[pianotech] Action Ratio measuring

[David Love]

David S.:

 

Here are some data worth looking at, perhaps.  The top lines are
measurements taken on my action model (Renner) using your system and the
bottom in color is my system of measuring the lever.  As you can see the
results are very close.  When adding some strike weight to the hammer, in
this case 2.2 grams, the balance weight changed at a ratio of 5.68
((83.5-71)/2.2 ). Pretty close considering the potential for measurement
error with measuring upweight and downweight.  But at least as I measure it,
it doesn't agree with the SWR exactly, or my distance ratio for that matter
(though that one appears to be a bit closer). Measurement protocols are
clearly a factor.

 

Which, for me, is really the issue.  On this model I've taken great pains to
ensure that each measurement is very accurate, having taken it several
times.  Even so, there is even a discrepancy between the two SWR
measurements with different strikeweights.  Yes, of course you can take an
average but when I've had individuals send me data in order to help them
with weight problems I've found that the SWRs are all over the place with
quite a wide range.   Totally unreliable may have been a bit strong, but I
have found it certainly much less reliable (and more time consuming) than
the distance method.  Some  of that may indeed be because of non straight
balance rail lines or capstan placement, knuckle placement etc.  However,
much of it, I'm convinced, is due to the difficulty in the protocols for
measuring upweight and downweight.  Just how fast should the movement of the
hammer be, how far should it travel, should the upstroke be the same speed
as the downstroke, faster, slower?  If I'm asking these questions you can be
sure that different people are taking measurements in different ways
yielding unpredictable results.  

 

Distance and weight ratios may be different as you are looking at them, I
don't know.  I'd be interested to hear the specifics. Typically they are
referred to as distance and force ratios and there is definitely a
relationship as far as I understand it.  The force leverage is calculated as
1 divided by the distance ratio and predicts how much weight is required to
balance a hammer of x grams.  So in the example below (mine) the 5.78 DR
would yield a FR of .17 which means that an 8.7 strike weight will balance
at 50 grams at the key (8.7/50 = .17.)  I would be interested to hear the
argument for gravity vectors impacting this, it's clearly outside the range
of my engineering chops.  

 

However, if the issue is predicting front weights by formula then
establishing a working ratio is clearly important but the distance method
achieves that goal.  Since you are using an average from a sampling which
smoothes out errors either in measurement or execution, I presume that a few
percentage points difference is not critical to the outcome-there will be
built in errors along the way anyway.  

 

As a practical matter, at least for me, I have found that measuring the
action ratio by distance, either by product of levers or direct measurement,
is more reliable than taking sample weights and then determining the SWR.
Once the action ratio is determined, and sampling is useful as executions on
actions can be faulty, then the front weight target and strike weight
targets can be determined using a bit of algebra and your equation of
balance (thank you for that).  At that point I believe (correct me if I'm
wrong) your method is to use that as a guide to preset a smooth front weight
curve.  Others, like me, prefer to use those calculations as a guide to
ensure that the front weights fall in a specific range (not to heavy or too
light) and then set each note individually by uniform balance weight.
Because of inevitable small variations in the note to note executions
through the action (balance rail, capstan, knuckle, etc) I find that if a
smooth front weight is attempted that the balance weight will need to float
(assuming you have a smooth strike weight curve), something I believe you
mentioned in a post some time ago that you allow.  I prefer that the front
weight float a bit as the inertia differences owing to small differences in
front weights are small in the whole scheme of things compared to the
floating upweight or downweight curve that will happen if you try and smooth
the front weights. Sadly, in practice there is enough error in the
executions that you can't have it both ways, at least that's what I find. 

 

As far as the relationship between your SWR and the distance ratio and that
DR/WR > 1, I can't really comment as to why that would be or what that means
exactly.  I take you at your word that you have found examples of that
relationship as you are determining it and that they have seemed meaningful
in terms of action performance.  But since I don't know your method of
measuring distance ratio or the exact way in which you determine SWR re UW
and DW measurements or what and how you are assessing "the best action
setups" it's hard for me to comment.   I'm interested though! 

 

 


U

D

SW

FW

KR

BW

F

	SWR

		
	


61 

81

12.4 

10.0 

0.50 

71

10.0 

	5.81

		

71 

96

14.6 

10.0 

0.50 

83.5

12.5 

	5.79

		

 

 

	
Ea

Ra

	Ratio

		

Key

260

128

	0.49

		

Wippen

63

99

	1.57

	Action Ratio


Shank

17

127

	7.47

	5.78

 

 

 

 

David Love

www.davidlovepianos.com

 

 

-----Original Message-----
From: pianotech-bounces at ptg.org [mailto:pianotech-bounces at ptg.org] On Behalf
Of David Stanwood
Sent: Sunday, September 23, 2012 3:55 AM
To: pianotech at ptg.org
Subject: Re: [pianotech] Action Ratio measuring

 

>Date: Fri, 21 Sep 2012 07:56:11 -0700

>From: "David Love" < <mailto:davidlovepianos at comcast.net>
davidlovepianos at comcast.net>

 

>BTW just one added comment.  Calculating the action ratio by weight, as 

>in the Stanwood SWR method, I have found to be totally unreliable as 

>there are simply too many opportunities for measurement error.

 

>David Love

 

Dear David Love et al,

 

I've been trying to understand your statement: "Calculating the 

action ratio by weight".   Do you mean calculating the distance ratio 

by the weight method?  If so I think it is important to distinguish between
distance and weight ratios.  One is influenced by gravity vectors and the
other not.  So they are not necessarily the same.  I stand by my opinion
published in the PTG Journal that the best action setups exhibit a distance
ratio that is equal to or higher than the weight ratio, the higher the
better.  This relationship is effected by the set up of the arc geometry, an
important and deep subject indeed.

 

I defend the calculation of SWR or Strike Weight Ratio as a viable and
reasonably repeatable method.  Those of us who practice Component Touch
Weight balancing know that the determination of the SWR or Strike Weight
Ratio level is an essential ingredient for calculating Front Weight
Specifications.  The technique has been practiced and continues to be a
highly valued tool by many in our profession. 

Those of us who utilize this method know that calculating Strike Weight
Ratio on a single note and drawing a conclusion about the action as a whole
is, in your words: "totally unreliable".  However, as soon as a greater
number of measures are taken and averages looked at, the information becomes
more and more reliable and totally useful.

 

I suggest the same statistical approach when drawing conclusions about the
calculation or measurement of distance ratios.  Best to measure sample notes
across the keyboard and look at the average level of the action as a whole.
There are plenty of opportunities for error in measuring between distance
points as we see by the 

discussion and questions.   Also there are naturally variations in 

action ratio due simply to unevenness in the construction of action
components.  Balance Rail pin lines are not straight, capstan lines vary a
lot, and when you look closely at knuckle placement you will 

find there is plenty of opportunity for variation.   Add this to 

measuring errors and we have a good formula for cumulative error or
variation which result in action ratios that uneven by varying degrees from
note to note.  Statistical method is the only way to avoid or minimize false
perception and this means taking enough samples to draw real conclusions.

 

In my work and consulting we use three levels of sampling:

 

Minimal = middle c2,c#2c4,c#4,c6,c#6 or notes 16,17,40,41,64,65  for quick
evaluation

Partial = all the c, c#s    for determination of levels

Full    = all 88 notes      For StrikeWt calibration and 

documentation of the action "as was"

 

Hope this helps.

 

David Stanwood

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.moypiano.com/ptg/pianotech.php/attachments/20120923/2d1e6439/attachment-0001.htm>