this thread, we have tried installing magneto filters, rerouting cables to avoid high-noise areas, swapped the GEA24, swapped the GI275, and even replaced the P-lead interface with the UMA T1A3-4 tach drive sensor (approved for GI275 EIS since July 2025) all to no avail. We tried all sorts of experiments, including rotating the tach sensor with a drill (to simulate the engine rotating), powering the EIS (GEA24 and GI275) with a regulated power supply, powering just the RPM sensor with the power supply, and even testing a Hall-effect RPM gauge from Amazon and hooking it up to the UMA sensor to verify it played nicely with the Amazon gauge (it did, beyond an equivalent of 5000rpm). Eventually, we removed the GI275 and GEA24 from the airplane and constructed a basic wiring harness in the workshop. We have an oscilloscope and signal generator, so we had the signal generator feed into the RPM input port of the GEA24 to emulate a sensor (with the o-scope monitoring it). After playing around with the waveforms a little, we found that the duty cycle of the RPM signal directly impacts the EIS's ability to properly resolve a valid RPM reading at higher RPMs. Duty cycle | Max RPM before fail | Failure mode 15% 1200 rpm Fluctuates, then drops to 0 30% 2400 rpm Fluctuates, then drops to 0 50% 4000 rpm Fluctuates, then drops to 0 70% 4000 rpm Red X 85% 4000 rpm Red X The above data implies that in any case, the EIS (as configured for a Lycoming 6-cyl direct drive engine) only goes to 4000rpm before failing the RPM, likely by design (a Lycoming spinning at 4000rpm is ... bad). However, we can clearly see that the RPM performance drops sharply at duty cycles below 50%. As a side note, this may explain Garmin's position of the magneto being "at fault" regarding the erratic RPM readings on select EIS installations. I haven't had the opportunity to explore the magneto P-lead output on an oscilloscope but I suppose that different condensors/capacitors in the mag may directly impact the waveform, and in turn, the "duty cycle" despite it not necessarily being a clean-cut two-state pulse signal. Going back to the issue at hand, the UMA tach sensor, by construction, has a fixed duty cycle of 1/3, or 33%. Looking at the inside of the sensor reveals that the unit is nothing more than a wheel with a permanent magnet embedded somewhere in it, and this wheel rotates in close proximity to a PCB with a Hall sensor on it. Unfortunately this doesn't help the cause, but what can help is if the signal could be "conditioned" to an adequate format before being passed onto the GEA24. In this case, "inverting" the signal such that at 0V (low), the output is 12V (high) and vice versa, would effectively increase the duty cycle from 33% to 67% with minimal impact to the geometry of the waveform. There are several ways one can accomplish that, but ideally the "signal conditioning" circuit needs to (1) be reliable for long-term duty, (2) have fast switching characteristics, (3) be simple to construct (KISS method), and (4) be resilient to environmental variables like temperature and humidity. One possible method from this discussion is using a transistor, such as an IRF530 MOSFET, to drive the gate (or base in a BJT) to cause the transistor to conduct or cutoff between the drain (collector) and source (emitter): In the above circuit, the output signal is held at the MOSFET drain-to-source voltage. When the input is pulled high, the MOSFET starts conducting and effectively diverts the power passing through resistor R3 straight to ground. Thus the output voltage sees almost 0V. But when the input drops to a low voltage state, the MOSFET stops conducting and the output is virtually held at near 12V (assuming the GEA24 itself has a high impedance, which it does, so that only a minimal amount of current flows through it). I'm sure the real electrical engineers will hate me for the simplified explanation, but that's really the heart of the circuit. No circuit is perfect and the primary concern with this design is the MOSFET's ability to react quickly enough to sudden changes in the input signal. However, the IRF530 exhibits sufficient switching performance and testing with signals up to 13kHz (for the UMA T1A3-4 tach sensor which produces 2 pulses per engine revolution, that's equivalent to 390,000rpm) demonstrated that the circuit was still able to maintain a duty cycle of greater than 50% even with the imperfect rise/fall characteristics of the MOSFET. (Testing the circuit at near 13kHz. Yellow = input, blue = output) (Testing the circuit at 200Hz - equivalent to 6000 engine RPM. Yellow = input, blue = output) In the image above, the GI275 can be seen indicating RPM well above the 2200-2300rpm failure threshold I mentioned in my previous posts. In fact, this is near full power with the manifold pressure running well above ambient. Granted, the troubleshooting and testing here was all done with the UMA T1A3-4 sensor, which obviously works a bit differently than the magneto P-lead interface, so if anyone else is having issues and would like to keep the P-lead interface while implementing a solution, then a thorough review of the behavior and characteristics of the P-lead signal would be warranted. Edited September 16 by Minivation 2 1
Thedude Posted September 16 Report Posted September 16 Thanks for the very detailed updated, but how did this ever work in the first place?
