Wednesday, September 26, 2018

Sunken Float

During my last Annual Inspection I overhauled the Bing carbs as they had exceeded their 200 hour mark. I did this despite the feeling that leaving things that work alone is usually the best strategy.
My main worry was that after all the issues with the Bing floats that Rotax had for a long time, I should probably go and finally see how they were doing after such a long time submerged in gasoline. And indeed, I did find two that were a bit heavy. Mind you though that despite their weight I experienced no issues with the engine. Not at idle or low RPM and of course not at WOT.

A few weeks ago I started my engine in the morning and found it to be reluctant to come to life. It sounded a bit like a Diesel engine, with heavy vibrations and very rough running coming out of idle, barely maintaining idle speed. My first guess was that one of the cylinders was not firing properly.
Pushing the idle during warm-up close to the 2500 RPM mark made it better, and the engine ran fine at higher RPMs so I decided to go ahead and take off.
At my destination I throttled back to start the decent and the engine behaved a bit rough again and I could smell gasoline. Yeah, that definitely worried me, so I grounded the plane after my return to give the engine a thorough inspection before further flight. As a busy man I didn't get to it until last weekend where I took off the carbs opened the bowls and re-inspected the floats.
One of the just recently replaced floats (one of those that were supposed to finally end the problem of sunken floats) weighed in at 8.5 grams!

Notice the markings on the float in the picture above. This is definitely a new style float and it took just a few months to almost triple its original weight. Given the price of these parts this should never ever happen. The other new style float I found was at 4.5 grams and two original floats I left in the carbs were at or below 3.5 grams.
I replaced the 8.5 gr float with a new one and matched the 3 others for lowest pair weights and installed them in the carbs.
The engine fired up and ran just fine, no coughing, no fuel smell, no rough running at low RPM or idle.

I am getting tired of these Bing carbs and am thinking again about replacing them with a TBI...

-- Update --

I researched this matter a bit more as I was very annoyed by the fact that these "latest" floats were still showing the old problem of absorbing fuel. Turns out, I must have bought the carb rebuild kit right before a newer new float was released. I have the 861-185 version which was obsoleted around September/October last year which is right after I had bought my rebuild kit. The latest product number is 861-188 and the leisurely determined price is a whooping $151 for a pair. Which is about 15% of a TBI ....
I found details about the float tale here:

The following picture is a nice example how you can identify a float issue without even utilizing a scale.

When the brass pin that is supposed to push against the lever that controls the inlet valve on the carburetor is below the fuel line in the bowl, it cannot perform its function properly. The left one is clearly below that line!

Tuesday, January 2, 2018

Dynamic Propeller Balancing

Ever since I had my prop balanced in 2014 somewhere up in Phoenix I had the feeling that the guy who did it had not done a good job. He only had done it on direct drive Lycoming and Continental engines before and didn't care about inquiring about the details of the Rotax beforehand despite me offering to get him any information he might desire.
He claimed that the prop was already pretty well balanced when I got there (despite my toes experiencing quite some vibrations in flight) and he said he got it down to 0.02 IPS with one washer in one location on the spinner mounting plate. Pretty lucky, ey?
Well, I was suspicious but did not want to spend $4k to buy my own prop balancing equipment. Lucky for me, my friend Don deep down in Texas, had bought a PB-4 from SmartAvionics in the UK and was willing to lend me the tool to do my own balancing job.
This tool is a lot cheaper than traditional tools, i.e. Dynavibe, partly because they require a WiFi connected device for the visualization of the data and for the user interface. A smartphone or a tablet or a small laptop would all work with the PB-4 as long as there is a web browser installed.
I used my iPad Air 2 as it seemed to be the best compromise between display size and ease of use.
Before I bore you with the set up of the sensors and prop indexer, let me summarize that the first measurement indicated 0.4 IPS @ 4000 engine RPM with the balancing weight installed and 0.35 IPS at the same RPM with the weight removed. That means whatever the guy was doing the first time around was making it worse instead of better. Of course, nothing had changed on my prop between 2014 and now. So my suspicions were confirmed.

Now to the set up. Don had never used the tool before and likely wouldn't have cared to use it on a Rotax anyway as he has a Viking engine on his RV-12. So I had to find a way of attaching the sensors.

