Krill - The Next Generation

[andy_c]

Below is 1 kHz (left), 20 Hz (right), and the simulation file.

[andy_c]

Here's the schematic for those who don't use LTspice.

Edit: Output current was close to 3.5A peak for each sim. Output stage bias current was 262 mA.

To run the sim at different frequencies, just change the freq .param statement. It will do the same number of cycles regardless of the frequency.

[Steve Dunlap]

The other interesting thing is to show the base current in each output device going all the way to zero on alternating half cycles.

Base current of NPN output. I inserted a 10 ohm resistor to view the voltage across it.

[Steve Dunlap]

Base current of PNP output. Same 10 ohm resistor.

[andy_c]

Output device base currents at 1 kHz (left) and 20 Hz (right). Also shown are driver Vbe's for 20 Hz, showing that they turn off.

[scott wurcer]

In the idealized case crossover distortion has nothing to do with the output devices turning off. The logarithmic Vbe modulation between several amps and several micro amps is enough. I still don't get the fixation with "does the output device turn off".

[ostripper]

wow , andy and steve on the same page. We still come so close to 0 , what we are attempting is to "rip " the carriers from the bases of the "half" that just switched off ?? from what I have seen, the "krill" way hopes to eliminate higher order distortion by not having a Xover point. Short of biasing your AB into A and using MOSFET's , This does not exceed the new triple I now listen to.

This stored charge manifests itself as a slowly decreasing collector current

that is what gets "sucked" by my CCS's on the"off" cycle , the T3T triple works very good.. almost no H3 - or above .. at all.

Output device base currents at 1 kHz (left) and 20 Hz (right). Also shown are driver Vbe's for 20 Hz, showing that they turn off.
Attached Thumbnails

Obvious , it is still going negative. I have not gone for that result. I have instead "shot" for the best transition at Xover , resulting in the lowest H3/5/7 ... with a typical EF..
os

[Steve Dunlap]

Originally Posted by G.K.

Alternatively, you could just remove the driver bias resistor from the plain double EF (with standard Vbe multiplier bias) for the same behaviour.

Amp from post 3 with driver bias resistor removed.

[andy_c]

Originally Posted by scott wurcer

In the idealized case crossover distortion has nothing to do with the output devices turning off. The logarithmic Vbe modulation between several amps and several micro amps is enough. I still don't get the fixation with "does the output device turn off".

Exactly. I think this is just a "does it do what it says on the tin" issue.

[Steve Dunlap]

Originally Posted by andy_c

Output device base currents at 1 kHz (left) and 20 Hz (right). Also shown are driver Vbe's for 20 Hz, showing that they turn off.

Aren't simulations fun?