"Bump" Starting

When I purchased the airplane the starter was extremely weak.  When engaging the starter I would typically get one blade before the starter would stop.  It would take 4-5 "bump" starter engagements to get the starter to turn the engine over beyond one blade, but eventually the engine would crank over.  Once the engine got about 1 full turn it would start right up - fortunately it doesn't have mag or fuel issues, otherwise it would never start!

I tried looking at the usual suspects of corroded battery and starter cable terminal connections, but a thorough cleaning resulted in no improvement.  Eventually I replaced the starter and the battery, both of which were worn out.  This results in one-engagement starting about half of the time, but never more than two or three "bump" engagement starting.  While not ideal this was a vast improvement over the previous arrangement.

When the engine was hot it was even more difficult to turn over.  In fact it was so difficult to get the starter to turn the engine over that I took to avoiding quick fuel stops or unloading, which was beginning to intrude on my flight planning.  In late 2005 I had flown cross-country to a destination, filled up with gas, spent about an hour on the ground chatting with friends, and was getting ready to leave.  The engine was warm and heat soaked, but not freshly hot.  The starter refused to turn the engine freely, so I was "bumping" away in an attempt to get the engine to hit.  I started to smell burnt rubber, and a wisp of smoke appeared in the cabin.  Ack!  It was time to fix the problem for real.

Rather than try again with the "replace blindly" approach to fixing, I approached the problem more scientifically.

Starter Circuit Physics

In an unmodified Cruisemaster, the battery is located in the rear tailcone of the aircraft, just in front of the horizontal stabilizer.  This location is great for weight and balance to offset the heavy engine, but it means that the main battery "hot" lead is approximately 14 feet long from battery master relay to the firewall.  This is in addition to roughly another 1.5 feet worth of jumpers from the battery to the master relay, firewall to starter relay, and starter relay to starter.

For low-voltage high-current circuits like the starter, this is a real issue because the resistance of the wire (and the ground return path) becomes very significant.  Modeling the wire as a resistor, the equation is V = IR to calculate the voltage drop incurred through wire resistance.

Bellanca also tried to be clever and used a 3/8" aluminum round rod, sleeved in vinyl, as the battery conductor.  Presumably they were trying to save weight.  Unfortunately given that there's a need for weight in the back of the plane, and that there's a lead brick attached to the side of the battery box already, this was a penny-wise pound-foolish decision given the troubles aluminum causes for starting (see more below).

• Resistance of 3/8" Aluminum per foot (assuming an alloy and hardness of 6061-0):  0.000158 Ohms/foot
• Length of run:  14.17 feet
• Voltage drop at 200A:  0.448 V

This may not seem like much, but you have to add up the drops from each component of the system.  When diagnosing your starter, measure the voltage drop across each component and wire:  battery (has internal resistance), battery master relay, each jumper, firewall through-terminal, starter relay, and ground return path.

Typically you're going to see about 10V from the battery while cranking, and up to 0.2V of drop across a relay.  The small resistances of connections each add up.

The bottom line was that I was getting only about 7.0V at the starter.  I found that one jumper from the starter relay to starter was no good, and my ground path in the steel tubular airframe is also poor.  Measuring the airframe resistance is difficult, but you can tell something is going on just by observing which tubes and/or wires get warm while cranking.  High resistance will generate a temperature rise noticeable to the touch.

The real killer with Aluminum as a conductor is at the terminals.  Pure aluminum's resistivity is actually pretty good, after silver, copper and gold in that order.  Alloyed aluminum is a higher resistivity by roughly 1.3x but still not bad.  But compare the resistivity of aluminum with that of aluminum oxide:

• Pure Al:  2.82 x 10E-8 Ohm * Meter
• Al Oxide (14 degrees C):  1 x 10E14 Ohm * Meter

Notice the 22 order of magnitude difference.  It doesn't take much aluminum oxide on the surface of an aluminum conductor to turn it into a toaster.  Aluminum oxide also forms readily when aluminum is exposed to air and more so in the presence of moisture and high humidity, so if your connections are not 100% gas-tight you're vulnerable.  This is what was happening on the firewall terminal of the battery conductor, and it was smoking the rubber bootie covering the terminal.

The corrective actions that I took on my plane are:

First:  Remove and replace the aluminum rod battery conductor with copper wire.  This is an involved process that requires many hours of work and tailcone diving if fabric is on the airframe.  The rod in my plane was covered with a semi-transparent vinyl coating that had aged to a medium brown color.  The rod was attached to the airframe using lacing cord.  Removal required cutting the lacing cord along the length of the rod, then using bolt cutters to cut the rod in several locations (below the front seat, just in front of the rear seat, and below the baggage compartment).  Be careful with the bolt cutters since at least in the cockpit area other wiring bundles are laced to the rod for support - do not cut the wire bundles!  Gaining access requires removing the right hand cockpit kick panel, copilot's seat, rear floorboard, rear seat, and baggage compartment.

