Solenoid Actuator?

Yes it feasable. It requires pricise control of the the solenoid current and timing. Diskclavier works this way. Check them out you may get some hints. Also QRS piano roll company has a piano playing mechanism that works this way, not cheap.You will probably have little luck finding a ready made actuator at any price you can afford, so if you are serious will have to design and build your own. Good luck. Bob

Don't the player piano restoration supply houses have these as a stock item?

How about Allen Organ?

It is used in professional systems, but itsn't all that practical - in particular, with the space available you can't match the accelerations and velocities you get when you strike a piano key, rather than pushing it down.

E-mail me - I've got a handful of references on the subject.

Compumotor has an ironless linear electric motor which is powerful enough and almost compact enough for the job,

but it isn't cheap.

see also

Find an amusement machine vendor in your area who still does pinball machines, and get some drop target reset solenoids. I used to work repairing pinball machines (and juke boxes, Pac-Man, and so on - kewl job!), and was tempted to use them in a piano project of my own, which never materialized. (sigh)

If you've ever played pinball, you've seen drop targets reset - the stroke is maybe 1-1/2", and they don't have a lot of force (they raise those little plastic targets, which latch in place), but by eyeball and gut feeling, they looked like they'd be ideal. You'd have to carefully control the current, to give the attack and sustain that you want, of course.

Good Luck! Rich

The throw on a piano key is about 9mm - 3/8 of an inch.

To simulate the force that is required to accelerate the hammers at a

Modeling the 14-19 moving parts between the key and the hammer seems to be a fairly important part of the job - carefully controlling the current to do that is quite tricky.

Brent Gillespie's Touchback Keyboard project is described in the literature - I posted a couple of URL's earlier in the thread - and level of simulation required seems to eat up quite a lot of computing power.