Hi there,
I am playing with the amplifier circuit below in LTspice.
It seems quite stable when looking at gain/phase margins. However, when I feed it a square pulse, the result is this:
I don't really know how to interpret this, is it bad? if so, what could be done about it?
Looking forward to your comments, also in general about this circuit.
I am playing with the amplifier circuit below in LTspice.
It seems quite stable when looking at gain/phase margins. However, when I feed it a square pulse, the result is this:
I don't really know how to interpret this, is it bad? if so, what could be done about it?
Looking forward to your comments, also in general about this circuit.
Hi!
Just add a low pass filter at the input to limit bandwidth to between 50 to 100kHz.
You can start with 3.3nF and check FR and this overshoot.
It's always good to limit bandwidth at the input so as to avoid T.I.D. (transient intermodulation distortion) when signal moves fast than the amp slew rate.
Just add a low pass filter at the input to limit bandwidth to between 50 to 100kHz.
You can start with 3.3nF and check FR and this overshoot.
It's always good to limit bandwidth at the input so as to avoid T.I.D. (transient intermodulation distortion) when signal moves fast than the amp slew rate.
Maybe think about increasing the R26 so as to not load too much what will feed this amp.
Maybe instead od 680ohms you change to 4.7k and reduce the capacitor to 470pF.
Play with these components until you get a good result.
Maybe instead od 680ohms you change to 4.7k and reduce the capacitor to 470pF.
Play with these components until you get a good result.
👍 that's the first thing to try really. This is with 1.5nF
Its the input source that is the problem. You need the square wave to start at 0 volts.
This will copy and you can move it to your sim. Set freq and amplitude in the .par line. Open both sims together and use the copy function to draw around the generator and then move to the other sim and 'drop' the generator into place.
Ok...Mooly was faster,
but nevertheless
a further *.asc
I had to change some parts (same type, available in my lib)
100p in output stage removed
added 1µ to Bias Transistor
Quiescent current of the outputs is about 300mA, unnecessarily high, I think. With R5=1k2 you get about 50mA
but nevertheless
a further *.asc
I had to change some parts (same type, available in my lib)
100p in output stage removed
added 1µ to Bias Transistor
Quiescent current of the outputs is about 300mA, unnecessarily high, I think. With R5=1k2 you get about 50mA
Last edited:
Mixing other transistor types can also reduce overshoot. e.g. using MJE340/350 for Q5/Q13 instead of BD139/140.
Or keeping BD139/140 but using other output transistors, like 2SA1294/2SC3263, just to give an example.
At least in the simulation.
models included in the *.asc-File
Or keeping BD139/140 but using other output transistors, like 2SA1294/2SC3263, just to give an example.
At least in the simulation.
models included in the *.asc-File
Thanks everyone for you replies!
@Mooly and @catd, your knowledge of LTspice is obviously way beyond mine. The PWL voltage source and everything that comes after that is abacadabra to me. I will have to study! but now at least I know about the input filter to limit HF.
I suspect in real life the bottom Sziklai pair at least needs the 100pF, that is what I see in other amps, like the P3A from Rod Elliott.
About the idle current, I have built the Hiraga 20watt class A and now wanted to come up with something more powerful and running less hot. 300mA is a lot, sure it could be less, but then you get a bit higher THD at low power.
I will try different transistors as you suggest.
@Mooly and @catd, your knowledge of LTspice is obviously way beyond mine. The PWL voltage source and everything that comes after that is abacadabra to me. I will have to study! but now at least I know about the input filter to limit HF.
I suspect in real life the bottom Sziklai pair at least needs the 100pF, that is what I see in other amps, like the P3A from Rod Elliott.
About the idle current, I have built the Hiraga 20watt class A and now wanted to come up with something more powerful and running less hot. 300mA is a lot, sure it could be less, but then you get a bit higher THD at low power.
I will try different transistors as you suggest.
Could you please explain this statement?
When I raise the supply voltage from 35 to 36 volts, the idle curent goes up 4%. Not a huge difference. The value of R1 and R6 can be adjusted to set VAS current and DC offset. In fact, at first emitter resistors were included for Q6 and Q12, but I took them away again. Real life will show if the thing is stable enough like this..
real life is measuring a 20 centigrade rise inside an amp enclosure after one hour.
Hold on a minute here.
You are simulating it. What does it do in real life? Some ringing is normal and fine in most real amplifiers, look for sustained oscillation.
