Alt-BEAM Archive

Message #04393



To: beam@sgiblab.sgi.com, davidperry@mail.geocities.com
From: Bruce Robinson Bruce_Robinson@bc.sympatico.ca
Date: Sat, 12 Jun 1999 22:33:51 -0700
Subject: [alt-beam] Re: bypassing a nv


davidperry@mail.geocities.com wrote:

>> What configuration are you using for your Nv loop(s)?
>
> 4NV loop

I used Mark Tilden's H-bridge document to understand how they work. See
the version at:

http://people.ne.mediaone.net/bushbo/beam/BiomechMotorBridges.html

The document shows several versions of the circuit. The version Ian
shows at Beam-Online turns on the motors when the H-bridge input goes
high. In a typical microcore the Nv gives a low signal when it switches
on, and a high signal when it is off. So there's an incompatibility
here; using the circuit Ian shows, your motors would run nearly all the
time.

In his document in figure 3, Tilden shows the design for a negative
logic H-bridge. It's nearly the same as the positive logic one, with a
couple of minor changes. Using Ian's free-form layout, you replace the
middle row of 2N3904 NPN transistors with 2N3906 PNP tranistors and turn
them the other way. What you're doing when you turn them is swapping the
collector and emitter. BTW, there's a typo in figure 3: inputs 1 and 2
are labelled wrong -- read the caption instead.

There is one other problem with the H-bridge on a 4Nv loop. If you turn
on both sides of an H-bridge at the same time, you will create a direct
short through your transistors -- meltdown time. Depending on which
neurons your motors are connected to, it's possible to create this
condition when your Nv loop starts up saturated.

The guiding principle here is, it's impossible to have two adjacent Nv's
on at the same time, so if the two sides of one H-bridge are connected
to ADJACENT Nv's there's no problem. Any other arrangement is potential
trouble.

Tilden gets around this problem very neatly, using a 74xx139 chip to
ensure that you can NEVER turn on both sides of an H-bridge. One chip
will supply two independent H-bridges. The other solution is to use the
Pulse Suppression Circuit I recently posted -- but as shown that
requires an extra chip as well (a 74xx14). I'm experimenting right now
with a more compact version.

Hope this is reasonably clear. I can scan in some sketches for you if
you'd like -- don't want to bog down the whole list with them, though.

Regards,
Bruce

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com - Simplifying group communications



Home