Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[flameaway]

Yeah, so I was scrolling down through this thing thinking, damn this is really annoying.  Too bad I don't know how to use the forum tools well enough to do a better job.  Then I thought, "If it's pissing me off, I sure am glad other people are so patient..."

Ultimately I decided that since I just built an auto repeating water cannon out of three rocks and a toenail (this is a fun game let me glow for a moment) that no-one would believe I was anything but lazy if I didn't learn about spoilers...

Sorry for the sore mouse wheel fingers.

Spoiler (click to show/hide)

Wanted to see if it's possible to build a one cart repeating water cannon.

It is.

I built it in three stages.

Stage one:  Water loading and insertion into magazine.

Stage two:  Acceleration and firing.

Stage three: Minecart recovery.

Water is loaded in this test setup directly from an aquifer.  I use a roller on the ramp leading out of the water filled ditch.  Everything else uses the exploit discovered by QuantumMenace (kudos to QM... again).

Here's a movie of the magazine/cyclotron spinning a cart: (fast forward about two thirds of the way through the movie for the action)


http://mkv25.net/dfma/movie-2482-spinning

Here's my first test shot, firing an empty mine cart: (I one stepped through the action)

http://mkv25.net/dfma/movie-2483-firing

I didn't expect the cart to travel so far.  So here is another run with a longer test range. (one step)

http://mkv25.net/dfma/movie-2484-longerrange (I think it was starting to slow down just before it hit the edge of the map)

I'd noticed in my earlier experiments with loading multiple carts into these structures that the minecarts bounce around until they achieve a steady path for each minecart - no constant Newton's cradling with two minecarts.  Here is a video of the process happening.  {That's my best architect that gets blitzed by the carts.  I forgot to forbid the cart.  Stupid dwarf, or stupid me - whatever.}

http://mkv25.net/dfma/movie-2485-sortingthemselvesout

Here is a full speed test shot of the two mine carts.

http://mkv25.net/dfma/movie-2487-fire

So when am I going to get to the water cannon part?  Right now.

First test of the water loading system.

http://mkv25.net/dfma/movie-2488-loadsilver.  (Works great!  Getting water into the minecart was the hardest part for me to figure out.  Never worked with track rollers before).  You'll notice that the video makes hash of the text menus.  I checked to make sure the cart was loaded just after going through the trench and again after it had been running in the cyclotron for awhile.  Loaded in both cases.  You'll not be able to see that until firing I guess.  {How do you like my uber messy breadboard setup? Test of concept that worked, yay!}

In this shot, I load two carts with water...  To my dismay they unload each other in the cyclotron.  (I'd been hoping to increase the rate of fire by spinning several carts at a time. I still think I can do this across multiple z-levels - difficult since mine carts unload when falling z levels.  So I'm thinking about a bottom loading magazine with the firing mechanism on the top level - an elevator gun.)

http://mkv25.net/dfma/movie-2489-oneshotatatime

Note:  You'll notice the strip of acceleration ramps leading from the water loader to the spinner?  Looks like the collisions happen between the two carts because the one in the spinner is going so much faster. Accordingly,  I tried to inject the second cart with a bit more velocity.  Didn't work.  Might be able to fiddle with the timing by adjusting the speed of the cyclotron.  So a better injection procedure seems likely.  It also appears to me that carts in the cyclotron/spinner reach a max speed and turning corners doesn't seem to slow the cart down (or at least that 'friction' is greatly overcome by the impulse provide by the track ramps).  This means long stretches of straight track are not necessary to achieve an energetic cart.  (I noticed this when doing the preliminary fiddling with my magazine setup detailed here:

http://www.bay12forums.com/smf/index.php?topic=120039.0

Now the fun stuff:

http://mkv25.net/dfma/movie-2490-firstwatershot

Not bad at all.  Why does this thing throw mine carts straight and water crooked?  Anyone know?  (Is it because the 'water block' is passing through a fortification? I'm gonna setup several fortifications in a line and fire through them and see if the dispersion increases.) Did you see the explosions of blocks?  Yeah... At first I was all excited thinking I'd blown up a contructed wall with a fortification carved into it!  Yay!  As it turned out I was using a minecart full of rock blocks as the blocking cart before the fortification.  Still, a cool little unexpected thing.

http://mkv25.net/dfma/movie-2491-boom

Finally:

http://mkv25.net/dfma/movie-2492-onecartwatercannon

Full speed action shot of the water cannon firing. Rate of fire is aproximately one shot every 64 ticks.  I hope to improve this. You may have noticed that I'm using a smaller 2 z-level version of the cannon in this movie.  I think I can make a whole one of these fit in a 4x9 *edit across two z-levels* footprint.

