| |
In lugnet.lego, Brian Davis wrote:
> Stalled, the train motor pulls 950 mA, while the stall current of the NXT
> motor is a whopping 2 Amps. So a single NXT motor output should easily
> handle a twin-engine train loaded to the point where it stalls the
> engine(s)...there's the matter of what the peak voltage is, which for a
> system driven by the NXT will be slightly lower than what I *think* the peak
> voltage is from a train controler, but not by much.
So it sounds like a single loop should be no problem under most situations,
since I'm not sure I've heard of anyone running three motors on a single train
(there's the same issue with how much amperage a power regulator can push).
> > I'm not sure a 1:3 pulse of power would be enough to sustain
> > a desirable speed on many AFOL trains.
>
> You lost me here. I was proposing pulsing the power ("pulse width
> modulation") on a motor output, to simulate a more standard PWM scheme. With
> an NXT motor attached, you might actually get away without a bunch of
> rapid "on off" command in the software, because by watching the motor
> encoders for feedback the firmware on the NXt would do it for you.
It sounds like what you'd want to do is run three loops where only one loop is
receiving power at any given moment. Based on that, what I was saying is that
I'm not sure a train will run nicely if it's only getting power 1/3rd of the
time. It seems like if you're giving it full power for one second and no power
for two seconds, it's roughly equivalent to giving it a steady 33% power.
> Use an on-screen menu and the on-brick buttons to select which bank of loops
> to control, and then use the three motors as three inputs to "set the
> throttle". No, you wouldn't have a console of nine analog-appearing
> throttles sitting in front of you, but it would certainly work. Start small
> is good, and I agree... but I'm pointing out the level of flexibility
> inherent in the system.
Well, if you don't mind the fact that many people would have difficulty
operating it under emergency circumstances (such as one train derailing over the
tracks of another approaching train). And don't ever hand it to anyone whose
VCR still flashes "12:00". Actually, if you include the use of buttons in the
control scheme, you could control four single-loop NXTs by using three motors as
analog throttle controls, and the left/right arrow buttons as a fourth digital
throttle.
> And this is a good reason, but it would need more testing to see what the
> limits are. Conservative is good, but very few of us have 4 NXTs to try this
> out with as yet (yeah, I do know a few folks who could, but theytend to be
> hyperactively busy with their HW).
No one person should need to own all involved NXT bricks. Since this is
something that would ultimately likely appeal more to an LTC than a robotics
group, having a few members donate NXT bricks for any shows would allow one
person to do all of the programming without having to drop a bunch of extra cash
on the project. For instance, in our club, Steve Ringe does a lot of work with
the older Mindstorms, but I doubt he'd have bought four NXT sets (I'm not sure
he's even bought one), but all told I believe our club has at least six within
the greater Detroit area (more if you include all of our far-flung members).
> I agree, but it's not large buildings that worry me, but tunnels and multi-
> level layouts. If the Hobby train IR remote is anything like the PF remote,
> just point it at the ceiling. I can control my PF Bulldozer from the other
> side of a island counter just by pointing it up at the ceiling and bouncing
> the IR. But no way will it work for tunnels.
Heh. So I guess that wouldn't work at all for one of Ringe's more devious
systems...
|
|
| |
In lugnet.lego, David Laswell wrote:
> So it sounds like a single loop should be no problem under
> most situations... there's the same issue with how much
> amperage a power regulator can push.
Philo kindly corrected me, pointing out that the NXT outputs are regulated down
to 1 A, so you couldn't run a dual-truck train all the way to stall. You could
still get a good bit of the way there, however. Does anyone know what the peak
voltage and current a train regulator can put out is?
While somebody is at it, the next time a large train is run up an incline that
is going around a bend at the same time, what is the train regulator actually
putting out? Do you ever see a train regulator running "full on" and the train
is just making it (i.e., near the limits of the system)? We had this issue with
the RCX powering the GBC trains, in that if we ever do go to "full power" (and
from the RCX, that's 500 mA, max), we get instant train wreck (usually with
impressive colateral damage, but that's a different thread).
> It sounds like what you'd want to do is run three loops
> where only one loop is receiving power at any given moment.
Ah, I see. What I was talking about was running three loops in parallel, pulsing
the power to each of them, not rapidly switching between the three from a single
port. PWM under software control isn't as smooth as you'd like, particularly at
low powers, but it might actually be good enough using the NXT motor encoder
feedback trick. It needs to be tried.
> Well, if you don't mind the fact that many people would
> have difficulty operating it under emergency circumstances
> (such as one train derailing over the tracks of another
> approaching train).
Program in a global kill switch: if the touch sensor is pushed, stop *all*
trains on *all* slave NXTs. Ideal? No... but if what you want is handling an
emergency, that will do it. My primary point is that if you are using a series
of true computers, there's a lot of possibilities.
As an example, let's say a slave NXT is running three loops under steady-state
conditions, and suddenly notices either its internal battery voltage jump up
(I'm not sure how much that would happen if it's plugged in), or one of the
loops suddenly starts running under a "no load" or "light load" condition. The
slave can correctly assume that something has happened to that train, and if
that loop crosses any others, can shut down not just that loop, but the
intersecting ones as well, "safing" that section of the layout (possibly an
out-of-sight section) before sending a message back to the master NXT for user
display. Incidently, since the master NXT isn't doing anything power-intensive
like running the motors, you could wear it on your belt like a big pager,
controling the layout from anywhere you happen to be. Your control center is
mobile.
> Actually, if you include the use of buttons in the
> control scheme...
There are four sensor inputs as well... any one of which could read an old-style
rotation sensor. If you are OK with different style throttles, you could easily
run seven (three motors, four rotation sensors) or more (gang touch sensors,
using pairs of touch sensors for throttle up / throttle down. You can probably
stack three or four touch sensors off every sensor input port, leading to
perhaps 6 up/down throttles from just touch sensors on sensor ports alone).
> No one person should need to own all involved NXT bricks.
No, but to learn how to program this stuff you really need at least two. I find
BT messaging between NXTs to be fairly easy, and very simple under NXT-G... but
that's because I've had the experience. You really need at least two to work out
the kinks on such a remote control scheme, and probably more (response times
change as you add more NXTs into the system).
> So I guess that wouldn't work at all for one of Ringe's
> more devious systems...
Really, if the PF system is going to be embedded into trains, having a "latch
on" mode would be ideal, and solve many problems. It doesn't even have to be an
"exposed" mode to kids, just a behavior that the IR receiver can produce when
properly commanded (like by an IRLink from a NXT, or a special "train remote").
OK, I'd also love non-line-of-sight solutions like compact BT between the train
and the remote, or just good old RC... but I don't know if those are in the
pipe. I'm pretty sure IR is already in the system :-).
--
Brian Davis
|
|
|
