While trying to get the ATMEL 89C2051 programmer to work, I
have encountered a problem with the processor power supply 
(+5V, pin 20 IO) switching. Since I didn't have the BC107 transistor,
I used two KC 239's and I found the voltage on the pin 20 to be less
than 4V (the older design using the 74573, according to EI, Dec 96).
I tried to logically explain it in the following way (follow the
original schematics):
 Assumption Ub(T1) = 5V (optimistic version)
 Since the transistor conducts at min. 0.6V,
  Ue(T1)=5-0.6=4.4V and Ucc(20)=Ue(T2)=Ue(T1)-0.6=3.8V max.
  Probably even a little less than that.

Although the Atmel works from 2.7V, I did not know for sure if that's
enough for programming too (it didn't work for me so I was looking
for a bug). So, I modified the design to reduce the voltage drop to
less than 0.1V (0.03V typically) - see the 'picture'.

                     KF 517

         +5 V o--------\   /------- pin 20 (Atmel)
                |       V /
                |  ___ =====    91R
          56 K  --|___|--|    ___         2K2
                         |---|___|-\ |    ____
                                    \|---|____|--o Power='1'
                                    /|           12 (74574)
                         KC 239    V |                   ^
                                   |          used to be 3
                             GND  =|=

I managed to put all of this onto the original board.


                               Robert Svarc   OK1FEN
                               mail: svarc@student.fsid.cvut.cz