[rvmills]

Sorry in advance, as this is a long post, and will probably take a couple posts to get the pics in. But it may be of interest to the engineering minded (and others) among us. But lots of pics, so maybe it'll keep the attention of the Navy and Marine Corps pilots (I can say that, I are one! ) I just hope the length doesn't kill ya!

Shortly after purchasing my RV-6, I was told about Kahuna's website and his detailed prop clocking description. Since I have the same engine/prop combo (IO-540/Hartzell 2-bladed D-twist paddle), his success at reducing vibrations piqued my interest, and I've wanted to follow a similar protocol.

I started this last week with a trip to Golden State Propeller, owned by Eric Struve (the oldest son of the gent that gave me my first ride in an airplane a long time ago). My intention was to first check the dynamic balance of the propeller, and if it checked good, go forward with clocking the prop to see if I could reduce vibes. As it turned out, the DB on the prop was out...it likely had not been dynamically balanced previously. Due to time constraints and a desire to fly a bit with the newly balanced prop, I did not re-index the prop, but the vibes are lower after the DB, and I'll fly it as is, take a look at my engine mounts and consider replacing them at the next condition inspection, and consider re-indexing the prop after that (keep with the protocol! )

What I did do along the way however, was to take vibration readings before, during and after the DB, using the "Vibration" application for the iPhone. I saw a discussion of the app here: http://www.vansairforce.com/community/showthread.php?t=43372&highlight=iphone. So I purchased and played with the app, and joined the author's forum to discuss methodology and get smarter on vibration measuring.

In discussions on that forum with the author and a gent that did a semi-rough DB on his prop using the app, I found out a lot of good info (the most important of which is that those guys are a lot smarter than I am! ) But armed with a few new nuggets and some ideas, I set out to see what I could measure. One of the discussion points we had on that forum is that the app is probably best suited for relative measurements, versus being a tool that might replace an expensive dynamic balancing device. Of course, that's what I wanted to do...measure before and after.

I flew from Reno to San Luis Obispo, and on the way down, took measurements at 2000 to 2500 RPM (in 100 RPM increments), at 8500', constant MP. I took the readings twice for each...once with the iPhone mounted in it's cradle/Ram mount, and once holding the phone against the panel by hand. Here's a pic of the phone in the cradle:

And here's a sample of the output of the app. I believe this is at 2000 RPM:

With a hi $ vibration analyzer and solid mounting (like Kahuna used), one might expect to see spikes at the engine RPM, and 1/2 the RPM (1/2 rev as the vibe guys called it...they also said I might see spikes at 1/3 rev and 2/3 rev given I have a 6 cylinder motor...more about that later). From this pic, it looks to me like the Ram mounting was transmitting some odd vibe spikes, and the resulting plot is not really easy to work with or analyze. All the readings I took on the way down were similar, so I decided to try mounting the phone to the panel with some thin 3M rubberized sticky tape for the follow on work. That tape held the phone quite tightly, and though the rubberized surface add some "compliance" to the system, it seemed to give me a platform for some relative comparisons. In addition to tightening up the mounting method, I improved my data collection method, as I lost track of which plot was which RPM on the way down, as I captured screenshots. But I figured out a way to mark the screenshots to ID the RPM. So though I did not end up with good before and after data sets, and I won't go anywhere close to saying one could balance the prop with this (at least with my methodology) but it made for some interesting observations, that I'll share in the next posts. Perhaps someone will be able to take this a step or two farther, and see what they can do.

In the next two posts, I'll compare readings during the dynamic balance runs, and then in the third post, I'll show some comparisons between readings in flight after the balance, at different engine RPMs.

More to folla...

Cheers,
Bob

 

[rvmills]

Part 2

Once in SLO (KSBP), Eric set up his dynamic balancer (he and it shown below):

It has one accelerometer mounted behind the prop on the engine. Eric put a piece of reflective tape on one blade, and called that the index, or 12 o'clock. The readings are taken with a strobe, and show as movements in IPS (inches per second) at a clock position relative to the index. From reading and discussions, more than .2 IPS is a noticible vibe, less than .2 IPS is desirable, and less than .1 is good, with the goal being getting as close to zero as possible. My DB took 6 runs, and weight was added or removed from various places on the hub after each run (all runs done at 2300 RPM). The runs were as follows:

1. .45 IPS at 12 o'clock
2. .5 IPS at 6 oc (too much weight added!)
3. .3 IPS at 5 oc
4. .2 IPS at 3 oc
5. .15 IPS at 2 oc
6. .05 IPS at 12 oc

So Eric was able to reduce the vibes by about 9 fold from the initial reading, and he was confident I'd feel the difference (I do!).