Looks like I need to increase the flow into the water loading trench.  I'm getting dry shots.  Also a bit confused by the shot spread.  It's throwing water everywhere.

Another cool thing about the cyclotron doohickey is that you can set one gun up to fire in any of the four cardinal directions, just by placing four floodgate triggers instead of one.  One floodgate in each wall the cart hits during it's path around the machine.


Anyway.  I was happy to get this working.  First major DF project - well besides learning how to play...

Amazing game, big props to Toady 'n Three Toes.

Spoiler (click to show/hide)

*EDIT*



NOT built yet.

These are my new plans for a simple one cart repeating water shotgun.  I haven't messed about with increasing the aquifer flow yet, because I want to find the optimal size through testing. 

I'm a bit concerned about the recovery system.  Cart might derail on the central track.  I'm hoping impetus will carry the carts into the trench on the more indirect paths.  I don't want to use ramps and I don't think t junctions would work.

If you want a simpler alternative just pick one recovery path and stick with a one direction shooter. It'll require some fiddling with the design to get the most direct path that enters the trench from the left proceeding right, but that option will work.

Everything else in the design is a tested solution.

Question:

If I stick a wall dead center in the middle of the trench, but on the z-level just above, a cart striking it will drop into the trench - with zero forward velocity.  I've tried to drop directly on a roller and the cart hangs which leaves a cart stuck in the ditch.  Is this typical?

Spoiler (click to show/hide)

*Edit (again)

After a rare moment of intelligence I found a better way to handle recovery



As with real world politics, the addition of a gun barrel often simplifies matters.   I've included a gun barrel that directs a shot using two floodgates and a couple of levers to control pathing.  This small structure would be all of the machine that is exposed when dealing with the enemy.  Now I can simply direct the recovery for each different direction to a common hole using two extra z-levels at the top.

Makes the whole thing much cleaner.

<points at self and laughs>

You know... If I had made the common recovery hole the one on the lower left... well... things would be simpler on the loading level. 

duh.



I think this modification will keep the trench full:

 

I added a pump that is powered by the same gear assembly that powers the roller.  It pumps into the ditch which is itself connected to an aquifer.  I will likely have to change the geometry of the trench a bit.

Now, I think I'm ready to build this. (I also think I've figured out a way to load multiple carts in the same loop and keep them loaded with water.  (most of the time)

Yeah, that was a teaser, check back here tomorrow and tease me if it doesn't work.

Cyas

Spoiler (click to show/hide)

***EDIT***12/10/12

New gun works, very much better than the breadboard model.  It is still very slow, because:

1) I've choked down the speed of the recovery system, by not placing acceleration ramps.  I did this purposefully because I'd like to 'dynamically tune' the system.  That and test to see if a speed increase during the recovery process translates to a faster gun.  So rate of fire is fairly meaningless at this stage of testing.

2) I've only spun up one cart into the magazine, the upper cyclotron.  (I'll be trying to capture a movie of a two cart payload before my fort dies.  Might not be succesful though.. I hope that I have solved the problem caused by inserting loaded carts into an already loaded cyclotron.  The difference in speed causes violent collisions that unload the payload of both carts.  To address this issue I have created a path that spins an entering cart through an injection cyclotron that's almost identical to the actual magazine.  I'm hoping that this will get additional carts up to speed before entering the magazine proper.  There are sixteen possible positions for a minecart in the upper cyclotron (good argument for using a larger cyclotron for weapons of this nature.  Larger cyclos reduce the chances for entry collisions with additional carts).  Most of the time the cart will enter the upper magazine safefully.  The other times I expect I'll have a couple of carts semi-Newton cradling around until each sorts out a place for itself in the upper cyclotron.  During this process the cannon will be firing some dry shots.  But the carts should reload properly and continue to path through the machine.  Not sure I can test all this today because my fort appears headed toward a major tantrum spiral.  No biggie, I'm save scumming to get results anyway.  But it delays things a bit while I redo stuff.

