Since the signal is between base and emitter then what is the practical influence of that BC junction capacitance non-linearity that needs fixing?the non-linear base-collector junction capacitance.
But ok, since this discussion is straying away from the main topic then from my side we can close it here - it is unlikely that I will become an expert on that level 😉
This is getting beyond my expertise, but non-linearity can cause unexpected gain resulting in oscillation. (I am an electrical engineer, but my field is computer architecture. I can tell that quite a few members here specialized in signals or devices).madis64 said:Since the signal is between base and emitter then what is the practical influence of that BC junction capacitance non-linearity that needs fixing?
Ed
Oh, just imagine, emitter really tracks base signal, no BE capacitance recharge.Since the signal is between base and emitter then what is the practical influence of that BC junction capacitance non-linearity that needs fixing?
But BC capacitance are at full rail potential, exhibit full output signal swing and need to be recharged.
You can get a few ideas here https://www.pcbway.com/project/shar...audio_power_amplifier_a_rebirth_4a7881e9.html
So these caps serve as reservoirs for BC capacitance?But BC capacitance are at full rail potential, exhibit full output signal swing and need to be recharged.
No, they act as a brake.So these caps serve as reservoirs for BC capacitance
Slowdown driver stage.
I put some notes together on EF instability based on Cordell's book and my own experience. Its not perfect, but it does provide some general guidelines about how to make an EF3 bulletproof.
Here is the oscillation WITHOUT EF3 instability precautions :-
I used 10-15 nH PCB trace inductances in this model, whereas Cordell used 35nH in his which will in general result in lower oscillation frequencies. On small signal emitter followers, I've had problems at 180 MHz, and on a beta helper in a VAS circuit, 20-40 MHz. My first EF3 in 2006 oscillated at 4 MHz (used MJ21193/4 output devices) which was cured with base stoppers in the output devices and a 22 Ohm and 1nF driver base damping network. I've not had a EF3 problem since then, using the simple techniques 1-4 outlined in the 1st graphic above.
Here is the oscillation WITHOUT EF3 instability precautions :-
I used 10-15 nH PCB trace inductances in this model, whereas Cordell used 35nH in his which will in general result in lower oscillation frequencies. On small signal emitter followers, I've had problems at 180 MHz, and on a beta helper in a VAS circuit, 20-40 MHz. My first EF3 in 2006 oscillated at 4 MHz (used MJ21193/4 output devices) which was cured with base stoppers in the output devices and a 22 Ohm and 1nF driver base damping network. I've not had a EF3 problem since then, using the simple techniques 1-4 outlined in the 1st graphic above.
Last edited:
It’s a “ground is not ground” problem. See the parasitic inductances in Bonsai’s post.Capacitor to ground would do the same?
Thanks for doing that. I knew about base resistors and collector decoupling.I put some notes together on EF instability based on Cordell's book and my own experience. Its not perfect, but it does provide some general guidelines about how to make an EF3 bulletproof.
By now, I have seen on this board examples of everything that can go wrong in amplifiers. I look back on my design as having avoided all pitfalls, mostly by skill but sometimes by luck. 🙂
Ed
How do you measure that in real life?I've had problems at 180 MHz, and on a beta helper in a VAS circuit, 20-40 MHz.
Yes.Capacitor to ground would do the same?
If you can keep it to act as a capacitor in the whole its network. And namely this remark accurately provides us to the solution of placing it straight to the leads of the BC junction.
And if I happen to have a 20/40 Mhz scope I would not notice anything?With a scope
That is likely good enough. Oscillations tend to be gross - the voltages appear wrong and the amplifier is getting hot! It may even be jamming radio receivers.
Ed
Ed
Depends if it is digital or analog. 20 MHz analog scope will show a 100 MHz signal. Digital will not.
Then it would be easiest to judge their presence by power consumption (rail current draw)?Oscillations tend to be gross -
Current draw is only a clue. Use the oscilloscope.Then it would be easiest to judge their presence by power consumption (rail current draw)?
Ed
- Home
- Amplifiers
- Solid State
- Preventing Sansui AU-X1 output stage oscillation burn out