Re: gEDA-user: My first symbol...

[kmk <kmk@xxxxxxxxxxxxxxx>]

Phil Taylor wrote:

> You might make your pins 300 units long, that's geda protocol. 

Ok. They are 300mil now. (See attachment)



> Also, I would try to extend your rail pins a little further out. 

This was deliberate. With opamps there is almost always some kind of
feedback. This feedback circuit will have to detour the rails plus the
associated supply pin. The further the rails extend, the larger  the
opamp module will be. The kind of circuits I design, tend to include
lost of opamps...

At my former job we defined opamp symbols with the rails included in a
separate slot. This power slots of all opamps plus their corresponding
100n caps could be conveniently clustered at one corner of the sheet.
That way the schematics were less cluttered with boring power supplies
and it was easier to read the circuit that actually does the work. This
is especially adequate for dual or quad opamps.

The separate power slot also avoids potential mistakes if you try to
flip the symbol upside down to get the negative input above the positive
input. With the current symbol this also flips the rails and the
negative supply pin points upward.

Of course, the power slot differs in many ways from the other slots.
(only two pins, different graphics...) Is this possible with the gschem
symbol format?


> You'll find that when your
> pins are larger, the entire symbol should be larger.

Everything is larger now :-)


> It's okay for pins to connect to lines, because once they're in the .sym file,
> there is no meaningful 'connectivity' within the symbol itself.

??


> I try to call _pinlabels_ things that cannot be confused with _pintype_ ...
> such as Vo (Vout or Io or Iout) instead of 'out'.

Ok.


> If this is a specific part, with a specific package, you can add the footprint
> to the symbol itself.  (or not depending on your religion).

There already was an invisible attribute "footprint=pentawatt" included
in the symbol. Anything else I can do for the footprint?

> 
> Label your input pins exactly the way they are in the original manufacturer's
> datasheet.  
> 
> Are you using a micro or all analog PID?

All analog. This is for a low noise diode laser. I think, it's more
difficult to get rid of interfering overtone frequencies than to avoid
them in the first place. The characteristics of the controlled system
will never change. So there is no need to adjust the parameters of the
PID on the fly. The optimum parameters will be hard wired into the
circuit by fixed resistors.


> Are you editing the symbol in gschem or in text?

Both. I added free floating attributes with the editor and placed them
with gschem.


> If you're only sourcing _or_ sinking current to/from your peltier junction,
> you might consider using a single mosfet and a little op-amp, or no op amp at
> all if you're using a microcontroller.  This will give very good efficiency
> and little heating on your pcb because the mosfet doesn't disspate any heat.

Well, I know. But the benefit of avoiding any switching will outweigh
the drawback of heat that needs to be dissipated. Also, I need to go
from cooling to heating without glitches.

Thanks for your input,

          ---<(kaimartin)>---
-- 
Kai-Martin Knaak
kmk@xxxxxxxxxxxxxxx
Blog: http://lilalaser.dyndns.org/blog