Here are screen shots of just the machine

Aquifer lever:



Cart loading:



Magazine pre-acceleration and recovery:



Where the pain happens, just kinda slowly:




Succesful stop motion test firing.

http://mkv25.net/dfma/editusermovie.php?view_movieid=2494

Did you see that?  This movie is why I didn't wait to show complete results.  Check this out careful

http://mkv25.net/dfma/movie-2493-lastrecord

See all the water spots to the right?  This thing fired ten or fifteen times approximately ten timesand then shot the moving cart and the stationary cart right through the damn fortified wall...  I caught it during a record.  This is the source of my earlier problem when I thought I'd misplaced the fortified wall.  I discovered that I didnt' misplace it.  But I had no idea the mine cart disappeared because this cannon shot it through the fortification.  I thought it was one of those decontruction quantum dump bugs.  Look at the water payload.  The (833) units of mud laced water unloaded beginning at the stationary cart position and end in the tile of the fortified wall.  Yes, the wall is still there, how could it not be?  I checked, though. 

Has anyone seen a shotgun do that before?  Both cart down range, fast.  Is the range that this thing is tossing water similar to that achieved by other water cannons?

{I just watched this sequence more carefully.  Somehow the blocking cart ended up in the recovery path and paths right through the machine... behind the projectile.  How the heck did that happen?  The recovery channel is behind the blocking minecart (that's the whole point of it).  Did it bounce in there, when struck by the silver cart? It is a wooden cart, and much lighter.  Really interesting ***edit upon further reflection I'm fairly certain of the bounce theory.  We've seen this behavior before when loading multiple carts in a cyclotron.  Jostling unloads the carts at high speeds.  And every once and awhile the video shows the projetile unloading at the blocking cart position.  So the innocent little light weight *oaken* minecart gets slammed into and gains a tiny vector.  The silver cart starts dribbling it against the opposing wall until things get out of step and presto change-o the wooden minecart bounces into the trench just behind the silver mine cart. Both whip around and hit the far wall with the wooden cart bringing up the rear.}

Bad news is that I forgot to save the movie showing the auto repeating shots.  I'll try and catch this whole sequence on one movie.

In action:
http://mkv25.net/dfma/movie-2495-fullspeedrepeatingshots

This is pretty interesting.  The blocking minecart is unloading the projectile's payload, but only every once and awhile.  I wonder it thing is approaching some collision energy limit.  But it looks to me like the water gets deep enough to occasionally wash the blocking cart right through the fortification.  If this thing was fast I'd have been looking for an off switch.

I also have the aquifer flow at half what it's currently designed to deliver.  One floodgate is closed.  I suspect the cannon is experiencing a lot of dry shots, because it's taking two units of water out of the trench faster than it can be replaced. That's a bit surprising if true, consider this thing is operating slowly.

I simply don't understand why the payload is occasionally unloading at the blocking cart.  I'm not experienced enough with cannons know how put this result in context.  Hell this cannon I've got is the only one I've actually seen operating.  So I have nothing to compare this result to. 

Could use some feedback on this... please.

Here's the first test of the aiming mechanism:

http://mkv25.net/dfma/movie-2496-switchfiringdirection-failure

Reason: I did not place an acceleration ramp after the corner ramp.  The shot hit the wall rolled to the bottom of the direction change ramp and stopped... as you'd expect.

I've also tried inserting a second cart into the cyclotron.

Here:

http://mkv25.net/dfma/movie-2497-injectsecondcart

Failure:  This is pretty interesting.  This result indicates that the time spent in a cyclotron increases velocity past what is achievable when coursing around one circuit. In other words each circuit adds a bit of extra velocity. Therefore, unless I've missed something, corner track ramps when used for direction changes do not appreciably slow a mine cart.  After all they make up one quarter of the track positions in the cyclotron. So any friction caused by direction changes is more than a factor of four less than the total acceleration provided by the twelve acceleration track ramps.

There seems to be an easy solution to this problem with the current model.  Just let the second shot spin up to speed in another cyclotron before trying to insert it as a second cart in the magazine cyclo.

