N891JF Panel

N891JF Panel

created January 2013, updated May 2015 with defective DPDT ignition and fuel switches(below).


The instrument panel looked like it had everything you'd need to fly, but there were issues that required it's replacement. The radio was an anemic two Watt home-assembled unit with poor noise EMI resistance and very limited range. The transponder was an old Terra with an odd half-width footprint, so if it ever died nothing else would fit in the hole. The Terra I bought new 15 years ago also died...so been there, done that. The instruments had lots of issues, such as the Tiny Tach indicating whatever it wanted, and fluctuating wildly in the process, CHT gauge inop, and oil pressure gauge that read 7 psi when the engine was turned off. But the biggest issue to me what the fact that all wiring went through a single contactor mounted to the firewall...which essentially functioned as a key operated battery disconnector. Not what you want when the single electronic ignition is also going down along with every thing else in the plane if that contactor croaks in flight! The wiring scheme had to be revamped before I was going to fly it again, so it made sense to remedy everything else while I was at it.


Because the plane has a center stick, which is way more of a pain than you would think until you've flown one, the critical engine controls, radio, transponder, and iEFIS, need to be available for left-hand usage. The center stick will be remedied shortly, but I can live with that a little longer. Besides, the fuel tank was only clearanced for left-hand radio and transponder lengths, so they have to go back in the same place. And the mixture, throttle, carb heat, and ram air needed to be located on the lower left side of the panel to get the most direct path to the firewall, given the large 15 gallon fuel tank that only left a small place for the cables to pass around the sides. Abrupt bends to cables increase resistance substantially. To answer a question you're already pondering, the "iBOX" is MGL's brains behind the panel. The Explorer is practically just a video screen. Many thanks to Steve Glover at nvAero for cutting me a great deal on this (he's an MGL dealer).

The MGL will be connected to a fuel totalizer to keep up with fuel, but I'll also retain the old sight gauge for a reality check. I found the current implementation very difficult to read, but the addition of a red float ball should aid in locating the level. [Later...the sight gauge went away...I'd have had to extend it down from the panel several inches, which became a fuel leak hazard if it was disturbed].


I plotted these out full-size and used 3M Super77 spray glue to stick them to the carbon fiber. Lacquer thinner will take them off later. The reverse plot on top is for laying out the wiring on the back side. It's best to think this out in advance. Now it's hole-cutting and drilling time.


Progress is being made! I cut all the holes out in about three hours. The rectangular hole on the left is for the trim position indicator. It went a lot faster and looked a lot better than I expected. If I'd known it would be that easy, I'd have used CarlingSwitch breakers like I have in N56ML. The iFly 720 screen looks like a mirror compared to the iEFIS, but it's so bright it really doesn't matter, and after sitting in the plane with it, it's no problem at all. I may still put an anti-reflection thing on top of it. Both iEFIS and iFly 720 are touch screens, and record flight data to SD cards for later processing. This takes the laptop out of the plane forever!


If you're wondering how I fastened my iFly 720 in place, it's mounted by this little shoe through the built-in keyhole slot in the back. On the right, you can see how it's made...just a spring, some bent-up aluminum, and the shoe, which is a McMaster Carr 3/16" weldnut. The hard part of this is the bent aluminum, which is a leftover bracket made at work. Yours doesn't have to be this elegant. It takes about three seconds to reach around back of the panel and loosen the locknut enough to slide the GPS off to remove it, and maybe 30 seconds to reinstall it. So I can still use the excellent "street" mode when driving my car.


I did a trial fit to make sure all the long stuff dodged the tank and longerons, and it fits nicely. The "carbon fiber" look is real, although it's just laid up on a sheet of aluminum. The funny thing about this panel is that it was a reject from the contruction of N56ML. I was looking for a piece of aluminum that might work for a panel, and this thing was already bent at the bottom (for stiffening) and had the carbon fiber layer on top. At the time I made it, I thought is was ugly, but I like the dull finish and there sure won't be any glare off of it.


The surface on the panel was very flat black looking because I'd peel-plied it and left the peel ply in place until curing, so the carbon fiber weave was a bit of a stretch to see. You can see from this image that some epoxy (or even water, polyurethane, or other liquid) really brings out the weave and gives it an almost-holigraphic look. I tried coating it with epoxy to bring out the weave, but all I did was make a mess of it all by beading up into large beads of epoxy (I'll spare you that photo).