Removal of the rod and re-installation of a copper wire is a process that is much easier if the fabric is off the airframe.  The factory installation clearly happened prior to fabric installation.  If you are recovering a Bellanca fuselage I strongly recommend pre-emptively converting the battery conductor to copper wire to head off potential future problems.

I used "0" gauge stranded copper aircraft wire (MIL-W-22759/16 Tefzel insulation, this stuff is great but very expensive).  It is arguable whether "0" is necessary vs. "2" gauge, but since I didn't want to end up doing this again I decided to eat the extra weight penalty.  Do not go smaller than "2" gauge copper for a run of this length.

Second:  Replace a high-resistance jumper between the starter relay and the starter.  Freshly crimped terminals made a big difference, as this jumper was heating up in the old system.

Third:  Replaced the slightly corroded factory-installed firewall connection, which was an AN steel bolt, with a custom bolt made out of 3/8" copper rod and threaded using a coarse die.  You can get the copper from Online Metals if you don't have a local supplier.  I also ordered some bar stock and made copper nuts, one of which was soldered onto the end of the bolt to provide some turning leverage.

Fourth:  Moved the ground lug attachment point on the airframe to avoid a high-resistance tube.  You just have to experiment to find a better spot.

Starter Choices

When I purchased the aircraft it had the original equipment Delco starter, which is an automotive style starter with the field energized by the battery.  These starters are big and heavy, and are mid-range in terms of current draw, say 150A.

To address the airplanes starting issues, initially I took the dumb approach and decided to replace the starter.  I purchased a new Lamar permanent magnet starter.  This is a nice little starter, but I believe the permanent magnet starters draw even more current that the big Delcos and Prestolites.  As a result I think this starter made the situation worse.  For a "normal" installation with short battery leads I do recommend the Lamar unit, though.  Lamar had great customer service and it is a high quality unit.

After I replaced the battery cable and jumpers I tried cold starts.  The performance with the Lamar starter was much improved, but was still occasionally stalling up after 2-3 blades.  Given that I had addressed the R variable in the equation, and the V variable isn't really changeable, I decided to look at the I variable and see if there were lower-current starters available.

After some searching I came up with the Sky-Tec C12ST3 starter.  These starters are geared which allows for higher torque with less current draw as compared to a direct drive starter motor.  The downside is that they sound funny when running, but hey if it cranks well then I don't mind.  Empirical comparison on my engine shows that the C12ST3 is in fact cold-cranking reliably with no stalls and it is turning the engine over slightly faster.

The Sky-Tec C12ST3 unit is slightly longer but smaller in diameter than the Lamar.  It also is the lightest weight unit, only 6.5 lbs (10 lbs for the Lamar).  There are rumors floating around that the Sky-Tec unit destroys Continental starter adapters by not releasing tension when the starter stops.  I consulted several starter adapter overhaulers and could only find one that said this was true (the folks propagating the story against Sky-Tec starters).  Further Sky-Tec says that they've specifically added a clutch to the C12ST3 to counter any concerns about starter releases.  As a result I decided to go with the Sky-Tec, and time will tell who was right.

The only downside that I found to the Sky-Tec was that it turned the engine over fast enough that it exposed a slipping engine starter adapter.  The old starters never turned the engine over fast enough for long enough to notice this issue.  After verifying that it was indeed the engine starter adapter slipping and not the Sky-Tec starter clutch, I had no choice but to change the adapter for an overhauled unit.  I recommend Aircraft Specialties Services in Tulsa for overhauled adapters - great customer service.

Changing the starter adapter on the Bellanca is a major job since there is no room behind the engine to swap out the unit.  Consequently I had to remove the engine mount from the engine and float the engine assembly out a foot or so to get the clearance I needed.  This required many hours of hose, cable and wiring removal and reassembly, all for a 15 minute parts change.  Sure enough, upon overhaul my starter adapter's drum was revealed to be more than .030" worn, which is beyond the factory limit.  This is not a surprise considering the abuse from all the "bump" starts that my old adapter had inflicted upon it.

The Results?

Stay tuned, I'm still reassembling everything from the starter adapter change as of 3/10/06.  Preliminary tests with a cold engine look great.

The changes added 2.8 lbs for the copper cable relative to the aluminum rod, but minus 3.5 lbs for the starter, so a net weight loss of 0.7 lbs.  The CG moved slightly rearward which is beneficial for my airplane configuration.