Differential VAS topology

[danny92]

Hi everyone,

I've made this circuit, because I wanted to test the performance of a differential VAS amplifier. This topology seems to have lower distortion, and lower DC offset at the output. The major problem I've noticed with this design in particular is the lowered output excursion, caused by the Wilson Current mirrors. This design is based on one design of Bob Cordell, but it's not the same. In simulations the behaviour is very linear and has low distortions. I don't know if this amplifier with added S/C protection, DC protection and thermal shutdown could survive in the real world.

Thank you very much for your attention
Best regards,
Daniel Almeida

 

[Rafael L]

I think you do not properly simulating this distortion, 0,4%-20Khz is much THD

Try this:

http://www.diyaudio.com/forums/software-tools/101810-spice-simulation-82.html#post1743578

 

[danny92]

No, that distortion seems real, which is really bad, all the amplifiers I've tried to simulate have similar or worse THD performances, including a circuit directly from Bob Cordell's website, even so this is one of the designs with less THD at 20kHz.


Best regards,
Daniel Almeida

 

[Rafael L]

Try this:

 

[danny92]

Wow that gives me a THD of 4%, even higher, thank you very much for your help, the problem is bigger than I thought.

Best regards,
Daniel Almeida

 

[Rafael L]

Yes the amp was clipping

THD is 0,5%

I made some changes, now with this THD is 0,02%-20KHz

 

[danny92]

Right, so you've added degeneration resistors at the LTP, and lowered the CMiller capacitor, and the result is very good, the only problem I see is the 0.1 VDC at the output, but the amplifier is really stable and has a very good phase margin, CMiller could be decreased to 68pF or 56pF and I think that the phase margin will remain acceptable.

Thank you very much,
Best regards,
Daniel Almeida

 

[danny92]

Can you help me solving the same problem with MOSFET2 please,

Best regards,
Daniel Almeida

 

[Rafael L]

I do not know the problem, got 0,023% 20KHz which is a lot for this circuit. Need to review the models Mosfets, I know little about Mosfets.

For you to loop gain, ground the input and plot the graph (right mouse - add trace) -V(Vout)/V(a) and watch your phase margin in 0 dB.

 

[Walkalone]

Very interested in this topic.

Try circuit like Mongrel CX1.3VB U, simple change output stage with SF (Source Follower). And add an active CCS for Vas.
An active CCS for Vas, maybe, not necessary. Vas's Iq controlled by Iq of Ips' CCS.
And I think Wilson Mirror for Ips too.

Pic show THD @20KHz is very low.

 

[danny92]

Hi everyone,

I've made this improvement using TMC (Transient Miller Compensation), with this compensation topology, the THD1 (1W-150W) is < 0.0001%, the THD20 (1W-150W) is < 0.003%.

Best regards,
Daniel Almeida

 

[AndrewT]

I am trying to work out why Q7 & 8 are PNP? I'm lost.

 

[danny92]

Thank you for replying AndrewT

They are used to get a bias voltage to the differential pair of the VAS of nearly 3*vbe = 2.1 V, because at the output of the IPS/TIS the voltage is nearly 1.4V, and to prevent Q15 from entering in saturation the Vce voltage should be higher than 0.7V, in this case the Vce of Q15 is close to 1.2-1.3V to prevent the saturation. The 39k resistor allows a collector current of 1mA in transistors Q7 and Q8.

Best regards,
Daniel Almeida

 

[Walkalone]

Well, it's darlington differential Vas. What's about cascode?