Optical Prop Indexer

Close-up of tape installation

The prop indexer was a bit tough for me to get right as I misunderstood the purpose of the tape that goes on the back of the prop. I think it just needs a rough white or off-white tape to register the red LED light from the sensor. Instead of reading the manual I just tinkered around (waste of time, I know) until it worked for me which was a piece of white paper along the rotation axis of the prop (taped on with clear tape) and I added a piece of aluminum tape vertically along the prop blade in the middle of the white paper. That produced a stable RPM reading on the PB-4 side which was what I wanted to accomplish. You might get away with just some white tape on the back of the prop.

The actual accelerometer was mounted just outside of the view of this picture to the right and went on the back of the reduction drive. It has tapped holes there for M6 bolts which came with the PB-4.
Speaking of bolts... The angle that I built from scraps that holds the optical sensor, is mounted to the engine with M8 bolts that go into tapped holes opposite of the fuel pump location. That allows for a sturdy mounting point that gets the sensor close enough to the prop with enough clearance to allow for vibrations during start-up to not enter the propeller and cause havoc.

The cables were all hooked up and tied down with cable ties and lead straight under the canopy into the cockpit where they were attached to the PB-4 box.
That allows for an easy handling of the on/off switch of the unit and a close proximity to the iPad in the cockpit.
The balancing procedure is pretty simple. Remove all previously added weights from the spinner, Run it in polar view and find a reasonable RPM (I chose 4000 engine RPM to start with) and see where the dot settles. Take some readings and this will be your starting point - 0.35 IPS in my case.

Then you add a bit of weight - the manual says anywhere - but I decided to put it opposite of where I saw the imbalance in my polar plot. That got me straight down to 0.1 IPS. For the new point, add where you added how much weight and it will tell you where you should put the weight in the next step to improve the imbalance. Rinse and repeat until you are satisfied.

Some advice on "being satisfied". I had my value down to 0.02 IPS at 4000 ROM and was very happy. 4000 is not a realistic value for flying the airplane though as we usually cruise above 5000. At least I do, 5400 usually. So I ran the engine up to 5000 (I stopped there because the whole frame started to wobble around and my prop is set to not exceed over 5200 WOT on the ground anyway). That was disappointing as the point of imbalance changed now and the IPS had increased to 0.12 IPS indicating it would get worse at 5400.

So I went back and redid the balancing procedure to get it down to 0.04 at 5000 engine RPM with a slightly higher value at 4000, I believe it was 0.08.

Now how does it feel in flight. I have to admit I am not completely impressed with the in-flight improvements yet. I can still feel quite some action going on in the rudder pedals that I believe to be vibrations from prop imbalance. Don't get me wrong, it did improve a lot over the previous condition but it feels much worse than when on the ground at 5000 RPM. I suspect that the nose wheel being on the ground impacts how the forces can resonate which likely skews the readings. The perfect setup would be one where you can record the measurements in flight.
Another way of getting close to the in-flight measurement might be to raise the nose off the ground during the balancing job. Maybe by adding a pulling force on the tail tie-down point.

BTW, I ended up with about 3 grams of washer in two places each to make a big improvement over the naked spinner setting.

Blue is my 0.35 starting point

Point 1 was with the wrong weight in place, the blue dot marks the real beginning with the wrong weight removed. The yellow spot indicates where I decided it to be good enough and close to noise measurements. I wish the PB-4 would allow to record the vibration spectrum in flight without the prop indexer in place. Unfortunately, without a proper index signal (0 RPM reading) it wouldn't show any data at all. All I would like to know is what mixture of vibrations I am seeing in flight. It would be easy to determine which spike the one is that is caused by the prop as it would engine RPM divided by gear ratio in the spectrum. Maybe what my toes are feeling are not prop related vibrations and couldn't be fixed by tuning it down more. Would be nice to know and I just don't see a good way yet how to install the optical indexer in a way that's safe for flight. 

Oh, and at last - after all the on/off/on for the engine which really is tough on the battery - I finally pulled my original Odyssey battery and replaced it with a new one. After 4 years it was getting a bit weak on cold days and as a matter of fact, the first start that morning of the balancing job, the blade moved to the compression spot and briefly stopped before pushing through and starting. That showed me that the battery's life span had reached its end.