I attached the iPhone to the panel with the sticky tape, and took a 10 second sample during each run. Pics of run 1, 4 and 6 are below. The app takes 1024 readings per sample. The time length of the sample is determined by the speed (Hz) setting of the sample readings. The app author told me that the sample plots will show best when the speed setting is set at twice the speed of the engine/prop, plus 10%. Since the speed of the engine was 2300 RPM (38.33 Hz), and I would later be measuring 2000-2500 RPM (33-42 Hz) in flight, I used the fastest app setting of 100 Hz to give a common scale. The app results show in two plots; one in amplitude versus time, and one in amplitude versus frequency (or RPM). The scales are in g for amplitude and in Hz or RPM for frequency. There is a formula to convert g to IPS, but I did not make those calculations, as I was really just doing relative comparisons, the data was not really conclusive enough to justify it (IMHO), and the formula pretty much made my head hurt! Once the sample is taken, a touch of a finger on the plot pops up an inspector cursor (a white line) that can be moved across the plot to show readouts of amplitudes at specific frequencies. I placed the cursor at 38.28 Hz (close to 2300 RPM), and then took screen shots of the plots.

The amp versus freq seems to be the best measure for this use, and it somewhat matches what Kahuna's graphs look like. The Vibe app presents three axes (one from each of the iPhone's 3 accelerometers). X is red, Y is blue and Z is yellow. The way I had the phone mounted, X= pitch, Y= yaw, and Z= thrust line or longitudinal axis of the aircraft. Here is a pic of the panel mounted phone:

From my reading of Mike's site, he used the 1/2 rev spikes to show the decrease in vibes after re-indexing the prop. The gent from the Vibe forum also used those to balance his prop, and he said I may want to also look at the 1/3 and 2/3 rev spikes for a six cylinder motor (his was a 4 banger). In these three shots, you can see that that there are not real noticible spikes at 1 rev, but there are some spikes at or near 1/3, 1/2 and 2/3 rev locations. I still think the data is somewhat hard to make conclusive measurements on and don't feel I could make weight adjustments based on them, but I do think you can see the spikes change orientation as the vibe moves from 12 to 3 o'clock (X spikes lower, Y spikes higher), and then it reverses as the vibe moves from 3 back to 12 o'clock. However I did not see a real change in overall amplitude in the samples from the DB runs, even though the vibe did goe down, per the balancer. The Z axis was not as useful, and the spikes there my be due to movement of the aircraft against the brakes or the tiedown rope on the tail, or some buffeting of the tailwind we had that day. Here are the pics of the DB runs 1, 4 and 6:

Run 1 (.45 at 12 oc)

Run 4 (.2 at 2 oc)(X spikes down, Y spikes up)

Run 6 (.05 at 12 oc)(X pikes up, Y spikes down)

I printed out the screenhots I took, and tried to analyze them more closely, but honestly don't feel I could balance a prop with the app. However, given some further tightening of the mounting process, and with someone with more brain power to analyze the data, perhaps a good method could be developed. The author has just released an upgrade (that is at Apple now for implimentation) to show the amplitude in mm/sec, and may develop another upgrade to show it in IPS, and he was very helpful as I researched this. So who knows, it may have some more powerful uses down the road.

In the next post I'll show a few plots of the readings in-flight the day after the DB, in which I feel show changes in amplitude that are much clearer at different RPM settings.

Still mo to come...

Cheers,
Bob

 

[rvmills]

Part 3

The day after the DB, I flew for about an hour, taking samples again at 2000-2500 RPM, in 100 RPM increments. 7,500' MSL, 22" MAP were held as constants.

Subjectively, it did feel smoother at all RPM settings than before the balance. It also felt to me as though the engine was smoother at 2500 and 2000 RPM (2000 being the smoothest of all), with a bell curve of increasing vibrations in between, peaking at 2300 RPM. Insteresting, in that the DB was done at 2300 RPM. But if it's .05 IPS at 2300 on the balancer, and feels smoother at higher and lower RPM settings in flight, that seems pretty good to me (a very unscientific theory though!)

Here are the plots for 2500, 2300 and 2000 RPM. I figured out that I could change the scale to RPM before these tests, so the cursor is set to the engine RPM (or close), and this provides a good marker for the picture (to keep them straight for IDing later), and it also provides a good way to help find the 1/3, 1/2 and 2/3 rev spikes.

2500 RPM:

2300 RPM:

2000 RPM (actually does look smoother, I think):

I think that if and when I do come back to clock the prop, I'll take more before and after readings, next time using a bit more strict and standardized testing procedures (having learned from the experience), and should hopefully be able to see (and feel) the differences. I think that will be the best use of the app, which is a pretty neat little program, and provided some good fun (and a good excuse to do some nice flying over the California coast!) Thanks Mike for your great website and descriptions, and hope I can match your improvements over time!

Bottom line on this experiment:

Vibration app: $4.99 (worth it!)
Gas for the X-C and test flights: $150 (well worth it!)
Dynamic Balance: $200 (way worth it!)
Having a burger with your kid on the veranda, watching planes takeoff and land in San Luis Obispo, and him smiling: Priceless!!

