Amplifier misbehaving when operated at B>1

[kipman725]

Hello I designed the attached amplifier only using spare parts hence the rather strange output stage. I have only used one gain stage such that the amplifier is unconditionaly stable, the open loop gain is aproximatly 60dB and rolls off -3dB/octave f(-3dB)=30Khz. Anyway my problem is that although the amplifier works fine in the configuration in the attached file as soon as I try and give the amplifier gain via a resistor divider to the (-) input of the LTP the output can't go below -6V I am preplexed!

Aswelll as this issue I am also soliciting further improvments, the amps power supply is fixed as are the output transistors, I have the ptype complement to those fets and lots of the small signal transistors. Also I have tip127 pnp darlingtons.

[kipman725]

Figured out the problem, when voltage divider on output the base of the right hand transistor wasn't driven low enough for emitters in the LTP to go lower than aprox -0.6V gain+level shift is needed after the LTP to stop this problem. My quick solution to the problem is attached, some resistor values need changing to reduce disipation in the grounded emitter amplifier and current sources need changign for actual transistors but seems to work

[AndrewT]

the first amplifier has no VAS (Voltage amplifier stage).

It is trying to use the first stage (voltage to current amplifier) to then drive the source follower and push/pull follower stage.
Note the two diodes cannot properly bias the output stage transistors. It will work in a sort ClassB generating high distortion that will almost certainly sound terrible.

The bias voltage generator must also be temperature compensated to help keep the output devices at near constant temperature (=constant bias current) and thus avoid thermal instability (=amplifier blow-up).

The output voltage can be used to feed R2 and allow much higher peak output voltages before clipping of the signal (less distortion due to clipped output).
Look up bootstrap, see ESP and DX.

[kipman725]

Ok I understand now I was thinking that the output of the LTP was voltage instead of current. This design clips at 50Vpp output (8ohm) and has good (simulated) THD figures below that. Idle power consumption is high but with the parts I have I can't see an easy way of lowering this. Anyway design attached.

[Elvee]

One remark regarding the use of fourier in LTspice: you should include the directive:
.options plotwinsize=0
Otherwise, the output data is compressed and generates distortion artefacts.
With the statement, the distortion figure of your circuit improves by a factor of more than 10.

[kipman725]

Thanks Elvee. I have been reading the LTspice topic in software and now use:

.options plotwinsize=0
.param Freq=1k
.param FFT=65536
.tran 0 {14/Freq} {4/Freq} {14/Freq/FFT}
.four {Freq} V(Vout) 5 5

It's weird that LTspice dosen't rember the compresion settings without a spice directive.

On the subject of my amplifier I was just calculating the power disipation of my 1ohm resistors and rearlised that I'm loosing alot of power in them. This is an issue as I only have large quantities of 1/4W resistors and I would need too many so I think I shall switch to a VBE multi as AndrewT suggested so I can use smaller emitter resistors. I genraly don't like having to adjust circuits once they are constructed hence the strange pereferance for diodes.

[kipman725]

Hello quick question, how do I find the saturation current for a transistor from its datasheet?

I tried using the graphs from:
http://www.datasheetcatalog.org/data...ola/2N3055.pdf

but Is was far too small causing the needed VBE to bias the transistors to be too high. I presume working backwards from the simulation will work but I would prefer to do this properly.

[AndrewT]

why do you need saturation current?
Do you intend saturating the output transistors during normal amplification or during some form of overload condition?

[kipman725]

Ok new design, I had some BD135 lying around so I can use those for the VBE multi. I did some tests at different temps and if I bias the amplifier once it has warmed up a little and mentain close thermal coupling I should be able to get away with no emitter reisistors (like how it's done in intigrated circuits). Rather living dangerously though as I will smoke my transistors instantly in the case of a fault.

Interestingly my large amount of gain means that distortion is at least simulated to be low with a pure class B output stage. I wonder if I can get away with that, unfortuantly I don't have the facilites to perform distortion mesurements other than checking the waveforms are ok on a scope once built otherwise I would try that first. There was much more harmonic content >3rd harmonic when operating in class B but it was all quite low level hence didn't add much to the THD.

[kipman725]

Originally Posted by AndrewT

why do you need saturation current?
Do you intend saturating the output transistors during normal amplification or during some form of overload condition?

to calculate the Vb needed for a particular quiesant current using the transistor equation. I found it in the spice model anyway:
IS=2.72053e-14