Are you O.K. Kai? I was expecting answers to my electronics questions in second posting. Please be kind enough to consider those.

1)Any objection using LM317T for generating reqd. programming voltage?
2)Why PNP transistors (2907)for swiching?Can I use 4066/4051 types?Then switching 7805 or 7812 for Vpp is easier and I dont(possibly) bother about logic high and low actual voltages.
3)317s genarate emi/noise.What is recommended physical separation practice,- from rest of the sensitive area of the circuit.

Hallo Asok,

LM317 does not generate 'noise'. It's a linear voltage regulator producing as much 'noise' as LM7812 or similar. So, LM317 will not make any trouble!
Voltage switching by the help of LM317 guarantees, that there's no switching spikes, when switching from 5V to 12V and vice versa. C5, 10µF/25V Tantal provides a smooth rise and fall of voltage and suppresses any injected charge which might penetrate via Q1. So, this LM317 circuitry to switch between 5V and 12V is ideal and need no change.

2N2907 is needed for 20pin socket only! Rather slowly 2N2907 is superior to 4051/4066 solutions, because of its slowlier switching characteristics. A faster switch could introduce fast spikes at output with the danger of making trouble at microcontroller. So, this circuitry is also ideal and does not need any change.

As Steve already stated: Don't start to experiment without having a scope. You must have a look at the pulses! Otherwise, you are totally blind!

Please keep in mind, that I do not at all recommend to use this self-built programmer with today's PC. Atmel people tested the programmer with 80386 processor at 33MHz. So, only a fool will believe, that his 500MHz Pentium PC will not make any trouble!!!
Also, what will you do, when AT89C51 is obsolete? Atmel does not recommend AT89C51 for future use!!

Kai