Measure your own break speed

[Runnin8]

I posted this in the "cue balance point" thread, but thought I'd post a new thread that is more on topic. The best I did at the time was slightly over 24 mph. I'll have to re-test again and see if I've improved. I'd be interested in tracking the results.

Here's the wavepad link: http://www.nch.com.au/wavepad/masters.html

My son and I did his science experiment to find the relationship between break speed and a measure of ball spread. To find your own break speed 1) measure the distance between where your cueball rests on the table for a normal break and the headspot, subtract 2.25" for one ball diameter, 2) use sound recorder on a PC to record the sound of your break. We used a laptop with it next to the table. Most PCs can record at 44Khz, or 44 thousand times a second, more than accurate enough. Also, forget about having to get it right in the middle. Unless your break speed is near the speed of sound its a nit. Use a sound program like Wavepad to open the file and identify the peak where the cue strikes the cueball and then the peak at rack contact. If you highlight with Wavepad it will tell you to the thousandth of a second. 3) Your break speed in mph is

(distance-2.25)/12*60/88/(measured time)

Here's his result from 150 breaks along with an example of determining the time increment.

 

[Johnnyt]

I posted this in the "cue balance point" thread, but thought I'd post a new thread that is more on topic. The best I did at the time was slightly over 24 mph. I'll have to re-test again and see if I've improved. I'd be interested in tracking the results.

Here's the wavepad link: http://www.nch.com.au/wavepad/masters.html

My son and I did his science experiment to find the relationship between break speed and a measure of ball spread. To find your own break speed 1) measure the distance between where your cueball rests on the table for a normal break and the headspot, subtract 2.25" for one ball diameter, 2) use sound recorder on a PC to record the sound of your break. We used a laptop with it next to the table. Most PCs can record at 44Khz, or 44 thousand times a second, more than accurate enough. Also, forget about having to get it right in the middle. Unless your break speed is near the speed of sound its a nit. Use a sound program like Wavepad to open the file and identify the peak where the cue strikes the cueball and then the peak at rack contact. If you highlight with Wavepad it will tell you to the thousandth of a second. 3) Your break speed in mph is

(distance-2.25)/12*60/88/(measured time)

Here's his result from 150 breaks along with an example of determining the time increment.

Very cool. Thanks for sharing that. Johnnyt

 

[SPINDOKTOR]

I break at a consistant 24-25 mph, I can break with more speed but my accuracy fails dramaticaly. Useualy with a good rack and fast cloth I will reduce the speed to as low as 17mph for consistant breaks. Speed really isnt relative as long as the equipment is up to par. Naturaly on super slow cloth or dead rails your break needs to be much harder to attain a runable rack consistantly. So I dont put much faith in the speed of the break rather its the accuracy I try to attain. Break only as hard as your acurate.
I also teach this in pocketing balls, It makes no sense to send the cue ball screeming if you cant have pinpoint acuracy.




SPINDOKTOR

 

[jon21588]

Well I tried it out tonight. Its pretty cool. I recorded two friend's breaks and two of mine. One friend hit 18 and the other hit 25. I hit 27 and 33. I'm not sure if this is exactly accurate, because there was music playing in the background. I'm going to try it out tomorrow when there is no one in the pool hall. I'm also going to borrow my friend's j&j break cue to try it. I was using a sneaky pete with a medium Moori(sp) tip. Thanks for the info. Rep to you, sir.

 

[Runnin8]

Well I tried it out tonight. Its pretty cool. I recorded two friend's breaks and two of mine. One friend hit 18 and the other hit 25. I hit 27 and 33. I'm not sure if this is exactly accurate, because there was music playing in the background. I'm going to try it out tomorrow when there is no one in the pool hall. I'm also going to borrow my friend's j&j break cue to try it. I was using a sneaky pete with a medium Moori(sp) tip. Thanks for the info. Rep to you, sir.

At 27 and 33 you are generating much more power than I am- congrats. I re-tested tonight and am still consistently around 24.

I'm curious how the break cue works out- the phenolic will have less energy loss at impact than a leather tip but will be in contact for a shorter amount of time. It's the trade versus less efficient for more time to generate speed. Sounds like a job for Dr. Dave!

Let me know how it turns out.

Steve

 

[jon21588]

I certainly will. I'll probably do 5 tests with each cue and see how it turns out. I'm going to get the paper reinforcements to spot the head ball and the cue ball to make sure they are exactly the same distance apart every time. I'll update here on this thread.

 

[mosconiac]

Nice experiment! I'm curious why you had so many very low speed breaks (~10mph or less)? Did you intentionally do this to test under a wide range of speeds to gauge the quality of breaks even at low speed?

BTW, how did you achieve a break speed of 0?

 

[Runnin8]

Nice experiment! I'm curious why you had so many very low speed breaks (~10mph or less)? Did you intentionally do this to test under a wide range of speeds to gauge the quality of breaks even at low speed?

