Alt-BEAM Archive
Message #05180
To: beam@sgiblab.sgi.com
From: Steven Bolt sbolt@xs4all.nl
Date: Fri, 9 Jul 1999 11:45:39 +0200 (CEST)
Subject: [alt-beam] Re: Something funny with the 1382 voltage triggers?
On Thu, 8 Jul 1999, Bob Shannon wrote:
> Wrong again Steven. My fusteration comes from people who assume
> they know whats going on, and dont. They would rather find the circuit
> design to be at fault (ignoring the many working versions)
The 2-transistor latch/driver is in common use because:
1) It's heavily promoted as the first and simplest device to
build if you are new to BEAM: "First put this SE together and
make it work, then worry about walkers," or words to that
effect.
2) It looks simple.
3) It is easy to free-form.
4) It can be made to switch some motors on and off, so can be
good enough.
I suspect that appealed to its originator because of low parts
count. As a workshop kit - where you control all parts - it is
pretty effective. But common sense design it ain't.
> This circuit is proven to operate reliably with (measured)
> identical component values. Something different starts to happen
> when SMT parts are used however.
> BTW, the full sun lockup problem is effected by the total
> capacitance. Replace the 6,600 uf cap bank with 4,700 uf, and
> there seems to be no problem.
A common sense design would not care about such minor changes. They
would be within tolerance. For instance, the 4-transistor SunEater
can be built with a 2N3904 instead of a BC549C, a 2N3906 instead of
a BC559, a 2N2905 instead of a BC327-25 and with just about any
storage cap without significantly altering its performance. It will
also happily switch any small recorder or pager motor, even those
with a built-in cap. Solar panels can range from the largish
BP-378234 (8-cell panel) to a small calculator-type device.
If you want SMT, use the BC847C, BC857C and BC807-25.
SunEater_IV/V is much *more* tolerant, and uses a physically small
(8mm high, 13mm diameter) 47000uF supercap to provide motor bursts
as short or long as you like; switch-off is timed and doesn't
depend on voltage drop. This SE is smaller and lighter than
`classic' SEs, when they need an ordinary storage cap (as they do
in photovore designs).
If you feel that you aren't arrive unless you use a 1381 - makes
significant difference only in very dim light, btw - then add a
monostable and driver stage to it. Won't make your gadget larger or
heavier, as size and weight are mostly determined by storage cap
and solar panel.
Wilf Rigter has published a number of SE designs which look great.
Ken Huntington published his Micropower SE years ago - offers low
light performance close to what you get from a 1381 and uses only
very common parts. There is lots of choice.
> We can assume that this is only due to tolerance issues, but the
> imperical facts point in a different direction.
No, they don't.
> I heard today that SOT23 transistors are not made from the same
> dies used for the TO92 packages, due to bonding wire attachement
> issues. It seems that they are not quite the same
> parts. I'll be hooking up sets of TO92 and SOT23 transistors to
> a Fairchild dual curve tracer this weekend and seeing exactly
> what these differences may actually be.
You'll no doubt learn interesting things, Bob and your extra
knowledge will in future be appreciated by the many who will no
doubt continue to build that latch/driver, to their frustration.
But given the normal transistor tolerance - which is in the
hundreds of percents - it is simply better to design in such a way
that your gadget can live with them. This is straightforward low
frequency, low power stuff. There is no excuse for an SE design to
be so picky about components.
> Please remember that this thread is intended to be about the
> differences between SMT and conventional parts rather than the
> flaws in the circuit.
For this type of circuitry, the differences are very minor and
should not be blown from gnat to elephant by indifferent design.
---8<---
> > Even just about `correct' motors tend to help the latch/driver
> > oscillate, which it is all too happy doing (and obviously shouldn't).
>
> I've only ever seen the oscillations with one type of motor, and they were
> easily eliminated.
You might want to look more closely. There is a burst of
oscillation with almost any motor, especially in good light. It
usually dampens out.
> > Even the rapid start/stop needs of most photovores aren't
> > served well by more than a factor two.
>
> Perhaps the BEAM community should rethink the 'rapid start/stop
> needs' of a photovore. Rapid starts and stops waste energy, and
> have no place in an efficient design.
The rapid start/stop is not a matter of efficiency. It's about
making something visibly happen, like a `step' every few seconds in
not so good light, and a couple of steps per second when the sun is
found. Makes the gadget more attractive.