I finally gave up, and sanded it back to CF with a palm sander, then laid up a new layer of carbon fiber on top of the panel, cutouts and all. This time I put a layer of peel-ply on it to soak up excess epoxy, then when it was somewhat saturated I removed it and put on another layer, and then I replaced it a third time. And FYI, I've already checked the image above for images of Jesus and Mary, but no miracles here. Maybe an hour after mixing the epoxy, I removed the third application of peel ply after it looked as if it wasn't absorbing any more epoxy, and the surface didn't shine anymore. I then cooked it under infrared lamps for the night and the next day I cut the CF out of the holes. The result was exactly what I wanted, which was a clear carbon fiber look, but without the shine. I take a lot of pictures out of the plane, and am constantly fighting glare off of various things. The panel will not be one of them! This is another reason the white panel had to go.


As you can see above, we have the carbon fiber look, but no shine. This is exactly the look I was after, and I'd have saved two hours of work if I'd just gone ahead and laid it up. There's only about five dollars worth of CF in this, and pennies of epoxy. The CF was left over from my wings, many years ago. As you can see, I'm making progress, with only the trim position indicator and mixture, throttle, and carb heat control cables remaining.


Now "all" I need to do it spend several weeks wiring up all the stuff behind the panel. At least this time I can incorporate all the tweaks that I incorporated into N56ML in the beginning, rather than cobbling it together over time, so it'll be neater and more compact. And after that, I have to do the same in the engine compartment. The "iBOX" on the left is the brains of the iEFIS system, but there's also an "RDAC" unit out on the firewall that collects all the engine sensor input and channels it back to the iEFIS via a simple two-wire CAN bus. This is a good thing, because it keeps firewall penetrations to a very few wires, rather a huge bundle, and the connections are very easy to make since they are all made at the firewall. That means no standing-on-my-head to make connections...just standing upright at the firewall. Sounds good to me!


May 2014 - Here's the final panel, with everything turned on and working. From left to right, intercom, ICOM radio, trim indicator, Becker transponder. Next the big "glass" is the MGL Explorer, with screen split between HSI and navigation map. This is a touchscreen device, along with knobs and buttons. I like the way it works, and it was the first touchscreen out there that I know of. The screen setup is easy to comprehend at a glance. Next is the iFly 720 GPS, also a touchscreen device. Essentially I have two GPS's now, which sounds a little crazy but I can keep ;the Explorer on the "navigation big picture" while using the iFly for the details such as airspace, AWOS stations ahead, and the destination airport.

To the right of the iFly is the air/fuel ratio (mixture) meter up top, reading a $15 Bosch oxygen sensor in the exhaust. The compass is obvious, and below are the simple switches. Although I prefer rocker breaker/switches like I have in N56ML, these leave a hint of the original flavour of the plane, and still get the job done. The big red button is the starter, and the switch to the right with the green LED above it is the fuel and ignition swapper. The LED is bicolor, so it's green in the normal position and red when in the backup mode. The backup mode uses a completely different electrical system consisting of a 7Ahr backup battery located behind the aft spar, and separate wiring to the ignition and secondary fuel pump. It is independent of any other switches in the plane, so all wiring can fry, and this system (which runs on the other side of the plane) will keep soldiering on. A 30A Schotkey diode keeps the backup battery charged during normal operation, but fries in the case of a completes short on the main side, thereby isolating the two electrical systems.

At the bottom right of the panel you can see the terminal strips below the panel, which provides access for all major circuits, in case I want to add something later. Since the front deck is not removable, this allows simple changes without having to do gymnastics under the panel.

The three LEDs above the Explorer are green for "belly flap down", yellow for iEFIS Explorer warning (which monitors all engine functions as well), and red for oil pressure below 8 psi.


UPATE - May2015: After I had about 60 hours on the plane, I was starting the plane for the trip to the Chino Gathering, and it refused to start. I killed the battery before it would start, and that's saying something with an Odyssey 680 AGM battery. Joe Horton and I did about two hours of troubleshooting and concluded it had to be the CompuFire ignition coils, so I was no longer flying with him to Chino for a very cool Gathering. In the troubleshooting process, the DPDT main/backup ignition switch failed due to repeated cycling, which confounded me a week later after I'd installed the new CompuFire coilpack...but I could hear the contact rattling around inside.

