[Ktechlab-devel] Bug of the week

P Zoltan zoltan.padrah at gmail.com
Fri Apr 3 17:24:25 UTC 2009


On Thu, 02 Apr 2009 22:11:11 +0200, Alan Grimes <agrimes at speakeasy.net>  
wrote:

> P Zoltan wrote:
>>    Hello people,
>>
>>
>>   I've found a very annoying bug, yet I don't really have time to
>> investigate it deeply. Here is the entry in the bugtracker:
>> https://sourceforge.net/tracker2/?func=detail&aid=2687692&group_id=143459&atid=755588
>> . It's about connecting logic outputs together using resistors. Instead  
>> of
>> forming an algebric mean between the voltages, se simulator makes the
>> voltage on logical output aproximately equal.
>
> That's not a bug, it's a feature. ;)
>
> Open the settings menu, select "configure ktechlab", then select "logic"
> and finally set the "low output impedance" to whatever you think it
> should be. ;)
>

  Thanks, it seems that originally it was 150; setting it to 1 ohm fixed  
the problem. My new question is then: what should be the default value? If  
someone wants to convert a digital signal to analog, by using R-2R  
network, he'll have a surprize...


> What I've been mulling over is this "EMISSION COEFFICIENT" crap in diode
> settings...
>
> It is currently set to disallow settings below unity, (which prevents
> certain types of diodes from being implemented.)
>
> There seem to be four types of diodes out there:
>
> PN junction diodes.
> MS junction diodes. (Schottky)
> Germanium diodes.
> Selenium diodes.
>
> There needs to be a wizard that can work the gawdawful equations for you
> for your selected diode type and give you some user friendly parameters
> that will let you tune the variable for your specific part.
>
> The equations for this variable really are gawdawful, I have them for PN
> junction diodes and MS diodes, but it'll be a pain to copy them into
> e-mail and explain each of the dozen or so variables. =P
>
>

  This wizard and settings should be included in the not yet implemented  
component editor (in my opinion); the equations should be part of the  
model -- in some way -- but I think we'll need some brainstorming in this  
topic: we haven't discussed how could we edit/modify models, without  
recompiling ktechlab.





More information about the Ktechlab-devel mailing list