TEST VIBRATIONS 01

Pour I Phone

his is my first time to join a forum of any sort. So I'm hoping I've navigated the log-on/posting correctly. I agree the Vibration Application opens up some wonderful opportunities. And yes, I believe it is capable of performing a propeller dynamic balance. I own a 1946 Ercoupe with Continental C-75 and was always annoyed by the buzzy feel at cruise power on long trips. I started about 3 years ago to develop a dynamic balancer. Knowing very little about electronics , it took $1000+ invesment and ended up with a marginal system. I was continuing to throw money at it when I found the Vibration Application. Applying what had I learned earlier, the iPod approach proved so much easier. I have reduced the Ercoupe vibration ~ 70% on first try. I'm writing the procedure up in the format of an article. It's not quite done yet. I'll be glad to share what I know on this forum. The article is over 12 pages and 8mb..so it may take a while to get through it. Fist step is understanding where the vibration comes from. Assuming a conventional 4 cylinder , 4 cycle direct drive aircraft engine, you want to be able to record the vibration component at 1/rev. For example if the engine is running at 1200 RPM or 20 hz you want to see the vibration signal at 20 hz. At 1200 RPM the engine will generate vibration signals at half per rev (10hz) and 2/rev (40hz), etc..That's because the cam is turning at half the crank speed, 2 pistons ae firing per revolution, etc...The accelerometer pics up all these signals. The FFT page breaks these components out. So first step is take a reading and see if you can identify the 1/rev signal. I was able to get a good , repeatable signal by simply holding the iPod against the instrument panel. ...I'll continue later. Posted on 14-5-2009 at 06:10 PM Posts: 19 Registered: 13-5-2009 Member Is Offline

To continue.. I was hoping to attached an example recording from the Vibration application....but haven't figured out how to do that here...I have a data recording in flight at 2000 RPM that shows a 1/rev vibration at 33 hz and peak of .2g in Y and Z axis..a little less in X axis. There is a vibration peak at half/rev or 16 hz of ~1g. There are a couple other peaks but they are not coming from the engine since they are not a multiple or fraction of the 2000 RPM signal. It is the 33 hz signal that is coming from the propeller-crankshaft. It is the peak vibration signal we want to reduce. It is also the only signal we can change by adding a correction weight in the plane of the propeller. Vibration from the propeller-crankshaft is due the radial force of the unsymetrical mass rotating around the center of rotation. We can minimize the vibration by adding an equal force in the opposite direction with the correction weight. The challenge is to figure out where to place the correction weight and how much weight to add...I'll continue in next installment...If it's not too boring.. Tom Zucal Propeller Dynamic Balancing Installment 3... I realize I'll be out of town next couple days so I thought I would add some additional information on this subject....First of all there is a correction in the previous posting.. the 16 hz recording should read .1g (not 1g)...Before I get too far into explaining the balancing procedure there are several things that should be checked and confirmed before starting. The propeller static balance should be checked. This is normally done by installing an arbor in the bore and placing the prop on parallel knife edges. When the prop is statically balanced it won't move when placed in either a vertical position or horizontal position. With the prop installed on the engine, the blade track should be within 1/16 inch blade to blade at the tip. The engine rubber mount bushing bolts should be properly torqued. Ignition timing sould be adjusted within spec and all compression readings should be nearly equal. The other thing I would like to cover before going further is vibration limits. ..Vibration is normally measured in IPS (Inches per second), Mils (Milimeters) and acceleration (g). The Vibration application measures in g(s). The standards for aircraft propellers is .07 IPS peak or less is good, .15 IPS is okay and anything over .25 IPS is considered rough. At 2000 RPM, .25 IPS is about .15g, .15 IPS is about .08g and .07 IPS is .04g. Engine vibration is normally measured on the engine crankcase near the propeller. It would be ideal if we could place the IPOD on the engine to take a vibration measurement, but that would be unsafe with the engine running. If the measurement is taken off an airframe component in the cockpit, I don't believe we can make any hard conclusions about the vibration limits since our vibration data will be affected (either damped or amplified) by the airframe geometry, materials, engine mounts etc. ...I'll pick up with balancing procedure in next posting... Tom Zucal

posted on 14-5-2009 at 07:50 PM

Propeller Dynamic Balance Inflight viration data recording at 2000 RPM (33 hz). 1/rev vibration from propeller- crankshaft is clearly visible...Tom Zucal TomZucal has attached this image:

TEST VIBRATION 01.doc

-1-

Aircraft Vibration Thanks for the information Tom, I'm looking forward to learning more about your project. Another application I'm pursuing is the monitoring of industrial pumping equipment. Example: Our sales and service guys visit the customer locations on a regular basis, by developing and marking a repeatable position on the equipment a quick vibration signature can be captured and recorded. This data can be emailed back and compared with previous data for trends. Has anyone developed a excel sheet to graph and display the data? D. Pansier Aircraft Vibration Yes, I believe the iPod could be used for diagnostic purposes. I don't know anything about industial pumps but I would think you could detect an impending bearing failure for example by tracking vibration signal changes over time....assuming their failure modes are gradual. Even for the aircraft engine application, tracking the half/rev signal may indicate health of the engine, leaks in induction system, poor ignition in one cylinder, etc....key to both applications would be to take data consistently (same location, same RPM, etc)... Tom Propeller Dynamic Balance Installment 4....Before I go further into my method for balancing a propeller, I should caution any readers that I am not an expert in engine dynamics or vibration analysis..I do have an engineering background, work in the aerospace industry and have an A&P certificate..but that doesn't make me an expert.. I'm open to critque of my method and any recommendations...Also, for anyone with a certified aircraft, FAA Advisory Circular AC 20-37E outlines acceptable balancing methods. My method is not FAA approved and therefore should not be used on certified aircraft (or engine)...For the kitbuilder & experimental aircraft community, there should be no restrictions......TomAnyone aware of a low cost method to present the captured data in a 3D graph?

TEST VIBRATION 01.doc

-2-

D. Pansier

Propeller Dynamic Balance Installment 5... The next phase is to add a test weight in the plane of the propeller at different locations (different clock positions but same radius) and take peak vibration data recordings at each location (same RPM) and then plot that data in polar coordinates, we can find the location of dynamc unbalance. One way to think of it is, if we had a way to install this test weight at 360 locations around the propeller and we took vibration data at each location, the data plot in polar coordinats would (should) be a circle with the center somewhere other than the center of the graph (assuming there is dynamic unbalance)....because as we place the test weight closer to the unbalance location the peak vibration readings will be higher and as it is placed further away (opposite) side, the readings will be lower. Luckily, we can do this is in a minimum of 3 runs and not 360. ....First step is to decide on what RPM to run test points at and where to take the data. This may take some experientation and readjustment after getting into the process. I chose 1200 RPM because in my application, the engine speed was high enough I could take a strong repeatable peak vibration reading from a location on the instrument pannel....and the speed was low enough, I could run up on the ground with reasonable safety....(1200 RPM also is a convienient round number of 20hz which isn't important)...the next step is tofind a place in thecockpit to take data. In my appliation, I found holding the iPod perpendicular to the instrument pannel worked well. The X axis was top to bottom, Y axis nose to tail and Z axis was wing tip to wing tip. You may need to find anther location, floor board, longeron etc...some where you can record a strong repeatable data point. I found in my application, the Z axis data was most useful to work with. You will take all test data at the same speed and location. ..I took a baseline data recording without a test weight at these conditions below.....Tom Propeller Dynamic Balance continued...Note there is a peak vibraton of a little over .06g at 20 hz in the base line run above.......just left of the white vertical line....Next step is to make up a test weight. I used 2 AN970-3 washers and flat lead fishing sinker with 3/16 hole drilled into it. Total weight was 32 grams. For the tests, I could easily install this weight on the spinner attachement screws. These screws (8 locations around the spinner at the rear spinner bulkhead) are roughly evenly spaced and at about a 10 inch diameter. This gave me ~150 gram inches of dynamic weight. The weight isn't critical...but you don't want to use so much that you could damage the engine...I would think ~ 300 gram inches would be an upper limit. Also make sure the weight is attached well during test because if it were to liberate during the engine run, you could cause harm or injury to someone.. or property...I selected a spinner screw location at random and marked it '0' degrees. I installed the test weight and ran at the same conditions above...The result is as follows:

TEST VIBRATION 01.doc

-3-

Propeller Dynamic Balance continued... Note we have a peak vibration Z axis of ~.05g at 20 hz.....Now relocate the test weight at 120 degres from the reference location. The result is : TomZucal has attached this image:

Propeller Dynamic Balance continued...Here we have a peak vibration reading of .145 g at 20 hz in the Z axis...Now relocate the test weight at 240 degrees from the reference for the next run..Rsults as follows:

TEST VIBRATION 01.doc

-4-

Propeller Dynamic Balance Continued...........Note...I changed the scale for this reading above... We have ~.1g at 1200 RPM in Z axis.......Now we have three data points... .05g at 0 degrees, .14g at 120 degrees and .1g at 240 degrees. We can use this information to find the location of the dynamic unbalance. If we plot these 3 points on a polar graph and we draw lines between these three points to form a triangle and we find the circumcenter of the triangle..that will be the location of the unbalance. The circumcenter of a triangle is where the three bisectors of each leg of the triangle meet at a single point... In this case it is located at 160 degrees....So now we know where to place the correction weight to reduce vibration due to dynamic unbalance....In this case that's 340 degrees (180 degrees opposite the unbalance location).....The next question is how much correction weight to add at 340 degrees? There is probably a mathmatical way to calculate it based on our data...I sugesst it may be just as quick to try a couple weights and measure the vibration till you get a minimum reading. On my first try, I added two AN washers of about 7 grams (35 gram inches) at ~ 320 degrees (closest spinner attachment screw) and recorded the following:

TEST VIBRATION 01.doc

-5-

Dynamic Propeller Balance continued...Note the 1/rev vibration dropped down from .06g to under .02g ......So I conclude this process seems to work... I wish I had access to "FAA approved" test equipment to confirm results....A couple last comments...the FAA does allow perminantly attaching the correction weight with a screw drilled through the spinner back plate...(following using 'approved 'balance procedures..which as previously stated this method described on this forum is not 'approved').....If you come up with a correction weight solution of over ~ 150 gram inches, I would suspect there is a problem and I would consult with a qualified balancing shop. .....I'm open to pot shots, recommends ,comments, etc...I hope someone finds this useful... Tom Zucal I also want to say thanks to Bruce for developing the Vibration application. It's a great tool...Tom BruceTruax Super Moderator So now that the propeller is balanced can you actually feel the difference? Is there noticeably less vibration? Great write up by the way. Even though the method is not FAA approved it would allow an individual to determine the level of imbalance, apply a weight for test balancing and see if it helps. If it does and they want it to be permanent, they can call and FAA certified tester and get it done officially. If it does not improve the vibration then perhaps the problem is somewhere else. Bruce Truax Diffraction Limited Design LLC Propeller Dynamic Balance Bruce ..Thanks...Yes subjectively it does feel smoother. I haven't had a chance to take a vibration data reading to compare with my data at 2000 RPM before the balance procedure. I will post it next chance I get...I believe what vibration is left is due to other sources such as differences in piston & connecting rod weights, generator (turns 2/rev), etc....I agree this probably could be used to check balance on a certified aircraft...and still stay within the regulations....Tom Tom and Bruce, Great thread, and related to a propeller project I've been considering using Vibration to help measure results on. However, not being an engineer, and just getting to know the Vibration app, I'd like to ask a few (probably dumb) questions. For background, I have an experimental aircraft with a six cylinder motor and a two-bladed prop. There has been some success with reindexing the prop on this engine-prop combo to decrease the vibrations felt through the airframe. A good example of that process is documented on this website: http://www.mstewart.net/super8/propclock/index.htm In that process, the aircraft owner used a portable measuring device that measured in IPS and displayed results over a wide range of RPMs. I'm trying to get my arms around how best to use Vibration to somewhat emulate his results, or at least get a qualitative measurement of vibration. I'd like to run the tests in flight, with my IPhone fixed to the instrument panel, and mesure vibes at engine RPM settings of 2600, 2400, 2200 and 2000 RPM (or 43, 40, 37 and 33 Hz). Of special interest (as Bruce mentioned in his post) is the peak vibe at 1/2 of each of those speeds/frequencies. In playing with the app, it appears that the settings for the Time Series goes from 10-100HZ, but the Frequency page shows .1-1 Hz (or .6-60 RPM). Tom and Bruce, First question I have is about the best combination of settings to use during the samplings. Should I set the Sample Rate to the freq I want to measure (ie 43 at 2600 engine RPM, 40 at 2400, 37 at 2200, 33 at 2000) and then set it half of those to see the 1/2 rev vibe, and take a snapshot of each, and note the max g spikes over the sample time? Or should I choose one Hz setting and just do all of the measurements there (at all engine RPMs) and compare the results? I figure I'll use a sample length of about 30 seconds for each reading (or as close as the program will allow for each. I also plan to have the DC Remove selected to ON, and the Hamming Window OFF, and will just leave all axes (X,Y,Z, Horiz & Vert) ON. Does this sound like an appropriate plan? (and Bruce, what did you use?) Bruce, you mentioned that your good/fair/rough thresholds at 2000 RPM were .25IPS=.15g, .15IPS=.08g, and .07IPS=.04g. Would those numbers hold true (or close) for the other engine RPMs I'll be evaluating? I'll be running these tests under the eyes of a friend that runs a prop shop, and will run a dynamic balance with his equipment after reindexing the prop. However, the in-flight measurements with Vibration will be very interesting to see, to supplement the subjective seat of the pants feel. If either of you have recommendations on how best to set up the app for these tests, it would be very much appreciated! Thanks, and cheers! Bob Some suggestions and information Bob, First here is some theory regarding frequency measurements. 1. The highest frequency you can measure is 1/2 of the sample rate. So if your sample rate is 100Hz the highest measurable frequency is 50Hz. I would recommend not putting your highest desired frequency right at the top of the plot. Sample at least 20% faster than you need to to resolve your frequency.