So I swapped out the DPDT switch with a spare and it ran fine...until I was flying back from Sun n Fun, picking my way around some light rain south of the pan-handle, and the engine went instantly dead. But I'd noticed the bright green main ignition LED disappear at the same time and immediately flipped the DPDT switch below it to the backup position, and the engine roared back to life before the prop even got stopped. This was a little unnerving, given that all I could see was swamp in all directions! After that, I was in no mood to even touch that switch again until I was ready to refuel at Tuskegee. On the ground, I confirmed that there was no main ignition circuit, but the backup worked fine, so I flew home at high altitude, as I'd been doing ever since I got out from under the cloud deck.

Only 4.9 amps passed through this switch at any given time (4.1 for the ignition, and 0.78A for the fuel pump), but switching a load is very hard on the contacts, and I switched them one time during runup during each flight as part of the runup procedure...my "mag check".


Here are the two failed switches: the one on the left is the most recent failure. Note the discolored and bent contact "spring" (1), and also the hole burned completely through the brass spring (2). It matches a nub on the silver contact, where it was welded in place (3). The crazy thing is that it was welded to the contact in a very high resistance way...enough that the contacts couldn't flow 5A anymore, and it did this "steady state"...after flying for half an hour. Let me tell you, there's nothing like the engine going silent over a swamp.

The upper contact point completely missed the lower contact...the angle was so severe that the brass spring contacted the lower point, and acted as the upper contact. It's not thick enough for that, so a hole was eventually formed as brass was transferred to the lower contact due to arcing.

The switch on the right was the first to fail...with the contact spring breaking off completely! Notice the angle of contact between the upper and lower contacts...nowhere near parallel (4). You'd think the designer would want to maximize the surface area of contact for max electrical flow, but instead, it's concentrated on one small point (5). All of the contacts are at this angle when closed. This means lots of power through a small point, kind of like "spot-welding". This contact simply broke in half at (6), apparently after being weakened by high temperatures. The lower points are installed at an angle that's adverse to properly alignment as well.


I expected to find fried contacts in these switches due to switching 4.9 amps, and although the contact areas are far less than optimal, it was that thin brass strip that's the weakest link, simply overheating, weakening, and eventually fatiguing to the breaking point. The brass "actuator" strips measure .004" x .075", which is a total surface area of .0004 square inches...virtually nothing. I realize this is a very short distance, so total resistance is reduced, but I have a hard time believing that this cross-sectional area is sufficient to flow 16 amps of current at 28VDC. 16 gauge wire has an area of about .002 square inches, 20x greater, and copper is twice as conductive as brass (40x). But again...it's a short distance. I'd bet that the Honeywell switch has thicker strips, but I'm not willing to sacrifice $40 to cut it open and find out. I guess I'll have to wait for it to fail to check that.

Anybody building an aircraft electrical system should know that "AC" current values are fairly meaningless for DC applications, and the Otto switch is labeled 10.1A @ 125VAC. But before I bought these, I checked the data sheet for this particular switch, and it's listed as 16A at 28VDC. That should be plenty conservative for less than 5A that I was pulling with the ignition and fuel pump. You could argue that the load is inductive, given that the ignition is coils, but it's not the contacts that are failing due to arcing, it's the thin (and misaligned) brass connection between the contacts that was the problem with these two switches.


First order of business was to replace that DPDT switch, and given that this was the second failure in 40 hours, I wasn't going to put another one like this back in the hole! So I opted for a Honeywell 2TL1-3D instead, which is rated for 28Vdc @ 20 amps, rather than the Otto's 16A @28VDC rating (per their spec sheet). This is a $40 switch from Newark (part number 23F1980), as opposed to $17 for the Otto from AS&S. Just to give you an idea of the quality of the Honeywell, it's about 5 times more difficult to "switch" than the Otto. That means the contacts are moving a lot quicker and more reliably, minimizing arcing. The Honeywell is rated at 20A @ 28Vdc, and I believe it! The Otto was supposed to be good for 25,000 cycles, but I can tell you it was well under a hundred for the first one, and well under two hundred for the second one! I should caveat this with the fact that I ran one of these very same switches for 1130 hours in N56ML without a single problem, although that system was probably drawing about 4A rather than 5A, due to using the lower current single Bosch coil.