This pretty much completes the main testing I wanted to achieve with this iteration of the cannon.  Oh there's some other stuff I could do, but the fort is dying and response is getting slow.  So at this point I'm gonna call it a wrap on this trial.  I have some plans to redraw and a more coherent presentation to put together on this run.

Oh yeah, I'm gonna open up that other floodgate and see if increasing the aquifer flow decreases the number of dry shots. (No luck, my game just crashed)

At this point, I'd really appreciate some feedback on this cannon.  The two carts through the fortified wall was very surprising.  As was the cycling speed considering I set up the slowest recovery system possible.  I'm going to go watch some video of other cannons in operation. (I didn't want to watch them before getting a working model myself)

Initial impressions:

1) Design geometry seems to effect the shot spread.  Maybe this is just because the second machine was producing much higher cart velocities and any angular momentum has less time to deflect a fast shot.

2) As expected shorter recovery paths seem to increase rate of fire.  (I'm concerned about skipping carts across the loading pool.  But if the path itself is short low velocity return may not matter too much.)Means my 4x9 2-z level design has a chance of attaining a much higher rate of fire... You know, if it works.

3) The cyclotron appears to accelerate the cart to the max speed allowed by the game, which indicates to me that the cart is not being significantly decellerated during direction changes.  Either that or three acceleration ramps are enough to attain max speed.  That's the max number of serial acceleration ramps this cyclotron has.  Three and direction change is the pattern in this machine.  Do you think a bigger cyclotron will create a faster projectile...?  I really don't.

4) Rate of fire is the obvious limiting factor, balanced by the ease of setup.  The rate of fire limit is ameliorated when you toss empty minecarts (because you can put several empty carts in one cyclotron.  I'm trying to decide if that would be an efficient solution to seiges.  Be a messy clean up everytime you used the thing.  And, because heavier carts likely do more damage you'd have to build a lot of metal minecarts, or fill up a lot of wooden ones with heavy stuff.

[Urist Da Vinci]

Wanted to see if it's possible to build a one cart repeating water cannon.
...
Finally:

http://mkv25.net/dfma/movie-2492-onecartwatercannon

Full speed action shot of the water cannon firing. Rate of fire is aproximately one shot every 64 ticks.  I hope to improve this. You may have noticed that I'm using a smaller 2 z-level version of the cannon in this movie.  I think I can make a whole one of these fit in a 4x9 *edit across two z-levels* footprint.

Looks like I need to increase the flow into the water loading trench.  I'm getting dry shots.  Also a bit confused by the shot spread.  It's throwing water everywhere.

Another cool thing about the cyclotron doohickey is that you can set one gun up to fire in any of the four cardinal directions, just by placing four floodgate triggers instead of one.  One floodgate in each wall the cart hits during it's path around the machine.


Anyway.  I was happy to get this working.  First major DF project - well besides learning how to play...

Amazing game, big props to Toady 'n Three Toes.

Cool. Could you post screenshots of both z-levels for the design in the final video?

If it could fire in more than one direction, wouldn't the cart return path below get complicated?

Your simple, single cart design might make water cannons more accessible to the majority of players, who can't handle complex designs such as that made by QuantumMenace.

[flameaway]

@Urist
 
I was like a little kid, getting this thing working.  Learned alot about mine carts, which had intimidated me prior to this.

The breadboard setup that I cobbled together works but has some problems.  I figured out a cleaner design over night and I'll put that together and see if it works.

The 'cart recovery' exits from the cannon and the entrance back into the loading system lie outside the rest of the machine, so the routing for the four 'shootable' directions can be set up in a ring around the water loading station. But yes it's more complicated, and no I haven't actually setup my idea for it yet.

I also don't like that when the example gun is repeating, the projectile exits along only one arm of the cyclotron, wasting three quarters of the acceleration available to the structure.

So, I'm going to put an improved version together in the morning and if it works effectively I'll post a design that has:

1)  A fairly simple recovery system for a gun that fires in all four   three directions (can do four but haven't include it in this design). (The spent cart drops through one of four exit holes that all exist outside of the structure of the rest of the machine.  They fall onto the water loading level which is just a one tile wide ditch, so there is plenty of room down there for pathing four carts back into the loader.