Simulators available to normal folks do not do transients well. They do not take into effect real world components variations or wiring artifacts, stray capacitance. So even if you got a perfect response in the simulation, your real amplifier may be a radio station depending on layout. It probably will not get the low distortion promised by a simulation. It may operate perfectly fine.
I just tested a real amplifier with a square wave that gave the same real trace on a real oscilloscope. That amplifier is perfectly stable, low distortion too.
You are simulating it. What does it do in real life? Some ringing is normal and fine in most real amplifiers, look for sustained oscillation.
Simulators available to normal folks do not do transients well. They do not take into effect real world components variations or wiring artifacts, stray capacitance. So even if you got a perfect response in the simulation, your real amplifier may be a radio station depending on layout. It probably will not get the low distortion promised by a simulation. It may operate perfectly fine.
I just tested a real amplifier with a square wave that gave the same real trace on a real oscilloscope. That amplifier is perfectly stable, low distortion too.
I know, but the simulation can give me at least an indication wether or not it might work in reality. I will try to build this on a breadboard (when I find the time..) and I guess if it is stable there, it will be anywhere.
Maybe I will use shottky diodes in stead of emitter resistors for the VAS, then the emitters will not be degenerated for AC.
I have found that if the basic schematic is good, I can design the PCB and only have to adjust component values - unless I mess up the PCB layout.
From what I see, it should work fine. But to look at a sim for details like ringing is wasting your time. Everything will be different in real life. Breadboarding is also a waste of time with an amplifier. You're right, if it works on a breadboard it will work anywhere (within reason). Going to a dedicated PCB changes everything, so why bother?
For quick prototypes I use blank, copper clad PCB. Drill the component holes where you want them, then "connect the dots" with a resist pen, bluing ink or nail polish. Use your favorite acid to dissolve the excess copper. You can do double sided this way easily. I cut the PCB to size before doing anything. These days I can use the PCB payout programs, then print onto transfer paper and iron the pattern down. Fix defects with a knife or resist pen/nail polish or whatever. Now you have a quick and pretty prototype PCB.
If you made mistakes, no problem. Develop the final layout on that prototype PCB, now you can correct your pattern and send it out for professional PCBs if you want. Or, print it out on transfer paper again, transfer to the PCB material and etch that one. Fast, cheap and it works.
From what I see, it should work fine. But to look at a sim for details like ringing is wasting your time. Everything will be different in real life. Breadboarding is also a waste of time with an amplifier. You're right, if it works on a breadboard it will work anywhere (within reason). Going to a dedicated PCB changes everything, so why bother?
For quick prototypes I use blank, copper clad PCB. Drill the component holes where you want them, then "connect the dots" with a resist pen, bluing ink or nail polish. Use your favorite acid to dissolve the excess copper. You can do double sided this way easily. I cut the PCB to size before doing anything. These days I can use the PCB payout programs, then print onto transfer paper and iron the pattern down. Fix defects with a knife or resist pen/nail polish or whatever. Now you have a quick and pretty prototype PCB.
If you made mistakes, no problem. Develop the final layout on that prototype PCB, now you can correct your pattern and send it out for professional PCBs if you want. Or, print it out on transfer paper again, transfer to the PCB material and etch that one. Fast, cheap and it works.
It's a good idea to have a minimum value of resistor at the VAS emitter so as to reduce the gain, improve linearity and stability.
When you connect the emitter directly to ground (pure common emitter) you move towards extracting the maximum gain of the transistor and the gain of this stage gets more dependent on HFE, which is not linear and is dependent of temperature and each transistor unity.
We know negative feedback controls everything, but the more linear and less dependent on component and temperature the better.
I follow the procedure to first simulate the amp in open loop and see if everything works fine with minimum linearity.
Than I close the loop and move forward to final adjustments.
...quite a lot.
I included temp in the attached asc-file
Run the sim and plot a outputs emitter resistor current.
assumed: all BD139/140, output- and the bias regulator transistors are mounted on the same heatsink.
Main heat source is the output transistors, and there is some temperature gradient towards the others.
The formulas used are a rough estimate, you can play around with it, but you can see that there is some need for changes to the circuit.
T1...T3 : Heatsink temperatures
T4 : air temperature in an enclosure
Between 25° and 125° junction temperatures the output transistor bias current rises from 40mA to 1.7A
Last edited: