I was already aware, after searching around the net, of the different opportunities for a power hammer builder; my project starts from the air cylinders I've found on ebay: they seemed cheap enough to push me in the direction of an air hammer; this was not the only reason, of coarse, in fact among the other features the single blow was something I wanted. I'm not that expert and prefer to work slowly, at my own pace, and I saw that mechanical hammers are hard to make so sensitive on the clutch to get single blows.
Now, with my new pair of cylinders at hand, I'm designing the rest of the machine around them.
First, I must say that since the cylinders are quite short for the usual set up (inline mount of the hammer head on the shaft), I'm using a parallelogram suspended hammer head with no slides (the head weights around 30Kg, the die nearly 5; the total is approximately around 70lbs).
This solution makes it possible to avoid precision machining (no sliding parts), to multiply head speed using a simple leverage, to get a deep throat under the hammer (some 12 inches), to get a reasonable jaw spacing (10 inches with only 3.3 inches of cylinder stroke) and also it keeps the hammer always parallel to the anvil during the stroke.
The coupling of the cylinders on a common pivoting base is working well in my tests and shows no sign of vibration or distortion, so I think the two are working at (grossly) the same speed with no need for any tuning.
The lever and a rubber dampener (1/8 inches) provide a shock absorbing effect to the shaft of the pistons to protect them.
For maximum power on the downstroke I use very short and pretty large tubes on one side ( 1/2 inch ); on the upstroke side this is not required since the air exits through a couple of fast valves applied directly to the cylinders.
Initially I was biased toward a lighter hammer head (in the 40lbs range) to reduce stresses on the hammer structure and on my floor (which is not intended for such an industrial use), but I couldn't find a suitable piece of steel in my junkyard explorations; the same happened for the anvil block, nearly impossible to find of the appropriate size and shape; in the end the two are pretty well matched but really on the heavy side in respect to the size and demultiplying of the cylinders.
The need for a lever action rises from a couple of calculations on the stroke, speed and force of the cylinders under different load conditions (with and without lever).
The results were that, provided the nominal force and no lever, the acceleration of the head would be some 80gees, the final speed of the stroke quite high, the energy of the impact provided only by the air action...sounds good, doesn't it?
Thinking all this over (and after a very poorly conducted speed test on the air cylinder with NO load) the bad news came out: the cylinder cannot run that fast, not even without any load!
A rough estimate of the average speed was 1-2 m/s, way slower than the calculated one. This happens because the air in the tubes cannot travel too fast, and since the tubes are much narrower than the cylinder(s) the air in them has to travel n times faster than the cylinder itself (where n is the cylinder/tube section ratio, that can be well over 10). On a long term basis one should also note that very high speed of the piston could wear the seals faster, shortening the life of the system.
The lever overcomes these problems applying a conversion factor of 3:1, so given a cylinder speed of 2 m/s I could reach a head speed of 6 m/s; in the end I have more power from the cylinders since they work slower (hence the pressure inside them stays higher througout the whole stroke because the air speed in the tubes is lower) and have a larger contribution of the gravity force because of the longer fall of the head.
For an easy movement and a better precision in dies alignment under different work conditions (size of the piece being struck) the size and ratios of the lever are to be chosen so the angle of travel is not too large (30 degrees in this case) and using the range -5 to +5 degrees as the hitting range; in other words when the lever is horizontal the hammer is not at its lower position but a little higher (i.e. one inch for a hammer used for stock up to 2 inches); this means that even if the head moves back and forth during the stroke (due to its circular path around the pivot) this movement is very little in the working range (maximum 2.5 mm or 0.1 inches, in my hammer).
Cutting, boring and grinding to shape the dies was the hardest part because they are large (for my shop...) and quite hard (the W500 tool steel is similar to H11 and is unhardened yet but, being an air hardenable steel, it doesn't came completely annealed from the mill factory ,in my opinion). The upper and lower dies are different, one being a single block with 4 holes (for the bolts); the other one being a smaller block with dovetails, kept in place with two pieces of 1 inch plate bolted down to make a matched dovetail recess; even in this case no machining si necessary and since the holes on hammer and anvil are the same size and based on the same square pattern I can use one or the other die on each side (and also turn them 90 degree if I need). All this was a bit complicated but guarantees that future dies can be of either kind.
The air supply is a small shop compressor (bigger is better for continuous hammering) and maximum pressure is 8 bar (around 130 psi); I have an additional tank placed very near the main valve to provide max airflow at the moment of the strike and to keep the pressure at a reasonable value for a number of strikes even with my small compressor.
The arrangement of the air valves is also unusual because I've found useful to introduce a short delay between the opening of the escape valves and the activation of the main valve on striking; this way the force on the pushing side of the piston is not facing any counterforce on the other side since the pressure on this second side has already fallen to ambient; I'm still working on the correct timing for this to work at its maximum, so that it will improve the hitting power of the hammer as if you could remove the spring of a threadle hammer when pushing on the threadle: all the force of your leg would be accelerating the hammer instead of pulling the spring at the same time.
To keep high efficiency and low vibrations the weights of anvil and hammer are in a ratio of more that 20:1 (as I said the hammer had to be way lighter, increasing this ratio even more in the original project), the sizes of the dies are generous giving them stiffness and wide surface of contact with the head and anvil; the vibration dampening is provided also by a thin layer of plywood between anvil and plate, a large 1.5 inches steel base plate, a 5 inches thick wood base platform and finally a 1 inch rubber dampener.
The hammer actually has separate regulations for upstroke speed and downstroke speed; future evolutions in my mind are to add a variable upper limit (for light work needing short strokes) and later I plan to add other mechanisms to make it a cycling hammer (but this can add some problems and needs to be well designed to work smoothly...the simplest way is to add a small motor with a cam that pushes the pedal valve I'm using now, but it must run slow enough...).
Thanks to all those who already have (and have shown) a working junkyard hammer because they gave me the hope that it was an achievable goal with minimum tools and knowledge.