Action Ratio Measurement Accuracy

[Bill Ballard]

---------------------- multipart/mixed attachment
At 10:53 PM +0100 10/28/02, Richard Brekne wrote:
>Where's Bill and Jon anyways :) ?

Off doing other things. Family Court, my off'n'on project to import 
the PTx Archives into a database, and pushing a real wonderful 
Steinway Model F (1883) vertical out the door. Or were you asking 
about the other BB? Excuse me, what was the question?

At 4:19 PM -0500 10/28/02, Farrell wrote:
>On average LAR values were about 6% less than MAR values, and CTAR 
>was about 2% less than MAR.

That's a half full glass. What's interesting is the match between the 
MAR and the CTAR. I'm no statisticificator, but I'd guess that the 2% 
disagreement is actually within the statistcal margin of error 
considering that they are on either side of the dimensional line. MAR 
is linear, and CTAR is static weight.

Each of these has its own uncertainties, its inherent problems. Given 
those, I'd say that the 2% difference is close enough to be a 
handshake. Thank you, Terry for provide the "triangulation" on this 
subject, with your three measures of an action.


Farrell wrote:
>I'm disappointed they are not all closer, although it is easy for me 
>to imagine why the LAR might be inaccurate - it is very hard to 
>identify the exact point of contact between the capstan and wippen 
>heel and to a lesser extent, the knuckle/rep lever contact - indeed 
>these contacts move (I measured at about 1/2 blow position).

These are moving points as you suggest, as well as difficult points 
to define, given that they're buried underneath felt or leather. I 
need to build a small mirror stand to make all of this more easily 
readable, but it all requires plenty of line-of-sight from the side 
of the parts.

Farrell wrote:
>To a small extent, the CTAR value will depend on friction as it 
>relates to one's ability to get a real good up and down weight - as 
>well as identifying exactly where the contact point is on the wippen 
>heel when measuring Wippen Radius Weight.

Difficulties like this cloud the measurements. A magnifying mirror 
would help, or a flourescent crane lamp with magnifier.

Friction is a strange player in all of this. Friction is supposed to 
be equal in either direction, and in the best pianos it is. But the 
business of "friction gradients" skews our observations. Our only way 
of avoiding them is to allow the parts to  move at least as fast as 
not to be overcome by the gradients. They no longer need to be 
creeping along at speeds close to a stall, all that is asked is that 
the Down Weight and Up Weight are measured at the slowest speed at 
which they can match (like modems). As they might say in medical 
research, not enough is understood about friction gradients.

Farrell wrote:
>As long as one has real firm and steady measuring platforms for 
>precise measurements, and has taken all the slop out of the action 
>(firm rep lever, etc.) I should think the mechanical or manual 
>method of direct action ratio would be the most accurate.

Again, I think the two are roughly equivalent. The MAR should really 
be an angular measurement as I mentioned in August. I would choose 
mine on the basis of what it was that we wanted to regulate, to set 
up. Usually, the results of this set-up ("action hanging") are 
experienced and described by the pianist in terms of weight or 
resistance. If I had a nickle for every pianist who said that the 
hammers were sitting too far from the strings or that the keys went 
down too far, and a nickle for every pianist who described it in 
terms of resistance ("flies away from you", "drives like a truck"), 
and if I had to pick between the two piles of nickles, I'd pick the 
pile accrued by the latter. (I'd be much richer.)

At 10:53 PM +0100 10/28/02, Richard Brekne wrote:
>The Stanwood is the best for key to key ratio analysys. Problem is 
>that it will show you exactly what the ratio is for each and every 
>key....including any and all small ratio problems. As Ron Overs site 
>states..... it doesnt take much of an angle in knuckle placement to 
>make a big difference in the resulting ratio. You need a lot of 
>samples to get an average that you can reasonably assume is close to 
>the true Ratio.

Eight note samples are a good quick look. I'm much happier with 
31-note samples (C,C#,E,G, plus the nats at the ends with a A#86 to 
round it off). When I actually get to doing the work, it's on the 
basis of a 100% sample.

At 10:53 PM +0100 10/28/02, Richard Brekne wrote:
>I like to use what you call the Manual Action Ratio to get a basic 
>view of what the design was meant to be. I personally use 5 mm key 
>movement and I use a block under the key to insure that.  I always 
>get more consistant results using several samples.

Again, I'd be interested to see this measurement converted to 
angular. Just to confirm that it wouldn't make a difference.

At 10:53 PM +0100 10/28/02, Richard Brekne wrote:
>I could never get consistant results with measureing distances 
>needed using Rons Overs method or the similiar one found in various 
>other places. Probably a personality fault :) Tho conceptually this 
>should give you the same kind of general ratio information as your 
>"MAR".

I think this is where the appearance of linear distance being moved 
will actually lose something due to the vector loss occurring when 
lever arms move down from the horizontal.

Attached are three graphs of the different readings. (This is a 56K 
.jpeg. PDFs are available, but as Kent Swafford points out, 
considerably bulkier.) Something happend to the CTAR on #31. These 
graphs call for a larger data set.

Bill Ballard RPT
NH Chapter, P.T.G.

"May you work on interesting pianos."
     ...........Ancient Chinese Proverb
+++++++++++++++++++++


---------------------- multipart/mixed attachment
A non-text attachment was scrubbed...
Name: 3-pane.jpg
Type: image/jpeg
Size: 74398 bytes
Desc: not available
Url : https://www.moypiano.com/ptg/pianotech.php/attachments/61/fa/bf/f4/3-pane.jpg

---------------------- multipart/mixed attachment--