Alan,
Sometimes I have an idea of the layout beforehand, but sometimes I
don't. Generally, I'll put a voltage regulator circuit in a section
of the board that minimizes the wire length to the switch on the
panel, or to the dc socket on the back panel, however the circuit is
configured. Same for RF output; I'd prefer to have it at he back of
the board. In the case of this IF board, the input from the crystal
filters comes in at the 3" wide left end of the board, because the
switchable SSB/CW filters I'll be using will be on a sub-board. They
will be fed by the RF input filter/RF amp/mixer/post-mixer amp board,
and obviously I want the input to that board near the back of the
case, but it's not as critical, since it's simply an antenna
connection through coax. I'd like to try to minimize the number of
wires and cables I have flying over the PC boards, though.
When I did the SKN Special trans-receiver, If you take a look at the
web site, I printed out correct-reading images of the boards and glued
them to cardboard, and did the final planning by shuffling them
around. I had to reposition the VFO compartment to the side under the
top lip of the case after I modified the low pass filter to a better
design that turned out to need more length. I was watching "Rides"
the other night, and they were converting Jay Leno's Toronado to rear
wheel drive, and then "oops...we forgot the air conditioner...now
where do we put that?" It ended up in the trunk. Sometimes planning
will carry you only so far!
I did autorouting almost exclusively when I first started with Eagle.
The rat's nest simply looked impossible. One thing I use a lot while
routing, either auto or manual, is "show ground"...click eye icon and
type gnd in the command line text box. That helps in that I know
which end of the component to focus on; i.e., the signal end, and
whether or not I need to reorient it. The triangular
emitter-base-collector transistors drive me batty, trying not to have
to run a signal through the triangle. The pad clearances are pretty
close as it is on a TO-92. Also, I'll use the "i" info icon to show
what signal I'm dealing with, and check the schematic.
I aim to minimize the length of the signal paths whenever possible.
With autorouting, especially if you have it set on a less intelligent
setting like 20, as opposed to 50, the signal paths may end up being 3
or 4 abreast, running around the edge of the board. That's annoying,
when the components they are trying to connect are only 1 inch apart.
Often, you can reposition a couple of resistors to allow the track to
run beneath one, bisecting it, and eliminate one signal running around
the edge of the board. It takes some juggling around. The routing on
the SKN Special receiver board became a whole lot simpler once I put
in the multiple power connection points. The tradeoff of much more
sane RF paths vs. a few extra connections to the +12V jack was easy to
make, in that case.
73,
Ted
www.kx4om.com
--- In Homebrew_PCBs@yahoogroups.com, "Alan Marconett" <KM6VV@...> wrote:
>
> Hi Ted,
>
> That IS a lot of parts for a discrete RF design! I can see how
working back
> and forth on the schematic and layout would be an ideal way to get good
> routes. Do you already have a parts layout in mind before you do
this? I
> got into the habit of laying out wire-wrap projects by inserting WW
sockets
> into prototype board beforehand to get an idea of the layout. Discrete
> parts as well. But then I would solder ALL of the sockets down
first. This
> "tightly interactive" annotation between schematic and layout is great!
>
> On this board I found that the "flying wires" and parts profiles made it
> fairly easy to initially get all of the packages down and roughly
> orientated. Quite a difference from working with a bag of WW sockets!
>
> I won't be using a ground plane, but it's nice to learn how you've
done it.
> Hopefully I'll be able to MILL isolation traces, but I may also try
> Press-n-Peel Blue, just to be able to compare the results.
>
> I have been wondering how much of the board to manually route, and
when to
> just let the auto-router do it. Last night I got my first exposure to
> routing power traces by hand, and watching the flying wires
disappear after
> a net was routed (and the rat's nest updated). CUTE! I thought I'd run
> into a snag when I routed a ground trace from an IC over to a servo
ground.
> I couldn't delete my added route! But I found out that if I ripped
it up
> (back to an air line) that the rat's nest update would then delete
the bit I
> wanted (I already had the two nodes connected; I didn't want a
ground loop).
> I also found out how to rip up the entire board, which I needed!
>
> 73's,
>
> Alan KM6VV
>
> >
> > Alan,
> > That's right, it isn't very high for digital circuitry. A 16F877 and
> > a few other supporting ICs will run the count up in a hurry. That's
> > why I mentioned that my 215-pin design was discrete components only.
> > The design was the IF system in Figure 6.50 of "Experimental Methods
> > for RF Design", the one using two pairs of J310 JFETs in cascoded
> > arrangement. There are a lot of components in that design, and I had
> > about a 1 square inch open area left on the 3 x 4 Eagle board when I
> > was finished.
> >
> > I manually routed the board as I went, by the way. I added the
> > components for a functional section to the schematic, for example the
> > JFET pairs and associated circuitry for the board signal input from
> > the crystal filter, laid them out on the board in a logical position,
> > and routed; then I moved on to the next section, etc., following the
> > signal flow from the schematic in the figure.
---snip---