I'm so appalled at the construction and reliability of these switches that I'm going to call Otto and show them these pictures. It occurred to me that perhaps these are cheap counterfeit switches. I hope there's a logical explanation. Another possibility is that the design or place of manufacture has changed. Whatever the cause, I have a feeling when their lawyer finds out these switches are being sold and used for aircraft applications, they will no longer appear in the aviation supply catalogs, suitable or not.

Although arcing was not an issue in these failures (they didn't last long enough to become an issue), I should mention that arcing that destroys points is easily fixed by a $30 solution, called an "arc suppressor". See Arc Suppression Technology's website. For a DPDT (double pole, double throw) switch, you'd need two of them, one across each related pair of contacts. I should thank my electrical engineer buddy at work for pointing me towards these, as well as many years of other good advice and continuing education.


iFly UPATE - October 2016: I've put almost 300 hours on the panel now, and I really like the iFly 720 user interface. It's WAY more intuitive and user friendly than the iEFIS (don't get me started on that) and is also more intuitive to use than Foreflight. But after about 100 hours of flying, my iFly 720 "bricked" itself during a routine OS update, requiring me to overnight it to AdventurePilot for an emergency repair (this happened four days before a planned flight to Chino, CA).

Two years later, at about 200 hours of use, the touch screen went south, to the extent that the unit was unusable, requiring another repair, which cost me about $140. This is not an uncommon issue either, as I know several folks who've had the same problem. When I got it back, my non-glare screen protector was missing (it went with the touch screen) and the screen was so highly reflective I could barely see it. So then I had to order another non-glare sheet and cut it to fit, since iFly doesn't even sell one to fit their unit! Yes, whine, whine, whine!

I've recently built a dual frequency Stratux ADS-b unit. It works great with my iPhone and iPad, but NOT with the iFly 720. Hmmmm. I've spent at least ten hours screwing around with this issue, including reloading the OS, but to no avail. The symptom is that although the Stratux is "connected" to the unit in the WIFI NETWORK SETUP screen and shows the Stratux's IP address (correctly found through DHCP, so the iFly unit is certainly communicating with the Stratux), there is no ADSB connection once running the NAVIGATION app. In NAVIGATION, when I go to INFO and CONNECTED DEVICES, the Stratux/Flightbox shows as "chosen", but the note below it says "Selected device not connected". So part of the unit thinks it's connected, and the other part thinks not.

The bottom line is that my ADSB doesn't work on this unit, all the while working fine on both iPAD and iPhone, which are happily running the iFly app and receiving and displaying ADSB data the whole time! The ADSB tower is only a few miles from both my airport and my house, so I don't have to fly to receive ADSB data...or test the iFly, but I have flown with both iPad and iFly, with the same results as I get on the ground (obviously), usually tracking 14 satellites and receiving three local ADSB towers.

I've opened a problem request with AdventurePilot, and they've had me reload the OS again, to no avail, and try a few other things, but it still shows "selected device is not connected". Given that I've never used the WiFi before (it's famously slooooow), it may not have ever worked. For $39.99, they will "evaluate and test" it for me, if I'll send both iFly720 and Stratux to them, and then will tell me how much it will cost to fix it! That was the straw that broke the camel's back...I've had enough frustration with iFly hardware, and would rather not pour more money into that little black box.

The iFly software works great on the iPad. The iPad is actually a little brighter than the iFly, and it's absolutely bomb-proof hardware by comparison. I'm using an old castoff iPad 2 that I inherited from my daughter, and despite many years of constant use and abuse, the touch screen still works great (way smoother and more predictable than the iFly), it has a huge battery with many hours of runtime, and I can run all kinds of other apps on it at the same time! If the iFly loses power for more than a few seconds, it's a 3 minute penalty before you get it back online! Also, the iFly brightness drops off quickly when viewed from the left in portrait mode, which is exactly how mine is mounted...and portrait mode only works one way, not the other...so there's no cure for that. The iPad remains bright at far greater viewing angles than the iFly. And the iPad boots in just a few seconds, as opposed to the 3 minute boot time of the iFly.