BTW, how did you achieve a break speed of 0?

We were looking for the full curve of spread versus break speed so we intentionally hit some very soft breaks. Essentially, on an unbounded table the balls would spread further the harder you break, just physics. But with the rails pushing the balls back in play there is a maximum spread achievable that can be computed for any number of balls. In playing others, I noticed that often they spread the balls great without great power and we looked into it. If you look at the curve it flattens out pretty early because the balls are hitting rails and bouncing back in play. I had hoped to get Hillbilly to crack some for us as data points, but when we were doing the experiment he was out of town. I think that at higher break speeds you'll only get a slight increasing slope because of increased likelihood of sinking more. One weekend we'll do the full experiment up to 35 mph (with Charlie's help) and I'll post it.

The zero break speed was a calibration datapoint for the camera analysis. We videoed each break and used video analysis software to measure the location of each ball, calculate the distances, and then determine mean minimum distance between balls (the average of each balls distance to its closest neighbor). With zero speed, i.e., undisturbed rack, we came up with 2.2" with is only 0.05" from the what it really is- 2.25 or one ball diameter.

I plan on polling the community to help come up with some consensus on how we can score "best" break numerically taking into account clusters, cueball control, etc, and then re-reduce the data. What it will show is what most conventional wisdom already reports- that beyond a certain speed there is little to gain from hitting them harder.

Thanks,
Steve

 

[Jude Rosenstock]

I just wanted to comment on how brilliant an idea it was to use sound as the unit of measure for determining break speed. It's such an obvious and cheap solution. In fact, I really wonder if it could produce more accurate results than a radar gun.

 

[Bob Jewett]

... Use a sound program like Wavepad to open the file and identify the peak where the cue strikes the cueball and then the peak at rack contact. If you highlight with Wavepad it will tell you to the thousandth of a second. 3) ...

In the sonogram you posted, there are two short sounds at the start and then a long sound which I assume is the rack breaking and lots of balls hitting each other. Are there really two sounds at the start? They look to be maybe 20 milliseconds apart and roughly equal amplitude and duration.

 

[Jude Rosenstock]

In the sonogram you posted, there are two short sounds at the start and then a long sound which I assume is the rack breaking and lots of balls hitting each other. Are there really two sounds at the start? They look to be maybe 20 milliseconds apart and roughly equal amplitude and duration.

It could easily be something else the shooter did. If all of the samples had that pattern, I'd be curious to find out what it was exactly.

 

[DaveK]

I just wanted to comment on how brilliant an idea it was to use sound as the unit of measure for determining break speed. It's such an obvious and cheap solution. In fact, I really wonder if it could produce more accurate results than a radar gun.

This technique was discussed in threads a couple of years back, and yes, it it a good idea. Regarding the accuracy of the results, the method seems good as far as the distance measurement goes (not a big problem with an error of +- 0.25" ) but the question would be what is the error in the time measurement ... something PC's have not been great at, at least to the resolution required. If the PC method can be trusted to be +- 0.015sec, then the speed could be calculated +- 2mph according to by very brief analysis.

Dave

 

[juggler314]

This technique was discussed in threads a couple of years back, and yes, it it a good idea. Regarding the accuracy of the results, the method seems good as far as the distance measurement goes (not a big problem with an error of +- 0.25" ) but the question would be what is the error in the time measurement ... something PC's have not been great at, at least to the resolution required. If the PC method can be trusted to be +- 0.015sec, then the speed could be calculated +- 2mph according to by very brief analysis.

Dave

I'm pretty sure the timing is MUCH more accurate than +-.015 seconds. The Analog to Digital converter will use the system clock which pulses on the order of nanoseconds. Even for cheapo consumer gear you'd expect jitter to be no more than 100 nanoseconds. So your individual 1/44100 second time slices shouldn't be off by more than 100 nanoseconds (1/10th of a microsecond) I didn't do the math - but I'm pretty sure that's way under 2mph - more like "negligible".

Using this method your results should be far more accurate than your likely error in taking the distance between the cue and the rack (keep in mind you'll probably be taking the shortest path distance, but if you don't hit dead on - which you probably wont if you are trying to max out your break speed - that path will be slightly longer - this slight difference will account for well over 100 nanoseconds of time - thus the sound card error is way smaller and can be considered negligible).

The errors in choosing exactly what part of the waveform corrosponds to hitting the cue and hitting the rack will likely also overshadow any timing errors in the computer by at least an order of magnitude.

If we take the distance measurement to be the most error prone, then our speed measurement shouldn't be off by more than say 1/4 of an inch over what is it 1/2 the length of the table for a head on break from the spot = 4.5 feet. So .25/54 = 4/10th of a percent which on a 30mph calculation means an error of less than 1/10th of a mile per hour enough for bragging rights.