2)  A repositioned water loading system that fires into the bottom cyclotron directly south of the ultimate exit -- the one that fires east in the current model.  This will allow the cart to be accelerated for almost the entire 'length' of the path available to the cyclotron.  This will only work for a one direction/one cyclotron level design.  However, my ultimate aim is to create several cyclotrons that are stacked on top of one another. Each will feed into the layer above, ultimately reaching a top layer that will have the firing mechanism.    In this case, the cart/s will already be accelerated by the multi-level rise through the machine. I'm hoping that this will allow multiple carts in the path, in order to increase the rate of fire. (the recovery system slows things down, because a fast moving cart skips across the loading trench.  Any suggestions to fix this would be appreciated.  Propsed solutions should be as simple as possible since this whole project is geared toward minimalism and simplicity)

***

A fairly short tutorial exploring QuantumM's track ramp exploit.

Spoiler (click to show/hide)

For now I'd like to include a schematic of the cyclotron design (it's complete) with an explanation directed towards players that haven't approached mine carts yet.  (I'm a bit loath to start trying to explain things on a forum full of smart people - most of whom have been playing much longer than my eight months. But here it goes...)

With track ramps one of the main things that helped me start putting together predictable results was to understand how the track ramps are named.  When I think of the direction northeast, I generally think of a vector heading from center toward approximately 1:30 on a clock face.  In the game, the corner track ramp that would provide that vector is call Southeast.  So, if you start trying to cobble track ramps together based on a normal understanding of map directions, the machines you make will not work predictably.

Here's how I picture it.  Come from a direction; go in a direction.  So a Northwest, NW, track ramp comes from the north and heads west.  Or comes from the west and moves north. A SE track ramp comes from the south and heads east and vice versa.  Think of things this way and you can hook up designs that route the minecart along effective paths.  If you build a design and it hangs in a position, you've misplaced a ramp (or struck a two headed ettin who catches the gold mine cart and then simply walks into your now undefended fort - albeit with a bruised heart).

Another thing that is critical to understand is how the corner track ramps work:

They will:

1) Move a cart around a simple ninety degree direction change, if confined to a single z-level (no acceleration provided because the cart has both paths open for travel)

# = wall

Code: [Select]

  #  #
# SE NE
# NW

That's what happens in this example with the SE section of track.  It moves the cart from the south around the corner to the east.

Yeah, but what about that NE track ramp?  That's coming from the north and going east, right? The cart should derail, right?

Wrong.  (This is the exploit: the game just sets the cart at the top of the NE ramp and it rolls down to the east, pretty slick for us) So think of such ramps as energy coming from the closed path running from the wall and along the open direction.  So, in general a track ramp on one level will move a cart around a corner if both directions named have a valid opening to travel.  If only one of the directions named has a path, the track ramp will accelerate the cart along that direction.  So, in the code example I provided, if there is a wall to the north of the NE track ramp the cart will hit that ramp and speed up traveling east.  Likewise the NW track section will push a cart north if it has a wall to the west.

So that leads us to a one level cyclotron. (of course, various designs are possible)

Code: [Select]

SE NE NE NE SW
NW          SE
NW          SE
NW          SE
NE SW SW SW NW

This will spin a cart clockwise, and all the walls are on the outside of the structure. The center can remain hollow. (I want to put a smaller cyclotron in the middle spinning the other way - just for the cool looking effect.)

Here is an alternate design with the walls on the inside: Note: If you build this one, four single-tile outside walls are necessary to keep the cart from careening off the track.  (You'll see where they have to go when you try and run it. Tip: just outside the corner direction changes along the direction of travel)

Code: [Select]

SE SE SE SE SW
NE  #  #  # SW
NE  #  #  # SW
NE  #  #  # SW
NE NW NW NW NW

Notice that in both cases the machine moves the cart clockwise. Acceleration ramps provide direction and velocity, so if you swap the acceleration ramps the machine will spin the other way.  If you can look at these two and understand how the track ramps path the cart around, then you understand one level track ramps. If not, just study it a bit and you'll get it.

2) Now track ramps are ramps, after all, so they will also push something up or roll down one z-level if a channel exists through the floor above an up track ramp.  If there is a channel above a track ramp the cart will exit one level above and move in the direction of travel provided by the adjacent track ramp on the new level.

So:

X = channel

SE (with a channel over it) looks like this on the next level up.

