Rotor vibration monitoring

Oskar

Member
Joined
Mar 10, 2007
Messages
331
Location
Auckland, New Zealand
Aircraft
R22, MTO Sport, GyroBee, Mosquito Air
For many people the bells and whistles on an aircraft are just as important as the aircraft itself.

So I decided to add a few bells and whistles to the new electric helicopter I’m busy testing.

Measuring vibration is often a pain because the equipment usually consists of add-on bits and pieces that are often held together by cables ties and tape. It would be much nicer if all the necessary bits were installed permanently, and vibration displayed on the instrument panel. So that’s what I did. A LCD screen on the instrument panel can be toggled to different displays, and one of them is now configured to show vibration data, both horizontal and vertical.

Another issue, especially with helicopters, is that it’s not all that easy flying and recording data at the same time. A device that records vibration and which can then be analysed back on the ground is a much safer way of balancing rotors. The electric helicopter already has one of those (it records all the data on the CAN bus) so I simply dumped the vibration data onto the CAN bus and the “black box” records it automatically.

Here are the first results, just quick run ups on the ground. The vibration monitoring starts measuring at 180rrpm and the rotors were spun up to 400rrpm. Only horizontal vibration is shown, the left axis is phase and the right axis acceleration amplitude. As the rrpm increases the acceleration increases, but the phase stays constant (as it should).
Rotor vibration monitoring
I then added 0.5g weight to both blades, and already it is clear that B is the heavy blade.
0.5g added to blade A:
Rotor vibration monitoring

0.5g added to blade B:
Rotor vibration monitoring
 
Oskar
Excellent work, with the demise of the PB4 and a bit more work you could have the replacement.
What units are you using for vibration?
Are you filtering the vibration to Rrpm frequency or are your measurements total?
Are you measuring pk-pk, RMS or zero-peak?
Mike
 
Oskar
Excellent work, with the demise of the PB4 and a bit more work you could have the replacement.
What units are you using for vibration?
Are you filtering the vibration to Rrpm frequency or are your measurements total?
Are you measuring pk-pk, RMS or zero-peak?
Mike
Thanks Mike,

At the moment I measure acceleration and scale the amplitude result for convenient CAN bus comms. Converting to any other unitary base is just a matter of one line of code.
 
Oskar

What are you using for sensors

Doug
Doug,

The position sensor is a hall effect sensor which picks up a magnet attached to the MR pulley.

Rotor vibration monitoring

The accelerometer is an old school +-6g, 2 axis, linear accelerometer. Sampling needs to be synchronised with the position sensor which is very difficult to do with the new digital accelerometers.

Oskar
 

Attachments

  • Rotor vibration monitoring
    IMG_20240212_164833.jpg
    15.4 KB · Views: 5
Oskar
Would you like to share a little more information on the balancer. I assume you used an accelerometer module like a adxl335/345 or something similar.and an Arduino board. Would you share the code.

Thanks Doug
 
Oskar
Would you like to share a little more information on the balancer. I assume you used an accelerometer module like a adxl335/345 or something similar.and an Arduino board. Would you share the code.

Thanks Doug
Doug,

I might have made it look easy, but that definitely wasn’t the case.

About 5 years ago I tried to develop a rotor balancer. I got it to the point where it worked really well on the bench, but as soon as I installed it in a gyro or heli it turned into a random number generator. I spent about 6 months trying to get it working but ended up giving up.

A few weeks ago I resurrected the project. I used the same sensors and software, but changed the hardware and system design. And now it’s working beautifully in the heli! Of course the most important factor was that Jupiter and Mars happened to line up, as soon as they start diverging the balancer is going to revert back to a random number generator…

Getting back to the Arduino, the ADXL should work fine and it might be possible to get a balancer working on the bench using an Arduino. What I would not do is try to get a balancer working in a helicopter using an Arduino, I use a CY8C5888AXI and it’s powerful enough for the job.

On the software side, the biggest problem handing out software is that you create an expectation of fixing everyone’s system engineering problems. I really don’t feel like spending the rest of my life answering questions like “Why isn’t it working?” or “Why don’t you use I2C instead of CAN bus or Bluetooth (like the PB4)?”. I’ll manage that issue differently.

Oskar
 
Oskar

Thanks for the info. I understand the At Mega stuff but the Cypress is way over my head. Dont want it for a helicopter want to retrofit my driveshaft balancer. Have been using the adxl and a 2 channel scope 1for the accelerometer and the other an optical reference sensor Thanks Doug
 
Oskar

Thanks for the info. I understand the At Mega stuff but the Cypress is way over my head. Dont want it for a helicopter want to retrofit my driveshaft balancer. Have been using the adxl and a 2 channel scope 1for the accelerometer and the other an optical reference sensor Thanks Doug
Doug,

If the accelerometer signal is reasonably sinusoidal a 2 channel scope is a really good way of measuring. The latest digital scopes will do most of the measuring for you.

If there is an easy way to attach a small magnet to the moving bits I would recommend using a hall effect sensor for the position sensor though, something like a SS443A. Once set up it's much more reliable than optical.
 