BUT, the iPad is way too large to fit in the small space left by the iFly, so I'd need an iFly Mini, which can be bought used from ebay for $125 or so. But if I have to buy something, it should be optimized for the job, so a search for "brightest tablet" turned up the Nexus 7. It's even brighter than the iFly or iPad, almost exactly the same size as the iFly 720, it's a lot faster (with a quad-core processor and 32G of RAM), and the Nexus 7 is more flexible than an iPad (at least it's not "locked"), because it runs the Android OS and applications. It's only .35" thick, so it would even look better on the panel.

So I bought a used 32G Nexus 7 for $150 from ebay, and was looking forward to using reliable hardware for my iFly software. Yes, 32GB of RAM is overkill, considering the iFly 720 uses an 8G SD card that's only half full, but it was a good deal! So I got it, installed iFly software on it, and despite what their website says, the iFly app doesn't see the ADS-b on the Android-based Nexus either, despite a clear indication that it's linked through wi-fi! Again though, it works fine on iPad and iPhone, so clearly I know how to set up the software. If there's some obscure port setting or something else I'm supposed to do (although I've tried everything available, I think), please let me know. I'll appreciate the tip, and I'll be even more irritated with Adventure Pilot for not making their instructions (or the tech support) more informative.

So then I bought an iPad 2 (again...$154 on ebay) because it fits the panel nicely in the footprint of the iFly 720. Guess what...even though it's connected through wi-fi, neither Foreflight nor iFly software will "connect" to the Stratux from within the app. Go figure. Still working on this, obviously, but at the very least, I can run either Foreflight or iFly software on a very inexpensive box with a huge battery, great readability, bulletproof touchscreen and buttons, fast processor, etc.

One problem iPads have is overheating when running full brightness continuously, especially in hot weather or direct sunlight. Mine never had that problem until I flew to OSH this summer on a record-breaking hot day, and then only when I landed for refueling and the iPad was in the sun. When I got ready for takeoff it said it needed to cool off first..."turn me off".

The cure for that is to direct some air on the back of the unit. I saw Bill Clapp at OSH with an iPad on the panel, and he showed me the sheet aluminum holder he'd bent up to retain the iPad, with holes drilled in the back of the bracket to allow air to circulate on the back. There was a 1" hole in the panel with a SCAT blast tube aimed at it...end of overheating problem! I'll have a huge hole in my panel when the MGL iEFIS comes out, so I can simply mount a large diameter fan to the back of the panel blowing on the back of the iPad. Also, make sure you power it with at least 2.1A of power, more if possible. I bought a 3A power supply for about $5 off ebay that will provide plenty of power.

And if the Mini croaks, to replace the it with something newer would cost peanuts compared to another iFly GPS Unit. I don't know why AdventurePilot even bothers with hardware....they should be content with the $79 annual iFly720 subscription price, and concentrate on their software, rather than dabbling with hardware that's no competition for the much larger tablet market.


Here's an old iPad 2 running Foreflight compared to the iFly720 with identical conditions (direct sunlight from behind, with exactly the same anti-glare material applied to both screens). The iPad is the clear winner here.

Update on the iEFIS- December 2016 - After flying behind the iEFIS for two years, all I can say is that it's been an exercise in frustration. It works (when something's not dead), but it doesn't work well, the user interface is counter-intuitive, and the software interfaces (loading, updating, data recording) are confusing and user-hateful. I could write a book on this, but refuse to waste the time, so here'a a quick synopsis:

  • Hardware:
    The iBox (brains of the system) had an issue at installation where the OAT sensor wouldn't work. This underlines the problem with an EFIS...if some little part of it doesn't work, you have to buy a WHOLE 'NUTHER BOX to fix it. It was under warranty, so MGL sent me a new one.

  • After putting 40 hours on the plane, two of the buttons on the main "Explorer" video screen that operate the majority of the system quit working. "First time that's happened", I'm told. So they sent me another Explorer unit. If the first one lasted 40 hours, how long will the next one last?

  • Then, the "new" iBox croaked (I forget what it was this time, but it was major), and they sent me another one.

  • Early in 2016, just after the three year warranty expired, I flew down to my father's farm. When got back in the plane to return a few hours later, the iEFIS was totally black...no sign of activity, or even life. Well, I still had the iFly on the panel, but I had to fly 250 miles with no engine information at all! Not smart, admittedly. So I called MGL and they agreed to swap it out, but when I went to swap it out, it fired right up and worked fine! I know what you're thinking...this thing had power from both main and backup batteries, and I tried them both, both on the day it happened, and the next day, and it was dead by all appearances. But before I removed it, I thought "what the heck", and flipped it on...it lit right up, and the Hobbs meter had recorded the flight back home! Go figure, but I call this "unreliable", and intermittent.