X NE #

The cart moves to the east and stops on NE.  (note that the example moves clockwise, going the other way the bottom ramp will inject the cart up one level and one tile south). The X will actually show through the channel as a  SE down ramp.

Okay, but this stuff won't help you set up routes, right?  And you have to have at least a one stop to start a cart spinning in a cyclotron.

Go to 'h'

Position your cursor over the spot you want a cart to enter the machine (don't put the spot inside the machine or you'll kill dwarves.  Put the spot one tile away from the opening into the machine. There doesn't have to be a track or anything at all in this spot (better if there isn't).  Hit 'r' for new route, 's' for new stop.  The new stop should be highlighted, if it isn't, make sure that it is.  Once the stop, not the route, is highlighted, hit 'v' to assign a minecart.  Esc, and you are done.  A dwarf will grab the specified minecart and plop it down on that spot.

Step one complete.

Step two:

Once the cart is sitting where you want it.  Hit 'h', highlight the stop and hit 'enter'.  This will bring up the conditions for moving the cart from the stop.  Hit 'x' three times to delete the default conditions.  Press 'n' which will produce a single new departure condition.

You will want to push the cart into the machine, not ride or guide, so hit 'p' to change the method of departure to push.  You'll also likely have to change the direction the cart is to travel in, so hit 'd' until you see something like  this:

"Push cart <desired direction>"

One more step and you are ready.  You have to tell the dwarf when to push the cart.  'c' will cycle through a bunch of options.  Set it to 'always'.

Unpause and if you've done it all correctly a dwarf will wander by and give the cart a little push into the machine. The cart will skid along the ground and enter the machine, no track necessary.

Immediately pause the game and forbid that minecart.  Otherwise it will eat all of your dwarves as they come along to put the cart back on it's stop.  They don't care that it is spinning at lethal velocities, they just don't like an out of place minecart...

There you have it.  If you've done it all correctly you should have a working cyclotron with a cart buzzing around inside of it.  Being able to do this opens up a bunch of other options, most of them weaponizable.

I would appreciate any corrections or improvements to this post. 

I'm still learning myself.

Captain Obvious rides again...

[flameaway]

It occurs to me that you could use something like this in interesting ways...

(sorry if this has been covered)

For example, since one of these cyclotrons will toss a mine cart waaaaaayyyyyy across the map... without track...  that you can save yourself the cost of a track structure and combine some exploratory mining in the same process.

Just dig a single wide tunnel between appropriately place cyclotrons and use this as your material transit system. Just off the top of my head the whole system could probably be made inaccessible to drunken dwarves fairly easily. Toss carts between supply points using the directional gun barrel doohicky I detail above (as a switch).  Cyclotrons transit downward between z-levels just as easily as up...

Setup gun barrel positions and supply them from a central loading station.  The filled cart whips out to the correct  barrel, fires in the appropriate direction and gets tossed back by the recovery station using a return cyclotron.  (probably down a different, return, tunnel.)

[Oaktree]

It looks like there are four issues to be looked at for using mine carts as weaponry:

1. Loading - in this case dipping the car in a liquid-filled trench and getting it out.  It looks like rollers are needed for this most likely.  I experimented a bit with trying the ramp exploit to get loaded carts out and did not have consistent luck.  (Had it working for while with water in one location, but no luck with magma.)

2. Transit to firing position - this is using ramps and such to get the cart from where it was loaded to a position for firing.  Usually bringing cart up or down z-levels.  A complication in this is using switches or walls to allow for multi-directional firing.

3. Firing - Getting the cart into a position and velocity so that the cart contents are properly flung at the targets.  Additional factor here is the security of the firing position to keep the enemy out; usually using fortifications and open gaps.

4. Recovery - Getting the empty mine cart back to the loading location in order to repeat the cycle.

==

For recovery I have simply been putting a vertical shaft directly under the firing location. The cart slams into a wall - shoots the contents through a fortification located directly above the wall, and then drops down the shaft.  At the bottom of the shaft is a curved track ramp; this causes the car to roll off at which point it can be accelerated further as needed.

From multiple firing points the carts can probably be directed back into a common drop shaft down to the loading location.