Here are the results for horizontal balancing of the rotor. For some this will be over the top, but Mike G and a few others will appreciate the numbers.

From post #1 I already knew that blade B was the heavy blade and that I needed to add weight to blade A. The question is how much weight?
I found some wide tape and wrapped one layer on the tip of blade A. Weight of the tape was 4.4g.

The polar plot moved from point 1 (52@230) to point 2 (55@290).
1 to 2.jpg
4.4g was too much weight, just over twice as much as it should be. To balance length wise about 2.1g of weight needs to be added to the tip of blade A.

The rest is at 90 degrees, thus lead/lag adjustment. A visual check showed that blade B was slightly leading, so I adjusted the lead/lag screws on blade B 1/6 turn lagging.

The polar plot moved to point 3 (30@330).
2 to 3.jpg
1/6 of a turn was not enough, blade B needed about 1/18 of a turn more lag on the adjustment screws.

I put 2.1g on blade A and lagged blade B slightly more. The polar plot moved to point 4 (25@80).
3 to 4.jpg
The weight is now perfect, but the lag adjustment was a bit too much. Backing off just a tiny bit (more a case of pressure on the adjustment screws instead of movement) resulted in vibration levels dropping to less than 10.
4 to 5.jpg
Conclusion:
- The blades weigh 7303g each and blade A needed 2.1g added tip weight to balance. This was the easy bit.
- The lead/lag adjustment screws have a 1mm pitch, blade B needed 0.22mm (less than a quarter turn) adjustment on the screws. The final adjustment is extremely sensitive and becomes a trial-and-error affair. The tiniest bit of movement has a noticeable influence on the vibration level.

Next step is vertical.
 
Oskar
Excellent, you obviously understand how polar charts work.
It's interesting to see that you assumed that the lead/lag (we call chordwise) move line would be perpendicular to the spanwise (tip weights) move line, and it was. That's almost never the case for gyrocopter rotors.

I'm still curious and ask again:
Are you filtering the vibration to Rrpm frequency or are your measurements total?
Are you measuring pk-pk, RMS or zero-peak?
plus
I understand that this is your electric heliciopter?
These results were on the ground with zero pitch??
Did you have to use the tail rotor to counteract the torque?
The tail rotor is electric right?
Is your horizontal accelerometer orientated left/right or front/back?

Again Bravo, I'm really impressed.
Mike G
 
Oskar
Excellent, you obviously understand how polar charts work.
It's interesting to see that you assumed that the lead/lag (we call chordwise) move line would be perpendicular to the spanwise (tip weights) move line, and it was. That's almost never the case for gyrocopter rotors.

I'm still curious and ask again:
Are you filtering the vibration to Rrpm frequency or are your measurements total?
Are you measuring pk-pk, RMS or zero-peak?
plus
I understand that this is your electric heliciopter?
These results were on the ground with zero pitch??
Did you have to use the tail rotor to counteract the torque?
The tail rotor is electric right?
Is your horizontal accelerometer orientated left/right or front/back?

Again Bravo, I'm really impressed.
Mike G
Mike,

I think I now understand the question about filtering. Correct me if I'm wrong, I suspect the PB4 takes the accelerometer signal and sends it through a band pass filter with the centre frequency at the rotor frequency. I don't do that, I do a fourier transform so no filters involved.

I had a look at the software and realised that I had already converted from an acceleration base to a speed base, and if my calculations are correct the numbers in post #12 need to be divided by 56.9 to get ips zero-peak. That would mean I started at 0.91ips 0-pk and ended up at less than 0.18ips 0-pk. That scaling factor could quite easily be wrong, at the end of the day you don't need the correct scaling factor to balance rotors. I managed to balance the rotors having no clue what the scaling factor was.

Measurements were done on the electric heli with electric tail. Had the heli on the ground so could do measurements without tail thrust.

The horizontal accelerometer was located left/right, will move the accelerometer to see what the front/back measurements are.

Oskar
 
Very interesting work on machine vibration. If the number is correct at 0.18 inches per second (0 to peak), it is still a bit high according to global vibration standards. For those of you interested in vibration, the link below is a sample of machinery vibration limits and an explanation of why VELOCITY of vibration is used, not displacement or acceleration. (Vibration velocity is proportional to dynamic stress of the machine being measured). No connection with this company.


ps, I spent 30 years at Ontario Power Generation Research Division commissioning and, monitoring and maintaining machinery in nuclear and hydraulic power plants.
 

Attachments

  • Applying Examiner.pdf
    286.6 KB · Views: 10
Oskar
Yes Smart Avionics use a filter for the 1/rev for balancing and FFT for the frequency spectrum display.
Agreed you don't really need to know the actual vibration value for balancing (you just proved that ;)) but it's useful to know before to see if balancing is actually necessary and after to know when to stop.

