Alt-BEAM Archive
Message #09756
To: "'beam@sgiblab.sgi.com'" beam@sgiblab.sgi.com
From: Wilf Rigter Wilf.Rigter@powertech.bc.ca
Date: Thu, 3 Feb 2000 23:26:37 -0800
Subject: [alt-beam] Re: wilf
Hello bumpsterman.
The LEDPUMP is just a method of driving a LED when the power supply is lower
than the minimum forward voltage of the LED. Say you want to power a
circuit from a single 1.5V battery and you want to turn on some LEDs to show
it's doing something. Too bad, because the LED needs 1.6V to light up, so
what you gonna do? Use the LED pump, cause that way the LED will light up
even if the battery is right down at 1V. Running a bicore off solar and need
some LEDs to light up all the way down to 1V? Use the LEDPUMP. So the
LEDPUMP is really a general purpose solution for driving LEDs from low
volts. That's all folks!
regards
wilf
> -----Original Message-----
> From: Bumper314@aol.com [SMTP:Bumper314@aol.com]
> Sent: Thursday, February 03, 2000 9:38 PM
> To: beam@corp.sgi.com
> Subject: wilf
>
> Wilf, I couldn't seem to reach you though your other mail box so I'll ask
> you
> thought hte list. How am I supposed to hook up the capacitor and solar
> cell
> on the full range LED pump?
>
> Steve
9757 Fri, 4 Feb 2000 02:34:29 EST [alt-beam] Re: wilf beam@sgiblab.sgi.com Bumper314@aol.com In a message dated 2/4/00 12:27:26 AM Mountain Standard Time,
Wilf.Rigter@powertech.bc.ca writes:
> The LEDPUMP is just a method of driving a LED when the power supply is lower
> than the minimum forward voltage of the LED. Say you want to power a
> circuit from a single 1.5V battery and you want to turn on some LEDs to
show
> it's doing something. Too bad, because the LED needs 1.6V to light up, so
> what you gonna do? Use the LED pump, cause that way the LED will light up
> even if the battery is right down at 1V. Running a bicore off solar and
need
> some LEDs to light up all the way down to 1V? Use the LEDPUMP. So the
> LEDPUMP is really a general purpose solution for driving LEDs from low
> volts. That's all folks!
Right, so I want to make a blue LED run off a 1F cap that I let charge up all
day. So would I put the inputs of this LED pump in place of the LED on the
SIMD1 to get same power saving features of having a regular LED in there, but
just being able to drive a much more thirsty load?
Steve
9758 Fri, 04 Feb 2000 00:55:42 -0700 [alt-beam] VSPANS (was: The 240 microcore - was: Bicore question) beam@sgiblab.sgi.com Jean auBois At 09:50 PM 2/3/00 -0800, Wilf wrote:
>Here is a little thought experiment: in an open ended 6Nv (74HC14) chain
>with LED indicators, what is the maximum number of processes (output low
>pulses/LED on) that can be active at a time? I have not tried this but I
>predict that if you try to "stuff" more that 3 processes in a 6Nv chain some
>interesting things may happen.
Well, I've always used the following really disgusting "solution" to this
problem.
Let's pretend two different (and actually quite conflicting) conditions
exist at the same time. The first condition is that whatever signal you
are stuffing into the chain is of a sufficiently high frequency that it is
above the "knee" of the capacitor-resistor high-pass filter. At that
point, there would be very little attenuation -- in other words, the
capacitor-resistor portion of each Nv neuron would "disappear".
The second condition takes the first for granted -- in other words, you
still have a chain of six inverters but they are directly connected. In
other words, they'd look like:
in ----->o---a-->o---b-->o---c-->o---d-->o---e-->o----- out
Pretending that we are using a D.C. level input, if in=high, then a=low,
b=high, c=low, d=high and e=low. In other words, you'd see three
low-active "processes" at a, c, and e. Likewise, if in=low you'd get the
contrary state with three low-active "processes" at b, d, and out. So...
the maximum number of processes is going to be three in a six Nv neuron chain.
For what it is worth, I've tried this portion Wilf's experiment once upon a
time. I applied a square wave to just such a six Nv chain (made out of a
'14) and looked at the result on a dual-channel oscilloscope using the
driving signal as the trigger. At a sufficiently low frequency, you'd
stuff in a process and it would travel down the chain until it "fell off"
the end. At a somewhat higher frequency, you'd see two processes somewhere
in the chain. I then skipped a step, frequency/signal-wise 'cuz all I had
was a simple signal generator -- two processes in such a chain can be
separated by two or four inverters. Once I got above the "knee" however,
every other inverter showed exactly the patter described in the previous
paragraph until I got up to 1 Mhz which was the limit of the signal
generator. This was well within the maximum switching frequency of the '14
I was using, so I don't have a clue what happens if you go fast enough that
it would make a difference.
Interestingly, if it were a loop instead of a chain you'll notice that in
both cases the output matches the input so such a loop ought to be stable,
i.e. static. However, we really can't ignore the capacitor-resistor
portions of the circuit so it can end up being -- under Really Bad
Circumstances -- a very high-frequency oscillator which is only limited by
the minimum switching time of the IC. I -think- that this explains the
so-called super-saturated state that people report where all of the
attached LEDs glow, but dimly & the IC gets very warm. Anyhow, the signal
looks pretty high-frequency when you look at it on a scope and chips use
more power / get warmer as the frequency gets higher.
So... that's one possible clue into that part of the thought experiment
even if my underpinnings are shaky. Nv-only chains only really show
"interesting" behavior at relatively low frequencies. It reminds you of
Tilden's earlier name "VSPANS" which (hoping I'm close here) stood for Very
Slow Propagation Artificial Neuron (or nervous-net) Systems.
jab
Home