Amica Ampico Article
By Craig Brougher
|In the March/April '96 Amica Bulletin, an article appeared purporting to compare the Ampico B with previous versions. Some interesting observations were reported in this article which take exception to some of the very things mentioned in its developers' diary, its head of the Ampico Roll Arranging department of the day, bench measurements that anyone can make, and any good high school physics teacher.|
I thought it was strange that Amica's technical committee didn't catch at least a few of these errors, but perhaps, upon learning of them, they will consider rereading the article and perhaps correcting it for archival value, at least.
Authors writing about these instruments have (in my opinion) a responsibility to everyone to check their facts, and frankly, that is exactly what the technical committee should have done and didn't do, apparently because they didn't know enough themselves to object. It's understandable that sometimes an article slips through with a few mistakes, and since nobody is perfect, we all understand that. If the author finds it, it is his obligation to correct himself in a future issue. If he doesn't wish to do so, it then should become the editor's obligation to correct the mistake. But in this particular article, we have so many glaring errors that it might be wondered if the author really understands the Ampico expression system at all.
In the next several issues of AMD, I will try to post one or two of the exceptions taken with the article, and also invite comments. After all, we are all human and prone to making mistakes, so we all know how it feels. Some of the errors I see are in the lack of thoroughness of the statement, which, in my opinion, makes it just as wrong as if it were technically incorrect, because it insinuates a wrong conclusion.
You may not see it that way, or you may see what he is getting at, and that's what makes for good dialog-- as long as we can keep it technical and objective. I refuse to get down in the dirt for a fun subject like this. So let's keep it light, if you don't mind, and not take it too seriously. Here goes:
1. (In attempting to discredit the Ampico model B, fo which he said in
a former article, "The Model B skips notes", Jeffrey Morgan totally
discredits his own understanding of the reproducer in this one
statement) See p 71, Para 3. "Furthermore, the Model B crescendo
system does not initiate corrective action for stack transients as
do most earlier Ampico crescendo systems."
ANS: That couldn't be more wrong! If he doesn't understand the basics of ALL Ampicos, and the reason that the number of notes played are never compensated for on the roll (as in Duo-Art coding) then Jeffrey hasn't done enough homework to be writing articles.
The regulating pouch (or curtain) is a lightweight airtight pneumatic cloth which is placed between the "control" vacuum and each half of the stack. On the stack side of the curtain are two devices, the stack and the supply, which are separated by a grille (grid) having a blank space (no holes) in-between. Normally, without any control vacuum under the curtain, the supply would suck the curtain tight against its grid (a perforated piece of celluloid) and that half of the stack would then be cut off. But there is always a fixed "bias" vacuum called "first intensity" which keeps just enough curtain peeled back to play any or all notes called for at that intensity.
No matter what the demand or "stack transient" due to the number of notes being played, the curtain allows just enough flow to balance the control vacuum under it exactly without momentum or overshoot.
Let's suppose the stack is almost starved (except for the fixed leak which provides a tiny flow of air at all times) and one note is played down. That one note creates a tiny extra demand-- a very small percentage of the capacity of the supply, but is nevertheless sensed by the pouch because of a bit of air tension as the note valve begins to actuate. Air "blows" the regulator down just a very tiny bit-- IN PROPORTION TO the demand, and instantly the supply is uncovered to exactly the same degree because of a constant "weight" on the curtain by the control vacuum which has been set to a certain amount previously. Where will the curtain balance at now? It will try to come to rest at a place in which it can exactly balance the pressure below with the pressure above.
Does it make any difference now when two huge chords are suddenly played down at the 1st intensity? None whatever. The device simply balances by peeling itself away from the supply to the degree that the pressure above the curtain equals the pressure below. And the nice part about this is that there is no built-in error, like having a separate disk regulator valve in line but separate from the control device that is itself also affected by the stack flow but is not compensated for that.
Now by adding other inputs under the regulating pouch, one is able to change the balance point instantly, according to the roll coding, and have a dynamic performance which is faster than the requirements of the music.
The model B "makes its living" compensating almost perfectly for every change in stack pressure, including changes that no other piano can even begin to sense. It is the perfect regulator if there ever was one. So while stack transients (quick changes in pressure due to played notes) is still a problem for every other reproducer ever built, the Model B can totally ignore them because it is actually faster than the valves themselves.
It is incomprehensible that Jeffrey Morgan presumes to teach others that a proportional curtain (or "C" pouch as he calls it) doesn't compensate, and doubly incomprehensible that no one caught this absolutely obvious mistake, since it blows half of his article right out of the water along with it. The next installment will start sinking the other half of his "Observations" into obscurity. You will understand why I say that this article is raving and unsound.
2. p 70, para 2; "Ampico's highly stabilized expression system combined
with its spring loaded stack equalizer pneumatics make the task of
keeping up with normal stack operating demands (especially at soft
playing levels) much less dependent upon expression coding."
ANS: The spring loaded equalizers ( I call winkers) task is not to provide help in keeping up with stack demands. In fact, they actually add a little bit to the demand requirements. They act as eveners or capacitors which are there as a compromise between a little bit less expression capability at low expression levels, and the remote possibility of the regulator valve to overshoot, shutting off vacuum from the stack for a split second. As it turns out, they were really not needed, but historically, we should rebuild and replace them exactly as they were made.
The worry that the regulator valve could clamp the stack supply for a split second and miss a note or two caused Ampico to provide these equalizers that store a little energy in their springs and release it when the tension gets too low, creating just enough vacuum at the stack nipple to oppose the valve closing all the way. They are a "fix" for a possible problem with the regulator valve _at the expense of demand requirements_!
I have also played some sensitive Ampico rolls both with and without the equalizers and prefer it without. Some say they don't hear any difference at all (which doesn't surprise anyone, either ).
The objection I have to the statement is very simply this: Stack demands, caused by notes, chords, and expression from the roll, are compensated in the model A strictly and entirely by feedback from the regulator output to the three intensity pneumatics. Read the Inspector's Manual. Equalizers do not even figure into this, but rather increase, overall, stack demand slightly by storing a part of the energy in their springs, the rest being normal leakage which any pneumatic will have a bit of.
The next exception is a humorous one:
3. p 70, para 3. "Moreover, the rather cumbersome mechanical linkage
shown in this extremely ingenious design may well have impeded its
agility (...fig 8)."
ANS: He speaks of this device prior to this quote as having never seen production. My question is, "What is 'an extremely ingenious, cumbersome design'?-- That was never used in a piano?" Perhaps I was humming the wrong mantra when I read that one, but I don't think anything would have helped much. We used to call them Rube Goldbergs, Jeffrey. Something overly complicated and slow for what it had to do. Ingenuity to me represents the simplest solution to a problem, not the most expensive, difficult, cumbersome, and almost impossible to adjust solution to the problem.
4. p 70, para 1. "During those periods when the crescendo driver is not
engaged by an external signal (he is referring to the Ampico
Crescendo mechanism, but uses his own terms), the action of the
pallet applies inverted feedback from the stack to driver control,
thereby continuously tailoring the strength of the spring pneumatic
to the action of the main regulator valve."
ANS: Does anybody understand that? Why not just say that during periods of crescendo activity the Crescendos maintain equilibrium with their spring tension at the first intensity level by bumping a pallet valve. Does anybody notice something else here as well? Jeffrey says that this "inverted feedback" comes from the stack to the Crescendo. Could anyone simply look at a schematic and inform me via e-mail what he is talking about?
The set point, which Ampico calls the "First Intensity," isn't really inverted feedback in one sense anyway when you look at the entire system, although I will accept it on the grounds that it reverses itself. Feedback, though, is a controlling sample (of output) which usually is returned from the output of a system, back to the input, or at least across two stages or more elements of a system. It is bad terminology here. It would be better called a set point controller.
The supply to the Crescendos comes directly from the pump. There are three other tubes. One is its output to the spring pneumatic, and the other two control slow and fast crescendo/decrescendo and go the the trackerbar. So with that in mind, how does Jeffrey justify saying that any inverted feedback comes to it from the stack? Just curious, since it must be totally isolated from the stack in order to do its job and oppose the three intensity pneumatics whose supply happens to be the stack vacuum.
(Message sent Wed 22 May 1996, 14:19:03 GMT, from time zone GMT.)