Jerry
Agreed the ideal number that is usually quoted for props and rotors is less than 0.1 IPS. I think this comes from the sort of documentation you quoted but these values are more for large machines with a high machine to rotor mass ratios bolted to solid (or relatively solid) foundations. Because our rotors are at the end of a fairly long slender (hence very flexible) mast which is fixed to a machine floating in the air, we have a much lower effective machine to rotor mass ratio so 0.1 is a good target but not essential in my experience.
I think Oskar has done an exceptional job, and I agree with him that trying to give/sell it to others is fraught with the risk of endless questions.
Mike
 
Vertical vibration is proving to a bit more of a challenge than horizontal.

Firstly the symptoms:

With rotors up to speed and zero collective there is minimal vertical vibration. Pulling collective with the cyclic centred everything stays good.

A cyclic input with up collective, however, results in a bad vertical vibration. Sitting in the seat you don’t need a balancer to tell you that the vibration is bad, but the numbers confirm it. The balancer only measures one per rev, so it’s definitely not a 2 per rev vibration. The phase of the vibration clearly follows the cyclic as shown below.

Rotor vibration monitoring
Any guesses as to what the cause of the vertical vibration could be?
 
Oskar
I'm thinking about it. Most of my experience is with gyros so I may not be much help.

Can you confirm what exactly you are calculating for the phase angle? Is it between the tacho zero and the max vibration in each cycle?
I think your phase angle graph looks more like this :
Rotor vibration monitoring
Interesting that you get more or less a 180° phase change going from forward to Aft and the same for left to right. Usually that means going through a critical speed, I'm not sure what it means here but I think it shows a degree of reliability in your measurements.

The traditional explanation for vertical vibration is (as I'm sure you know) "tracking". This might suggest that your cyclic is not giving the same pitch angle to each blade.

Mike
 
Oskar
I'm thinking about it. Most of my experience is with gyros so I may not be much help.

Can you confirm what exactly you are calculating for the phase angle? Is it between the tacho zero and the max vibration in each cycle?
I think your phase angle graph looks more like this :
View attachment 1160532
Interesting that you get more or less a 180° phase change going from forward to Aft and the same for left to right. Usually that means going through a critical speed, I'm not sure what it means here but I think it shows a degree of reliability in your measurements.

The traditional explanation for vertical vibration is (as I'm sure you know) "tracking". This might suggest that your cyclic is not giving the same pitch angle to each blade.

Mike
Mike,

Your graph of phase angle is correct, nice job extracting that from a jpg picture. I was moving the cyclic by hand so the phases wouldn’t be exactly 90 degrees apart, but the effect is clearly visible.

The balancer has two inputs. The position sensor output is digital and consists of a short pulse as the magnet on the rotor moves past the sensor. The accelerometer output is analogue, and for this discussion I’ll assume it to be sinusoidal.

In the software I arbitrarily chose zero phase to be where the rising edge of the position sensor signal lines up with the zero crossing and rising signal from the accelerometer.
Rotor vibration monitoring
If the signals look like this the phase measurement result will be 90 degrees.
Rotor vibration monitoring
The absolute zero depends on a few things, including:
- Position of the sensor.
- Position of the magnet.
- Where zero phase is chosen in the software.
- Changing from acceleration to velocity base introduces a 90 degree phase shift.
- Measuring delays (for the slow speeds we run at that’s negligible).

Instead of trying to nail down absolute zero phase the easiest is to just install everything arbitrarily and then use a polar plot to find the relationship between the number on the screen and the real world (as was done in post #12).
 
The previous measurements were done on a tethered helicopter, to see how the balancer works in real life I installed it in my first electric heli (which is a Mosquito Air converted to electric) and balanced the MR.

It was a bit of a roundabout journey, first going the wrong direction (point 1 to 2), then overshooting going back (point 2 to 3) and finally ending up where I wanted to be (point 6). The lead/lag adjustment was again extremely sensitive, the last adjustment being 1/12th of a turn. With the polar plot now defined any future balancing adjustment will be easy.
Bal Air.jpg
For the last flight of the day the wind had got up to about 15knots and I decided to throw the machine around a bit (splitting the needles a few times) to see what the recordings would show.

Here are the results in two graphs, there was too much info for one graph. Vibration amplitude is in blue, phase in pink.
1.jpg
2.jpg

There are a few spikes in the vibration level and most of these corresponded to times when the rotor rpm was changing. The most obvious are when spinning up and shutting down the rotor, but also during quick stops when the rotor rpm first increases, then drops, and finally recovers to nominal rpm again. One per rev measurement is defined for constant rpm, if the rpm changes during the sampling interval the measurement becomes meaningless. I monitor changes in rpm and have realised that if the rpm changes too much it’s better to simply discard the data, otherwise it becomes a case of garbage in, garbage out. This applies to gyros as well as helicopters, and is the reason why balance measurements are best done in calm conditions where rrpm does not change much.

There is also a patch with very high readings a few seconds after the first lift off. I had landed again with blades still spinning and cannot remember high levels of vibration. The combination of 15 knots of wind while sitting on the ground with spinning blades was causing something to happen which I don’t understand yet.

Even with all the noise one can still see that the vibration is roughly 50 @ 280° which would require an adjustment of less than 1/12th of a turn on the adjustment screws. With a balancer now permanently installed and data for each flight logged it’s easy to keep track of any changes in vibration.
 
Top