> > Of course there are exceptions, like a SolaRoller which should run
> > its entire race on one charge, and certainly shouldn't use its
> > motor as a brake. Even there you might want to do something
> > different, as 0.7V may be too *high* as switch-off level...
>
> If your motor stops about 0.7 volts, then you sure dont want to
> leave the latch on in a solar roller application, or you will be
> using the motor as a break!
The pager motor on my desk starts at about 0.4V and certainly
contributes to distance travelled at that voltage.
> > But anything you do with that 2-transistor latch/driver can be
> > done better with other circuitry, as you seem to be finding
> > out, judging by your previous postings.
>
> Hmmm. Well, why does it work perfectly with thru-hole parts then?
It may not be too difficult to find thru-hole parts which fail, or
SMT parts which work. Small performance differences between
components are the likely cause, not the way they are cased or
mounted.
> I looked at lots of SE circuits, and lots of ways to shuttle
> energy between suspension system components. If you say I can do
> better with something different, please point the way, and I'll
> test it.
It seems to me that best performance of your `bot depends on the
relationship between masses at both ends of the spring, the spring
itself, motor torque and burst duration.
Keeping mass low should always be a good idea. Finding the optimum
compromise between motor burst duration, torque and charge time
will be easiest if the switch-on trigger level and the burst
duration can be continuously and separately adjusted. SunEater_IV
makes that possible. Mass can be very low when using SMT on a 0.5mm
epoxy pcb, and the small 47000uF super cap helps as well.
If you feel that the switch-on level is already optimized (I assume
you tested with various 1381s), then you might add an adjustable
monostable and a 1-transistor driver stage to your 1381 as I
described earlier; that gives you somewhat better performance in
dim light.
> > Earlier, you had much to say about advancing BEAM, if
> > memory serves; so why are you now so wedded to the first and worst
> > part of the BEAM circuit book? Very odd, methinks.
>
> Well, simply because I took the time to look for the best
> combination of circuits, components and mechanical design for my
> application.
It turns out thet your choice of circuit was unfortunate, but you
are unwilling to change.
> Also I've had more luck advancing BEAM by extending the
> photopopper design than I have had trying to measure implex in a
> microcore.
No doubt :)
I have never built a MicroCore - it seemed all information to be
gained from that circuit was already published. Reading is easier
than soldering. My first and only 2-transistor latch/driver was
built quite recently (after SunEater_IV), just to check whether I
understood it correctly. Ken and Ivor had already published plenty
about it, and the performance looked uninteresting...
> > It's a matter of reliably working with diverse motors and solar
> > panels, as your are finding out.
>
> Wrong again!
>
> We are dealing with the same solar cell, same capacitors, and same motors.
> Please re-read the thread and subject line.
You exchanged parts for equivalents and the design can't tolerate
that. Doh. It should not only be indifferent to the changes you
made, but also accept a wide range of solar panels, storage caps
and motors. Else it isn't much good, imho.
> > > How many centimeters per minute, under what lighting, and over what
> > > obstacles?
> >
> > A SolaRoller race, or anything like that, will be decided by the
> > quality of the motor, the mechanical parts and the solar panel,
> > with the electronics coming in as a distant fifth.
>
> Se lets set them as constants, or set competition criteria like sollar
> roller matches.
Well, why don't you. Yet another contest idea.
> > > Lets test this directly. I have little interest in arguing the
> > > point, but I am very keen to test it emperically, maybe we can
> > > all learn something new here.
> >
> > Like what? I'm arguing in favour of properly designed, described
> > and explained circuitry, to make BEAM a better learning experience.
> > You're turning it into a "who can find or buy the best motors"
> > contest. Can be fun, but it's a different topic.
>
> No, I'm trying to learn what the difference in circuit operation is!
> You seem to be trying anything and everything else but this.
That's correct. Imho there is little point in exhaustively testing
components for an indifferent design. I was taught that if a design
fails to work as specified when components vary within the normally
expected tolerance, then the design needs to be modified. Tinkering
with values is unlikely to help. Even if you seem to succeed, the
problem will come back to haunt you when production has started.
Best,
Steve
----------------------------------------------------------------------
# sbolt@xs4all.nl # Steven Bolt # popular science monthly KIJK #
----------------------------------------------------------------------
------------------------------------------------------------------------
eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com
- Simplifying group communications
Home