N201MKTurbo Posted September 16 Report Posted September 16 8 hours ago, Minivation said: I realize that this thread is over half a year old at this point but I am following up in case anyone else runs into this issue in the future. After nearly a year, we have finally isolated the root cause and have managed to resolve it. The bottom line is, the RPM signal being sent into the GEA24 sensor adapter is a pulse signal, and the duty cycle of said pulse signal must be greater than 50% in order for the GEA24 to process it properly at higher frequencies (i.e. RPMs). In other words, for every high-low period in the signal, the signal must spend more time in the "high" state than in "low." Since my last comment on this thread, we have tried installing magneto filters, rerouting cables to avoid high-noise areas, swapped the GEA24, swapped the GI275, and even replaced the P-lead interface with the UMA T1A3-4 tach drive sensor (approved for GI275 EIS since July 2025) all to no avail. We tried all sorts of experiments, including rotating the tach sensor with a drill (to simulate the engine rotating), powering the EIS (GEA24 and GI275) with a regulated power supply, powering just the RPM sensor with the power supply, and even testing a Hall-effect RPM gauge from Amazon and hooking it up to the UMA sensor to verify it played nicely with the Amazon gauge (it did, beyond an equivalent of 5000rpm). Eventually, we removed the GI275 and GEA24 from the airplane and constructed a basic wiring harness in the workshop. We have an oscilloscope and signal generator, so we had the signal generator feed into the RPM input port of the GEA24 to emulate a sensor (with the o-scope monitoring it). After playing around with the waveforms a little, we found that the duty cycle of the RPM signal directly impacts the EIS's ability to properly resolve a valid RPM reading at higher RPMs. Duty cycle | Max RPM before fail | Failure mode 15% 1200 rpm Fluctuates, then drops to 0 30% 2400 rpm Fluctuates, then drops to 0 50% 4000 rpm Fluctuates, then drops to 0 70% 4000 rpm Red X 85% 4000 rpm Red X The above data implies that in any case, the EIS (as configured for a Lycoming 6-cyl direct drive engine) only goes to 4000rpm before failing the RPM, likely by design (a Lycoming spinning at 4000rpm is ... bad). However, we can clearly see that the RPM performance drops sharply at duty cycles below 50%. As a side note, this may explain Garmin's position of the magneto being "at fault" regarding the erratic RPM readings on select EIS installations. I haven't had the opportunity to explore the magneto P-lead output on an oscilloscope but I suppose that different condensors/capacitors in the mag may directly impact the waveform, and in turn, the "duty cycle" despite it not necessarily being a clean-cut two-state pulse signal. Going back to the issue at hand, the UMA tach sensor, by construction, has a fixed duty cycle of 1/3, or 33%. Looking at the inside of the sensor reveals that the unit is nothing more than a wheel with a permanent magnet embedded somewhere in it, and this wheel rotates in close proximity to a PCB with a Hall sensor on it. Unfortunately this doesn't help the cause, but what can help is if the signal could be "conditioned" to an adequate format before being passed onto the GEA24. In this case, "inverting" the signal such that at 0V (low), the output is 12V (high) and vice versa, would effectively increase the duty cycle from 33% to 67% with minimal impact to the geometry of the waveform. There are several ways one can accomplish that, but ideally the "signal conditioning" circuit needs to (1) be reliable for long-term duty, (2) have fast switching characteristics, (3) be simple to construct (KISS method), and (4) be resilient to environmental variables like temperature and humidity. One possible method from this discussion is using a transistor, such as an IRF530 MOSFET, to drive the gate (or base in a BJT) to cause the transistor to conduct or cutoff between the drain (collector) and source (emitter): In the above circuit, the output signal is held at the MOSFET drain-to-source voltage. When the input is pulled high, the MOSFET starts conducting and effectively diverts the power passing through resistor R3 straight to ground. Thus the output voltage sees almost 0V. But when the input drops to a low voltage state, the MOSFET stops conducting and the output is virtually held at near 12V (assuming the GEA24 itself has a high impedance, which it does, so that only a minimal amount of current flows through it). I'm sure the real electrical engineers will hate me for the simplified explanation, but that's really the heart of the circuit. No circuit is perfect and the primary concern with this design is the MOSFET's ability to react quickly enough to sudden changes in the input signal. However, the IRF530 exhibits sufficient switching performance and testing with signals up to 13kHz (for the UMA T1A3-4 tach sensor which produces 2 pulses per engine revolution, that's equivalent to 390,000rpm) demonstrated that the circuit was still able to maintain a duty cycle of greater than 50% even with the imperfect rise/fall characteristics of the MOSFET. (Testing the circuit at near 13kHz. Yellow = input, blue = output) (Testing the circuit at 200Hz - equivalent to 6000 engine RPM. Yellow = input, blue = output) In the image above, the GI275 can be seen indicating RPM well above the 2200-2300rpm failure threshold I mentioned in my previous posts. In fact, this is near full power with the manifold pressure running well above ambient. Granted, the troubleshooting and testing here was all done with the UMA T1A3-4 sensor, which obviously works a bit differently than the magneto P-lead interface, so if anyone else is having issues and would like to keep the P-lead interface while implementing a solution, then a thorough review of the behavior and characteristics of the P-lead signal would be warranted. You should re-adjust the cam in the mag. If it is not properly adjusted, the dwell can be too short. Thus jacking up the pulse width. It is an easy adjustment to make and can be done without all the special tools called out in the service manual. You just have to buy a new screw. Spruce sells the screws for $5. 2
Minivation Posted September 17 Report Posted September 17 (edited) 13 hours ago, N201MKTurbo said: You should re-adjust the cam in the mag. If it is not properly adjusted, the dwell can be too short. Thus jacking up the pulse width. It is an easy adjustment to make and can be done without all the special tools called out in the service manual. You just have to buy a new screw. Spruce sells the screws for $5. You very well may be right. However, the concern here is that the EIS exhibited the same problem with the P-lead pickup interface or the UMA tach sending unit, which works completely independently from the mag. In my opinion, this points to a fundamental flaw in how the GEA24 (all variants) defines what counts as a pulse. Edited September 17 by Minivation
N201MKTurbo Posted September 17 Report Posted September 17 26 minutes ago, Minivation said: You very well may be right. However, the concern here is that the EIS exhibited the same problem with the P-lead pickup interface or the UMA tach sending unit, which works completely independently from the mag. In my opinion, this points to a fundamental flaw in how the GEA24 (all variants) defines what counts as a pulse. Perhaps, I hope you get it sorted out. 1
Recommended Posts