Square waveform test/simulate problem

[mclarenpingu]

Hello All
Could you help me to solve some problem??
I have a question of Audio Power AMP square wave testing
My AMP is Leach 150 (but have a little modify)
The square wave test is a very easy way to know high and low frequency response
but not I have a problem in P-Spice simulation

Test condition:
input square wave: 1KHz, 5v to -5V
Output waveform: 1KHz, 5.xxV to around -VCC (-50V)

this is confuse me, if I use DC to simulate,
the result is
input DC 5V, output almost DC 5V
input DC -5V, output almost DC -5V

that is when I use DC to simulate, the output voltage will follow input (because DC feedback)
but if I use square wave to simulate, when the input toward low, the output will pull down to -VCC
that is very strange
I have some picture I will upload tomorrow, because now my computer can use P-spice
thanks everybody to look my question, please help me!!!!

Thanks very much!!!

[unclejed613]

try 1Vp-p instead of 10Vp-p. you could be getting your input stage so far into saturation that it has unpredictable outut currents. besides that, if you have an inductor in the output rail, a square wave signal will have ringing on it anyway. the best way to test a SPICE model for frequency response is to do an ac sweep, using the .ac simulation command. using a square wave to check frequency response is a "quick-and-dirty" method used with capacitive and transformer coupled amplifiers, and doesn't really tell you much with a solid state design, other than the slew rate, or how much the output choke rings. you wouldn't want to apply such a large input signal to an amp in real life, and you shouldn't in a sim either. your amplifier should have a gain of about 30 if you have 50v rails, so you reach clipping at somewhere around 1.5V input.

[AndrewT]

Hi,
I think your version of the Leach has a DC blocking capacitor in the NFB loop.
This cap rolls of the bass response of the amplifier to give a DC gain of 1 (+0db)
If you apply an AC signal, the amplifier will mulitiply the input signal by the gain to give an output signal that is either clipped or within the range of voltage the amp can handle.

If you input a DC signal then even when the DC blocker on the input is bypassed/omitted then the DC gain of the NFB blocker ensures that the output signal matches the input signal.

That leaves a problem. The impedances seen by either side of the input differential amplifier (LTP) are not matched when mixed AC & DC coupling is used, leading to larger variation in output offset than is necessary. It also allows excessive DC to pass to the speakers if a fault develops in the source equipment.

[mclarenpingu]

Hello
Thanks Master unclejed613 and AndrewT help
those two ways are reasonable and functional
if I change the input square wave to 200mV Peak-to-Peak, the output will almost normal, that mean when I input too large wave
the BJT will saturatoin

and accorading AndrewT's method, this is also reasonable,
My Leach150 have a blocking capacitor in AC NFB loop, since it call AC NFB loop
but in we input a DC voltage, the AC NFB will provide zero voltage to feedback BJT
so the input BJT and feedback BJT can't following each other, that make the output voltage is very strange.
I think I may use a fast clock square wave to simulate, or bypass the AC feedback capacitor, or the square wave simulation
have some problem in this architecture natural

thanks a lot,!!!!!!!!!!!!!!!!