  • This data displayed on this thing is simply all over the place....temperatures, pressures, and things like the vertical speed jitter over an amazing range. I've sat in the hangar with nothing else on, running on an isolated backup battery, hundreds of yards from any 120V power sources, and watched the vertical speed hit over 1100 fpm...while sitting in the hangar! EGTs dance all around, sampling several times per second and hot and cold cylinders swapping places by as much as 30 degrees per second. Hmmmm... my old GRT EIS would fly along for thirty seconds without CHTs for all six cylinders changing by a single degree, or charging voltage might change by .1 volt, recording the same temps very steadily, and able to hold altitude within 10' for half a minute at a time. Just to show you how bad this iEFIS is, there's a "flight profile" page where it shows a few statistics from the previous flight, and I've seen stuff like "max fpm climb" as 21,000'/min, or something similar. I'm no longer amazed at any numbers I see from this thing. I'll post snapshots of this stuff if I ever feel like I have that kind of time to burn. I have plenty of video showing the crazy antics of the display while flying. MGL quickly pointed the finger at my alternator, so I flew with it disconnected...no change. So then the ignition system was the culprit, so I did a test where I went to 12,000' and killed the engine...there was a slight improvement, but that may have been my imagination. Then it was sensor wiring and grounding. What you should know is that I wired it with very high quality left-over Mil Std space shuttle wire...twisted pairs shielded in nickel plated copper braid, and grounded on a common end to eliminate any chance of interference, so this didn't hold water with me, although I checked and rechecked it all. Then one day I flipped on the iEFIS while I was listening to the FM radio in my hangar, and the radio instantly turned to "white noise". The iEFIS ITSELF was the source of the EMI! My iEFIS is wired on its own separate bus, and it also has its own backup battery. When I'm in the hangar I flip the backup EFIS power button and there's nothing else running but the iFIS. The temps and VSI, etc still jitter, but not quite as bad as when flying...I suspect because there's no motion or engine activity. Anything monitored like temps, VSI, etc have magnified variations when flying. It's worth noting that the rock-steady EIS (old-school engine information system from GRT that I ran on N56ML) had but one simple little ground wire to the main ground point and delivered solid numbers, but there's no end to the admonitions in the iEFIS manual regarding grounding, including conflicting information as to how to ground sensors and unit to engine block and RDAC. Eventually I hooked an oscilloscope to the bus that powers the iEFIS, in my hangar, nothing but backup power, no powered devices in the hanger (but the o-scope) and registered all kinds of AC spikes routinely around 2V, and as high as 10V on the power line between iEFIS and sealed lead acid battery. I'm just a dumb mechanical engineer, but you tell me, do you think it's the DC lead acid battery creating these random AC spikes or the iEFIS (that blanks out my FM radio)? I know which one it is! There was only one piece of MGL hardware that hadn't been replaced, so I bought a new RDAC-XT...no change. More later on this...I need a drink...but other topics will be:

  • Data logging (or the lack thereof) and crazy file naming convention (where all data files have the same name), same time stamp and file size for all data files, date format in a form that is difficult to search (given that there's no time stamp" etc.
  • Fuel level calibration is very coarse and hoaky, based on liters and very large fuel tanks (which we don't have), leading to interpolation insanity, and in the end a fuel tank level that means very little, especially if the tank isn't a uniform cross section from top to bottom.
  • Temperature resolutions of 1.8 degrees F (due to the core system being designed around Celsius), so it has half the resolution of most EFIS units(skips every other degree F). Not that it matters, as much as the numbers fluctuate.
  • Annoying little things like a software glitch that makes the machine lockup (forcing a reboot), when a common button (navigate to) is pushed.
  • Steam gauges, and the beauty of troubleshooting and replacing one instrument at a time, better readability, etc. Beats disabling the airplane while you send your expensive unit back to the manufacturer so he can screw you further for their lack of reliability.
  • Oh, and I have an iEFIS for sale....bargain price....


    Return to Mark Langford's N891JF Refurb Project