Also, would an alternative way to handle firing be to move the car into position at relatively low speeds and then use the "barrel" to accelerate the cart for the last 6-7 tiles as it reaches the firing position?  Curved track ramps in serial order (aka "linear accelerators") can do this effectively and compactly. 

[QuantumMenace]

You could easily combine this larger square with the minecart grinder concept to make a "turnstile". Invaders that enter the square get smashed through fortifications they can't walk back through, great for managing hostile traffic. One of the corners needs to have no outlet or fortifications, so a bridge built there to stop the cart won't throw it out of the square. Note that using an exploit ramp in the corners like in the grinder will make the cart lose its momentum when it hits the corner, but guarantees that it can't be permanently stopped there.

[flameaway]

Easy way to load a whole bunch of carts.

Cart stockpile in the bottom of a one tile deep channeled pit.  Say 10 x 10

Fill the stockpile.

Fill pit with water,  Drain through aquifer or into caverns.  Or into another pit with pumps.

Use dwarves to place carts in loading position.

You could cycle a lot of carts very quickly that way.

Did you know that you can set up a pump stack, direct the flow directly down through a grate and spin a mine cart through the resulting column of water...

You get a water covered and dolefully empty mine cart.
 
"Urist! Take the cover off the damn thing first!  You drunken dumb ass!"

"Screw you, you faux super being from a universal epiphenomenal processing system! Kiss the beard on my ass!"

<looks at Urist, solemnly>

"You have a shaggy ass? Ever try a Norelco butt razor? It uses magma... and you can use it while sitting down."

"You made me build that same damn bed fourteen times all over the damn fortress... I'll jam your skull into your brain tissue."

"That's a no?"

"I need a drink."

"Whatever gets you through the working day..."

Anyway... I digress.

Anyway that would be a super fast loader.  Hell a water/magma shotgun would be a snap to build and as fast as the cyclcotron.

<sigh>

Setting up a couple of pumps to pump into a single tile hole fill the hole sorta fast. Draining the water through a hatch in the floor works fairly fast.  Pressure swith in the pit senses the presence of the cart.  Shuts the floodgate blocking the entrance to the pit and toggles on the pumps. Hatch opens, drops the cart (and the water it was briefly submerged in) onto an acceleration ramp in large open room with only the stretch of wall required to impell the minecart.  The water disperses easily in the open room, the mine cart jets off to kill goblins fully loaded.  Requires a lot of switching stuff that I haven't really thought through yet.  Still it's the fastest inline way I can think of.

I should probably go a build the machine I've been designing all day.

[mscantrell]

Excellent explanations!
Two Internets and six masterful socks for you!

[flameaway]

Okay I have a working machine with a structural problem and a dissolving fort.

I misplaced the fortified wall one square and the car won't path down the recovery hole after the first shot and so won't repeat. Easy fix and means the whole machine will work when I adjust that fortified wall.

But the first shot was successful (and either by chance or no a lot more organized with a greater range, impossible to tell a trend from a single shot)

Have to go into town for awhile but I should be able to get a few more test shots in and will post a movie of full auto action and rate of fire.

Redrawing the plans to match the actual machine will probably happen tomorrow.

I've also kept a log of the entire build and test process.  I'll squash that into some kind of logical format and post it, so that folks will see some of the common mistakes that arise during setup.  (took 6 shots before I achieved a good result, mostly due to misplaced track sections)

[flameaway]

QM, hadn't consider using a fortification that way.  You could knock em down into a path to your arena.

Or if you put one cyclotron inside another you could ping pong (frappe) invaders back and forth, (likely very quickly) through the fortifications between the two cyclotrons.

Two cars in the grinder (grinder was a nice idea BTW) work much more effectively.  The invaders can sometimes catch one cart, but a second one bumps into the stopped cart and gets them both moving toward a harmonious minecart pathing solution, (after while bashing spleens and breaking ears) Plus loading a grinder cart with 80 plus rock block increases the weight of the grinder cart dramatically, to around 1500 whatevers.  The blocks will stay in the cart until it hits something then you suddenly have a bunch of building material flying about annoying the local fauna.  I want to test to see if an actual difference in the level of damage dealt occurs with a heavier cart.

So I think using your fortification idea and bashing encroachers repeatedly back and forth through a fortification will help keep the carts flailing about by turning goblins into projectiles that transfer energy to stuck carts.