Measuring RPM From Video

Arborist Forum

Help Support Arborist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Chris sometimes BS is just easier to except than actual facts. Most have been conditioned from day one to believe complete and utter BS, so really I expect nothing else.

Sent from my HTC One_M8 using Tapatalk
Andy is right on.

I find that when ever the term "********" is stated on this site (and in this case it is always written out completely) , the only person full of BS is the one who made the unsubstantiated rebuttal.

As another example of putting time into testing, think of all of the work that redbull660 has put into his oil tests yet some of these armchair QB's completely try to tear him apart regardless of his hard work. It is downright ridiculous.
 
Andy is right on.

I find that when ever the term "********" is stated on this site (and in this case it is always written out completely) , the only person full of BS is the one who made the unsubstantiated rebuttal.

As another example of putting time into testing, think of all of the work that redbull660 has put into his oil tests yet some of these armchair QB's completely try to tear him apart regardless of his hard work. It is downright ridiculous.

you wanna fight about it?
 
I thought this one was interesting - I was cutting a small ash I dropped along the edge of our lawn while my wife was doing a little cutting with the WheelHorse. When I looked at the spectrum plot I got both:

Poulan 42cc with Kohler M12-Notes .png
 
Andy is right on.

I find that when ever the term "********" is stated on this site (and in this case it is always written out completely) , the only person full of BS is the one who made the unsubstantiated rebuttal.

As another example of putting time into testing, think of all of the work that redbull660 has put into his oil tests yet some of these armchair QB's completely try to tear him apart regardless of his hard work. It is downright ridiculous.

Those types think that they know everything and when someone shows them something they don't understand because they have never used it or someone knows more about it than them they get defensive. They are unwilling to learn and run people into ground just because they themselves won't try to understand something that may be very beneficial for them to use. Think about how many dealers are getting left behind because they don't want to learn autotune.

Chris I like what you are showing us, it has far more uses than just measuring rpm, it can reveal problems before we actually start to notice them.
 
Those types think that they know everything and when someone shows them something they don't understand because they have never used it or someone knows more about it than them they get defensive. They are unwilling to learn and run people into ground just because they themselves won't try to understand something that may be very beneficial for them to use. Think about how many dealers are getting left behind because they don't want to learn autotune.

Chris I like what you are showing us, it has far more uses than just measuring rpm, it can reveal problems before we actually start to notice them.
Just think. A few years ago the thought of things like finger ports or advancing timing was also "hocus pocus" lol.
 
I'm not sure what you mean by anti-node - those areas between the harmonics are just the background sound level at those frequencies, it's not driven lower by being between the spikes. It looks to me like the 2nd harmonic is some 18dB down from the fundamental, and none of the others is higher either - and in this context I see no real significance to it anyway. I'm not sure what you think they mean? They just look like a typical harmonic response of a system driven at a frequency.

Yes, but it just makes it harder to read, which is why a log scale is normally used for such plots.
The increased power seen for the higher harmonics might merely be due to the phone's mic being more receptive to these freqs. The mic and recorder design will be very compromised, and not particularly linear in it's response I'd imagine.
 
Attached is a short video clip showing 2.5s of audio track spectrum analysis from 4-stroking chainsaw. Each frame looks at 380ms of data.

The saw was someone else's Echo 590 with a muffler mod that was pig rich. It's 4-stroking like crazy until it get to the widest part of the log, then it stops.

Starting at 6.5s the saw is very rich and 4-stroking. At 9s the saw has reached the widest part of the cut. As the wood gets wider the load gets higher and the rpm drops, and at 8s the 4-stroking stops.

After it begins firing every stroke the rpm rises, but not enough to initiate 4-stroking again.

In the first frame the saw is running at 8950rpm (149Hz), but there is a smaller spike at 1/2rpm (75Hz) that is the 4-stroking signature. It disappears as the 4-stroking stops.

 
Attached is a short video clip showing 2.5s of audio track spectrum analysis from 4-stroking chainsaw. Each frame looks at 380ms of data.

The saw was someone else's Echo 590 with a muffler mod that was pig rich. It's 4-stroking like crazy until it get to the widest part of the log, then it stops.

Starting at 6.5s the saw is very rich and 4-stroking. At 9s the saw has reached the widest part of the cut. As the wood gets wider the load gets higher and the rpm drops, and at 8s the 4-stroking stops.

After it begins firing every stroke the rpm rises, but not enough to initiate 4-stroking again.

In the first frame the saw is running at 8950rpm (149Hz), but there is a smaller spike at 1/2rpm (75Hz) that is the 4-stroking signature. It disappears as the 4-stroking stops.


That's a really nice demo Chris. I'm really grateful for you enlightening/reminding me that we can use Audacity to confirm some of ears experiences. Since I'm not great at tuning by ear (but I'm improving a lot with help from a lot of the people here) I've found that using Audacity plots is helping me a lot to post-mortem what I thought I'd heard.

I replayed your vid once or twice and noted that your mic samples at 44.1khz, your DFT is consuming 16380 samples spaced over 0.38 second intervals. Did you deliberately choose these sample sizes and intervals to optimise the analysis? I ask this question from some background in DSP stuff, and from recent experiences using Audacity to inspect saw clips where I've observed that having too low a sample size will reduce peak definition whereas too high (I think) results (for me at least) in more noisy stuff (i.e. random spikes) appearing in the plot. Clearly some forethought is required to make best use of the data and processing algorithm, and I wondered if you had any views on this subject.
 
That's a really nice demo Chris. I'm really grateful for you enlightening/reminding me that we can use Audacity to confirm some of ears experiences. Since I'm not great at tuning by ear (but I'm improving a lot with help from a lot of the people here) I've found that using Audacity plots is helping me a lot to post-mortem what I thought I'd heard.

I replayed your vid once or twice and noted that your mic samples at 44.1khz, your DFT is consuming 16380 samples spaced over 0.38 second intervals. Did you deliberately choose these sample sizes and intervals to optimise the analysis? I ask this question from some background in DSP stuff, and from recent experiences using Audacity to inspect saw clips where I've observed that having too low a sample size will reduce peak definition whereas too high (I think) results (for me at least) in more noisy stuff (i.e. random spikes) appearing in the plot. Clearly some forethought is required to make best use of the data and processing algorithm, and I wondered if you had any views on this subject.
Matt, the parameters were not that scientifically chosen - I wanted a short sample time span so I could see the progression of the 4-stroking over a couple of second section of the video. Using the 64k and 32k sample FFT would not let me get down as short as I wanted, and in playing around with the 16k transform it wanted at least 380ms before it would plot. So that's what I used. 380ms is faster than the 500ms update rate of most real digital tachs.

Usually when I am looking at rpm I use 64k samples and a 1.5s interval, but that can be too long when you are looking for short events. Then you end up mixing data from before and/or after the event you are interested in, and things just get all smeared.
 
I use 64k samples and a 1.5s interval
Indeed for your mic 65536 samples (i.e. 64k) will take

65536/44100 = 1.48 seconds (i.e. ~1.5) to accumulate. So you were being scientific - intentional or not!

images.jpg

I'm not sure how Audacity implements a largely selected interval than it has adequately frames; a decent FFT will just use zero samples to bring the sample size up to the next power of 2. And yeah, if you include too big an interval it's easy to overlap into a period where the saw's running characteristics have changed, i.e. due to load or the bar angle being changed.
 
Back
Top