Sorry.. but sometime i think you make a fun here.
I didn't have a deal with PMA5-83-2W ever and i don't know it's working conditions.
On your schematic i see U2 that acts like a voltage source and a voltage divider wich is formed by R6 and Rload.
Iload is limited by R6 and U2 (TLV3701) only.
TLV3701 has maximum output current +/- 10mA that is bigger than twice Ig1=4.5mA. By the case pay attention to TLV3701 power supply. By the datasheet it is +/- 8V only.
U1 acts like a zener diode. The one has a dynamic resistance. This means one's sinking current depends on an applied voltage. Look at the figures 6-3 and 6-4. After appiled voltage reaches 1v the sinking current starts rising gradually and at 2.5v goes up vertically. According this plot U1 may sink more than 100mA.
As you wrote upper you need 2.5V at the load. Ok. (8-2.5)/1000 = 5.5mA. That is bigger than 4.5mA. 1mA extra current will be sinked by U1.
If you look at the datasheet you will see the gate current (IG1) may reach up to 4000uA.
With typical gate voltage (VG1)=-0.8V it gives us 0.8/0.004 = 200 ohms resistance.. not 1M.
Разносите территориально.
Следите чтобы токи не заходили на чужую территорию.
Экран над аналогом с хорошим заземлением. Если есть dc/dc - тоже экран.
Заливка платы землей и прошивка переходными.
Схема.
Дифференциальная пара CLK не имела опоры на землю с одной стороны, поэтому импеданс был больше требуемого.
Как следствие отражения сигнала и многократное срабатывание.
I'm sorry.. I was wrong.
The time constant of R1C1 is too large for 8uS pulse width. The Q3 is still closed at 9uS.
Try to decrease C1 value either increase the voltage pulse width.
It seems you use a PNP model of BJT instead NPN one..
At very start of simulation current flows via R2 and then one gradually reduces with charging of C1. It's possible if collector and base of Q3 forms a conductive diode.
