[Ktechlab-devel] Working build

[Alan Grimes]

Judging by the sourceforge statistics, we actually have a sizable user
base!!! (EEEK!!!)

I have not been able to fix the current branch. =(((

And I'm going crazy without a working ktechlab to play with my
brainstorms!! =((( I finally figured out how you are SUPPOSED to use
transistors for IV conversion... Current mirrors can bite me, they're a
fucking awful design!!! A transistor is inherently a current amplifier.
To use it with a current output DAC, you need to find one with an
extremely low Hfe (between 20 and 100 depending on the chip) and base
resistance, bias it negative, and then put a really low load resistor in
the collector circuit, if you have a differential output DAC, make a
differential amplifier with a quarter to half an ohm in the emitter
circuit... At least that's what ktechlab is telling me to do. =P

The transistors will need to be as linear as you can find and good for
at least 500mW of dissipation, possibly even heat sinked, the power
supply will be putting out some pretty heavy current! =P

The secret sauce is to hold the emitter circuit at -900mV. =P

Since the impedance of this circuit is extremely low, you don't need an
output buffer of any kind, so you have a one or two transistor circuit
instead of a pile of opamps with dozens of transistors in each... I'm
not saying this approach will technically outperform the opamp solution
but it is definitely the "purest" approach to the problem. Your only
other option is to use a 417 tube in the grounded grid transimpedence
configuration. That may not be as good though. =\

I think it might be a good idea to go through my commits between
Saturday evening and today and roll back some of the more drastic ones
and hopefully at least make connectors visible again! =(

I don't have enough skill with SVN to do this myself. =( I really do
want to keep some of the changes, especially a few that exposed some
latent segfaults, but I really need to do what's necessary to make it
usable again. =(

I'm looking at the SVN logs, the last "normal" commit was 309.

Here's a keep/chuck list for the other commits.

310 was the one where I unmasked the bugs... (probably/hopefully keep.);
In retrospect, I should have made these changes one at a time... =(

311 simulator cleanup.    (almost certainly keep)
312 (keep).

[[ this is where I made my big mistake of trying to clean up/refactor
the Connector heirarchy in hopes of isolating some of the
de-refferencing deleted memory bugs.. ]]

313 -- chuck/revert
314 keep
315 -- chuck/revert
316 -- chuck/revert

[[ I then comited far too much of my failed refactoring attempt and got
myself stuck in a big old hole...]]

317 -- maybe keep
318 -- would be nice if we could keep.
319 -- maybe keep.
320 -- revert
321 -- revert (see earlier).
322 -- revert
323 -- revert
324 -- revert

Maybe I should waste my time breaking the KDE4 build... Can that be put
in SVN pleez?

-- 
New president: Here we go again...
Chemistry.com: A total rip-off.
Powers are not rights.