Transistors used now can be considered 'slow', but it should not rolloff starting at 10kHz. Each transitor should work to 1MHz.
Made series of measurements with various power supply voltage between 24vdc to 15vdc. Seen no difference in fr response. But again i see no rolloff.
Compared with and without mp bypass cap on output electrolyt, seen no difference. But no rolloff.
Thank you Adason, you have done a lot of heavy work and verified my simulations. I have never seen any rolloff with old caps; if any, particularly electrolytics, they just lose capacitance and go leaky.
I certainly see that loading is the greatest influence on performance, particularly the rise of all hamonic (though profile is unchanged) with increasing load. My own thoughts are that a HPA for 16R cans might be better done with a pmos slave, rather than bipolar. Then the master current is little changed over the range.
HD
I certainly see that loading is the greatest influence on performance, particularly the rise of all hamonic (though profile is unchanged) with increasing load. My own thoughts are that a HPA for 16R cans might be better done with a pmos slave, rather than bipolar. Then the master current is little changed over the range.
HD
So what changed physically between rolloff and non-rolloff measurements? No rolloff without load? or new components?
So Adason, regardless of the roll-off, is what you hear acceptable or rejectable. Keep in mind that extending the frequency range will also increase the distortion higher up. Does your hearing still conform to your youth or is it just that the measurement hinders your ultimate enjoyment? My hearing is stuffed, I am seventy and listening to my signal generator, I cannot tell if there is anything above 9 kHz. The fact that my amps have a flat response between DC and daylight makes no difference at all. My question remains, if you cannot hear it, does it matter? I like music as much or even more now than when I was 16 years old.
Well you have a task now, it is to find a solution to the amp roll-off. Normally it should not be difficult. One should investigate the GBP (Gain Bandwidth Product) looking at the whole schematic holistically. It may be that the BW is gain limited and requires using transistors that has higher hfe (not HFE). I will really be interested in your findings. Since I do not have access to an extensive lab like in the old days, it becomes difficult to theoretically tinker and subjectively correct things only by ear.
The very sharp knee at just about 20kHz with no bypass cap is very interesting, that is not roll-off , it resembles a brickwall filter.
Can you look at say 100kHz BW, so we can see if the down slope pops back. If it does you have total cancellation of those frequencies and is a phase inversion problem.
The phase inversion could be caused by the output capacitor and inductance of the headphones and form a tank oscillator which has very high Q. Try placing a series resistor and analyse the system again to establish which direction it is going. You may be better off looking at this on an oscilloscope not a spectrum analyser.
Nothing changed. That's why it remains a mystery. Except early measurement vs later measurement. Nothing, absolutely nothing changed on the board.
I made the brick power supply and used it for few days.
But the measurements were done with original very well stabilized power supply, just like in first measurement. For consistency, plus it can be adjusted, so it allows measurement at any power supply voltage.
There was rolloff, now there is none. I can not explain it. Hence my theory.
Michelag, do you still measure rolloff?