TEST VIBRATION 01.doc

-6-

2. Once you decide the sampling rate then decide how many samples you want. The number of samples will determine the lowest frequency you can measure (1/ total time). Adding more samples will also help you get a little better signal to noise on your other frequencies. 3. The Hamming window has many advantages. The primary advantage is that it tends to smooth out the peaks which fall between frequencies. The reason for this is very mathematical but I recommend using the Hamming window. It usually results in better data. 4. Looking for maximum G spikes is not useful in balancing applications. Spikes are indicative of one time events. If you have an out of balance condition it will repeat with every rotation of the prop giving you a nice periodic signal which will be very obvious in the frequency plots. My recommendation is to use the log-log plots and the inspector cursor to look at individual peaks. 5. Unless you plan to measure how many G's you are pulling in a turn you should probably turn on DC remove. This will make the time series data easier to view. The next version of Vibration coming out in 1-2 weeks will have units of mm/sec. You will have to convert to IPS by dividing by 25.4. I think I will add English units in an upcoming release since it seems to be popular. One thing to remember is that IPS or mm/sec are measuring the motion of the iPhone. If it is mounted on the engine (probably voiding the warranty) it will measure a different number than if it is mounted on the instrument panel, particularly if the engine is mounted to the airframe with compliant mounts (rubber engine mounts). The numbers will also be different if the iPhone is mount in some sort of cradle. Each time you add compliance in the path from the source to the iPhone the amplitudes will decrease (unless you create a resonance in which case you will get crazy numbers). Having said all of this, you need to use the iPhone (or any other vibration measurement tool) primarily as a relative measure. You make a measurement, change something and measure again. Or as Tom was doing, make three different measurements with only one change (the location of the weight) and use that information to make a change and then compare the before and after results. I hope this helps. I will make sure that Tom reads your post. I am sure he will post some suggestions. Bruce Propeller Vibration Bob, In response to some of your questions, I can offer the following comments and hopefully answers: 1. If you have a 6 cylinder engine and 2 bladed prop,you will need to think about what frequencies will be generated by the engine so you can distinguish them from other vibration signals. I have to think about it myself, but you might see 1/3 per rev signals and harmonics for example. If the prop is a direct drive (no gearbox reduction), then you should see a 1 per rev signal. 2. I was able to get good readings at 2000 RPM. 2600 RPM might be running up to the iPods upper limit. If your goal is to balance the propeller, there is no need to run up to max RPM. You will see imbalance at part power. 3. I played with settings and found the longer the data reading taken (like 10 seconds), the more consistant and repeatable were the results....Not to get too side tracked, but before I found the iPod, I mounted an accellerometer on my engine with a very narrow band pass filter and found that I had to average that data over a period of seveal seconds in order to get repeatable readings..I'm not sure why that is, but the engine vibration levels seems to cycle up and down over shorter periods of time. 4. The vibration range 'limits' in g units will be different at different RPMs. I found a handy tool for converting IPS & Gs, etc provided free for the asking from ENDEVCO Company. It's a slide rule called 'Vibration Computer'. I don't know if they still give them away, but you might check their web site at http://www.endevco.com . 5. Do you plan to use, the iPod while your friend performs the dynamic balance? I think it would be interesting to see how well the iPod tracks the other balance equipment readings. I'd be very interested to hear what you learn. I have to defer to Bruce on the technical aspects of measuring vibration. As Bruce states, the iPod usefullness (and it's very usefull) is in taking relative measurements. I'm looking forward to trying future upgrade versions of the application. Bob....Keep us posted on what you learn during your project. Bruce..Thanks once again TomZucal Bruce and Tom, Thanks for the quick replies! I'm fathoming most of what you guys said, but still find some knowledge gaps in my noggin'! Should have paid better attention in ENG 101 30 years ago! The prop indexing experiment is very much a before and after experiment, so the relative measurment you both spoke of is exactly what I'm looking for. To expand on that, 2 bladed props are normally indexed so the blades are in the 1 and 7 o'clock position when the number 1 cylinder is at TDC of the compression stroke. In the website I referred to, the gent worked with the manufacturer to reduce annoying vibes in cruise flight (a common issue with 6 cyl engines and 2 bladed props, from what I gather, and one I see a bit of), by re-indexing (re-clocking) the prop to a 9-3 orientation at #1 TDC. He was able to reduce the vibe amplitude at 2400 RPM (engine speed) by 85% (in IPS) by doing this. The resulting decrease was less dramatic at other RPM settings, but there was a decrease across the board. The vibrations he measured were most evident at 1/2 rev (or seen at 1200 RPM at 2400 engine RPM). He felt a substantial difference, as well as seeing it in the measurements after reclocking. From what I gather, it's somewhat a matter of syncing the engine firing pulses with the prop blade thrust pulses (probably much more complicated than that, but it makes sense to me that way!). I thought I'd give it a try, and rather than just doing it all by feel, thought I'd use Vibration to quantify it a bit. Since the measurements will be taken in cruise flight, it might be a bit hard to adjust the sampling rate with precision, so I may just use 100HZ, so that I cover the double+20% you recommend Bruce. (I thought about it today, and may measure at 2500, 2400, 2300, 2200, 2100 and 2000 engine RPM, as those are the most common ranges for climb and cruise flight. That means I'd be looking at freq ranges of 41.66 to 33.33 Hz, and the Hzx2+20% range for the top of the scale I'd want to select would be 99.98 - 79.99. I may get more resolution at 2000 RPM by dropping the rate to 80 Hz, but it seems just using 100 may make it simple, and make comparing the graphs pretty straight forward...would you concur with that?) I'll use the 10.24 second max sample time size as you both suggest for each run, to give the best shot at clear data and good readings (I'm guessing that means the program is limited to 1,024 samples, as the time seems to go up in a linear fashion as the sample rate goes down).