As far as radar guns go - if you have it pointed in a direct line with the cue ball path the result will be more accurate, but the difference would be pretty minimal - enough that it's not worth even trying to borrow a radar gun...

 

[Runnin8]

In the sonogram you posted, there are two short sounds at the start and then a long sound which I assume is the rack breaking and lots of balls hitting each other. Are there really two sounds at the start? They look to be maybe 20 milliseconds apart and roughly equal amplitude and duration.

Sorry Bob, that probably wasn't the best choice to use as illustration, I just had it available. The mic will pick up a stick slapping the table, which I think this was- it was one of my son's breaks. Also, you can clearly see cueball bounces in the trace and based on % of time passing from break to hit, calculate the bounce distance.

It could easily be something else the shooter did. If all of the samples had that pattern, I'd be curious to find out what it was exactly.

Here is a composite of many breaks- two things are consistent the initial hit and the rack hit. In between you can see occaisional peaks due to bounce, like upper right highlighted break. I certainly found that it worked out well.

but the question would be what is the error in the time measurement ... something PC's have not been great at, at least to the resolution required. If the PC method can be trusted to be +- 0.015sec, then the speed could be calculated +- 2mph according to by very brief analysis.

Really, the accuracy of a PC's clock doesn't impact the accuracy of the technique because we're not using absolute time. PC clocks drift but not within the duration of the recording. What we are taking advantage of is the frequency of the analog to digital conversion in the sound card sampling at 44Khz, which is very tight. Very mature technology. Juggler nailed it.

Also, here's the technique for using a camera to get ball positions for the math minded.

 

[Tim5000]

Another interesting use of this sound technology...

Everyone has seen the still photo of a bullitt coming out of the barrel of a gun? They mic the gun, then the sound of the shot firing triggers the camera.

 

[DaveK]

I was curious about the resolution of the entire method, and commenting based on the application. As an example of the errors that need to be understood : what is the delta-time for 1-pixel-over selections on the screen at your given screen resolution ? (OK, I can figure that out myself) Is the waveform clear to the extent that you can decide where the impact was to 1 pixel on the screen ? If not, roughly how many pixels ?

Dave

 

[jsp]

...My son and I did his science experiment to find the relationship between break speed and a measure of ball spread...

Shame on you for doing your son's science fair project. Totally unfair for the other kids who are bringing models of solar systems and digestive systems.

Just kidding. Nice thread. I'll try it out sometime.

 

[Solartje]

i LOVE it brilliant idea. especally checking the quality. now just add cueball controle factor, and you probably have the 3 main things of the perfect speed to break at.

reps for you sir!

probably most interesting post i read in a long time.

so best speed would be around 13mph? slowest speed to still get a high quality. everything between 13 and 20 gives equal averages, and 20+ is just SO hard to controle, its not worth it.

ps did you keep any statistics on the amount of balls potted? (even if that is really in function of the breaker... much more then the speed, but still.. might give interesting results

ps2 to bad we dont have those camera at home to record quality settings. if everyone contributes , you would have 10.000 records in no time and a very acurate average to draw conclussions from, or could we do it ourselves without the need of expensive equipment? (just take a picture with any digital camera and upload in on a program?)

use it during the 1000 9ball games guiness world records

 

[DaveK]

Another interesting use of this sound technology...

Everyone has seen the still photo of a bullitt coming out of the barrel of a gun? They mic the gun, then the sound of the shot firing triggers the camera.

A good friend has a golf 'launch monitor'. It takes two pictures of the ball in flight very close together in time and is triggered by the sound of the clubface hitting the ball. It has a 'slow' setting and I want to try this out on my pool table to measure spin rates (but I'm thinking the golf machine is still too fast).

Dave

 

[Runnin8]

I was curious about the resolution of the entire method, and commenting based on the application. As an example of the errors that need to be understood : what is the delta-time for 1-pixel-over selections on the screen at your given screen resolution ? (OK, I can figure that out myself) Is the waveform clear to the extent that you can decide where the impact was to 1 pixel on the screen ? If not, roughly how many pixels ?

Dave

Ah yes, now we're getting to the soft stuff. Pixel clarity on the signal isn't really the error driver since we can zoom to as much detail as we want in the software, but once zoomed how do we pick out the collision. Truth is, we're dealing with room acoustics and what we are measuring includes echos off walls etc- thats the bad news. The same effect impacts both hits however with room acoustics being roughly unchanged- assume echo effects are similar- thats good. If we handle each collision the same, say measuring off the first jump in signal, which should represent the shortest distance sound had to travel, then we can minimize the error. An alternative is to perform some high order filtering. Certainly a place appropriate for error analysis.

Any audio engineers in the house?

Given the technique I certainly wouldn't pay attention to the second decimal place and would kind of wink at the first decimal. What we really want is a tool to help us baseline our results against the top players, provide some measure of consistency in our game, and quantitatively measure progress. For me, I'll round to the nearest integer and be satisfied.

Steve