Cheers,
Bob

 

[Lou]

Relative readings

The vibration transmission path(s) to the iphone sensor are going to have a large amount of influence on your readings. Not only are mechanical forces being transmitted via multiple paths (each engine mount, frame, longeron ....), but you have the propeller thrust pulsations on the airframe. When these vibrations reach the sensor, multipath effects of constructive and de-structive reinforcement will cause interesting anomalies in the received vibration levels at that particular measurement point. Your noise cancelling headsets use the out of phase injection of received noise to reduce the noise level in the headset.

As far as trying to actually do any balancing with it, I would vote no. There are too many measurement factors with phase problems due to multiple transmission paths involved. If you consistantly use the same measurement point/iphone mount, a baseline could be established for relative comparison purposes. These data could be used to indicate the need for further balancing later on down the road. Especially if a sudden increase in observed levels is noted, much like the vib sensors on the jets. I would think that this is where the 5 buck investment for the application might pay off in spades. If something begins to break down in the power plant, it is probable that vibration levels would begin to increase and you might see this reflected on the iphone.

You will see sweet spots and hot spots. Your configuration with a six cylinder and 2 bladed prop will have reinforcements on the 3rd blade rate and first engine firing rate (6th cylinder rate). At 2300 engine rpm that will produce a propeller blade rate of 76.66 Hz and engine firing rate 230 Hz. The 3rd blade rate harmonic will reinforce the first engine firing rate (hot spot). There will be many other relationships such as this that will affect the vibrations levels at various frequencies. Again, your specific measurement point will have it's own set of demons that will introduce problems.

Just my .02 input. If you can pull it off for 5 bucks my hat is off to you, but I am betting against it.

 

[rvmills]

Lou,

Completely concur with you...and so does the app's author. Our discussions were on target with what you are saying...that it's best use in our application would be relative measurements and comparison, and that anything that introduced compliance (like engine mounts, panel mounts, cradles, and even the rubberized sticky tape) would introduce either damping or aditional vibes. We also chuckled that trying to mount the iPhone to the engine for the test would likely void the warranty on the phone!

My intent was mostly to take before and after readings to see if I could see a difference in addition to feeling it (I think I did, at least a bit), and secondarily to take concurrent readings during the dynamic balance (with the "real deal") and see if they correlated (not so conclusive there, and probably would be a stretch). I really didn't set out to try to do the balance with the iPhone, and would agree with you about it's feasibility (I wouldn't take your bet! ). But it was a fun way to geek out a little.

Your comments on the firing and blade rates are interesting. That coupling of blade and engine pulses (I believe) is a big part of the prop clocking process that Kahuna undertook, with very good success. The blade rate at 76 Hz may be what I saw manifested in spikes above the engine RPM in my samples...I'd been wondering where they came from. But those, and pulses at 230 Hz may be beyond what the app can effectively measure. Interesting discussion points though. Thanks!

Cheers,
Bob

 

[Mel]

Very interesting, using the I-Phone as a check. Regardless of where the vibrations come from, these are the same ones transmitted into your instruments and avionics and the more you can do to reduce them, the longer things will last.

 

[rvmills]

Very interesting, using the I-Phone as a check. Regardless of where the vibrations come from, these are the same ones transmitted into your instruments and avionics and the more you can do to reduce them, the longer things will last.

Spot on Mel...and they are the same ones (or similar) to those transmitted to your be-hind, so reducing them might make it more comfortable as well (and keep Momma happy too!)

Cheers,
Bob

 

[Lou]

Roger that, I am on the same page with you now.

Sorry about that, I mis-understood the direction of your post. I agree whole heartedly that this would be a great exploitation of the iphone. Especially since the associated relative levels are displayed in a digital format (no interpretation needed). If you can make a mount that would produce repeatable results this would work great (a dedicated secure spot).

 

[Lou]

Corrected numbers for frequencies on your Iphone app

Bob,

I realized that I had given you the computed values for a 2 cycle vice a 4 cycle engine a few weeks back. If you are still looking at the frequencies on your Iphone app, here are the correct numbers and the method for computing them. Based on a 4 cycle 6 cylinder (not geared) engine and a 2 bladed propeller.

Engine RPM 2300
Engine crank shaft rate = 38.33 Hz (2300/60)
Engine cylinder rate (4 cycle engine) = 19.16 Hz (crank shaft rate / 2)
Engine firing rate = 115 Hz (cylinder rate * number of cylinders)

Propeller RPM 2300 (for a direct drive non-geared engine)
Propeller shaft rate = 38.33 Hz (2300/60)
Propeller blade rate = 76.66 Hz (prop shaft rate * number of blades)

You will see reinforcements at any coincident multiples of the blade or engine rates. The third harmonic of the propeller blade rate will match the second harmonic of the engine firing rate which will result in higher noise levels at that frequency (229.92 Hz).

Sorry about giving you the 2 cyc vise 4 cyc numbers earlier. Hope this helps decode the observed frequencies on the Iphone app. It looks like the cylinder rates are showing up nicely.

Regards,
Lou