TEST VIBRATION 01.doc

-7-

For the log-log plot, does that mean I leave both the horiz and vert axes' switch selections to ON? Couldn't quite figure out that from the manual, but it seems to make sense. I will turn the DC Remove on to mitigate gravity, and will turn the Hamming Window on, as you suggest (won't even pretend to understand why, but I certainly take your word for it!! Something I'll dig into later to learn more about this, and maybe win a bar bet sometime!). After I do the test at each RPM, I'll take snapshots, and then compare them on the ground on my computer. The comparisons between various engine RPMs is more for interest in where the engine/prop runs smoothest (for most comfortable cruise)...but the real test will be to compare like-RPMs before and after re-indexing. I probably won't be able to do much inflight analysis with the inspector cursor (which is pretty cool, by the way), so I'll mostly be SWAGging the amplitude with the scale readings, and looking for lower overall vibes and lower periodic spikes (at least that's what I think I should do so far). One thing I don't really understand is how to tell where the spikes are occuring. By that I mean, if I see a periodic spike, how do I tell at what RPM or Hz it is occuring? Seems to me the plot is amplitude (g in this case) over time. Is there a way of identifying where the 1/2 rev and 1 rev spikes are occurring by looking at the plots...maybe its at certain time intervals? (my knowledge gaps surfacing here!) All this assumes a good dynamic balance, and we'll do that after indexing to 9-3 (it has a previous d-balance now at 1-7). When we do that, I'll try to take some measurments as it's done at the shop, to see if the Vibration plots correlate to the shop's equipment, as you suggested Tom. With respect to the mounting, it will either be attached to the panel, or in the cradle I have. I do understand what you said Bruce about adding compliance. Velcro to the panel will certainly do that, so I'm considering using more solid sticky pads. If I crank down on the Ram mount, it might work well, though it has some angles, and might transmit some odd vectors to the vibes. I'll work on it, and even though it won't be true vibes at the engine (and no worries, the phone isn't going near the engine itself! ), the vibes I hear and feel in the pilot seat are transmitted through the airframe (and the air), so it seems like affixing the iPhone to the airframe in some way is not a bad way to go (and again, like you said, it's all a relative before and after thing). Thanks for suffering through my long posts, and for all of the advice. I'll let you know what I come up with after I do this next week (weather permitting!) Please let me know if you have any other thoughts. Thanks again! Cheers, Bob Propeller Balancing Bob...I think you're on track for correct settings to take the readings you want...To answer your question about the spikes...I think it's important to try and identify where the spikes are coming from...If you go to the first attached screen in the thread above, that is a reading I took at 2000 RPM in the air. It' a little hard to read, but at the bottom, the scale it says 6.25 hz per division. So the big spike to the right of center is about 5.5 divisions or 34 hz..or 2000 RPM. So I know that spike is coming from the propeller and crankshaft. The smaller spike at ~2.5 divisions is the 1/2 per rev signal which ( in a 4 cylinder engine) is coming from combuston process, and as you say maybe the prop indexing...There would also be spikes at 1.5 , 2, 2.5 per rev etc in a 4 cylinder engine if the scale went out that far... The spike on the screen at ~4.7 divisions, I can only assume it is coming from somewhere other than the engine since it isn't a multiple of 2000 RPM....although since the reading was taken from the instrument panel maybe it is picking up a response from the airframe to the engine..but it could be some aerodynamic induced signal while in flight. I could have checked that by reducing RPM an maintain airspeed and if that signal at 4.7 hz remained, that would have confirm it wasn't from the engine.... I think the more you play with the iPod, the more you learn about vibration, and your airplane.....I'm interested, what engine /aircraft do you have. Mine is a old Ercoupe wth a C-75.......Good luck TomZucal

TEST VIBRATION 01.doc

-8-