Voltage error in CFP VAS stage

[northumber82]

Did you breadboard the circuit before making a pcb, or just do simulation? I've always benefited from prototyping first.
When laying out the pcb, did you use a net list, or route manually? Errors can happen with manual routing.

Yes, built, and worked with old PCBs and breadboards. I ever use autorouter. In fact now it seems very strange that it doesn't work anymore.

Check all npns are npns, and pnps are pnps, and the pinouts.
The circuit is fine but it seems you have may a wrong gender transistor.

HD

Checked all pins with the multimeter diode checker on the PCB and all seems to be ok, checked even after soldering and that's ok again.

Isn't the current in Q183 much too low? (There's no current through R197)

Probably what you see is a mistake of a past simulation, and it seems that it doesn't pass any current in that resistance, but it actually works normally.

You might try some resistance in the Emitters of Q193 and Q194. Two transistors with the same part number will simulate 'exactly balanced', which is almost never the case when you build it.

With CCS loads throughout, the intrinsic V-gain is sky high. Any small imbalance and it will slam as far as it can go since there's minimal resistive load to hold it back.

Glad you're building .. and not just simulating😉

Rick

Generally, I simulate using the montecarlo, with tolerances for resistances of 5% and transistors of 50% even (to be sure that the circuit works in any case).

I tried to do two things: add two 100 ohm 1% or even 470 ohm 1% resistors to the emitters, or even replace those resistors with two other transistors, turning it into a wilson mirror.

Which mirror (Wilson), according to the formula (with transistors with beta greater than 300) should give a better tolerance of the resistances: if I'm not mistaken, calculated of 0.17% of current difference, if the transistors were for example 300hfe and 400hfe with 100hfe of difference.

But the situation does not change whatever.

 

[Mark Tillotson]

Are you sure its not oscillating?


I see no Zobel network on the output to load it at HF

 

[northumber82]

Are you sure its not oscillating?


I see no Zobel network on the output to load it at HF

Open loop gain is 160° phase margin at 0dB (10 montecarlo give all same results).

Verified with oscilloscope, but I see only DC without any type of oscillations. Tested any point.

As I said, it's such a strange situation that I can't explain it. The only problem may be the PCBs, otherwise it cannot be explained.

 

[Mark Tillotson]

Time to map out the traces as if its an unknown circuit, write it down, compare to what you'd expect.

 

[Rick PA Stadel]

Maybe it's too high a value for the driver Emitter resistors, R193 and R202? At 2k2 it will only draw away 270 microA before the OP turns on hard. Yet the VAS CCS offers 11mA (an entirely reasonable value), so the drivers, Q184 and Q190 would have to be real turds to be able to find someplace to go with most of those 11mA without slamming the OP's on hard.

Also, most designs would have a single resistor between the driver Emitters, rather than 2 with each contributing-to/connected-to the output.

If Mark's '.. map out the traces ..' try (more good advice) doesn't set it straight, maybe try lifting just the Base lead of each output transistor.

Congrats on the MonteCarlo-ing -- I've yet to figure out how to do that.

Cheers
Rick

 

[northumber82]

Maybe it's too high a value for the driver Emitter resistors, R193 and R202? At 2k2 it will only draw away 270 microA before the OP turns on hard. Yet the VAS CCS offers 11mA (an entirely reasonable value), so the drivers, Q184 and Q190 would have to be real turds to be able to find someplace to go with most of those 11mA without slamming the OP's on hard.

Also, most designs would have a single resistor between the driver Emitters, rather than 2 with each contributing-to/connected-to the output.

If Mark's '.. map out the traces ..' try (more good advice) doesn't set it straight, maybe try lifting just the Base lead of each output transistor.

Congrats on the MonteCarlo-ing -- I've yet to figure out how to do that.

Cheers
Rick

Lowed down resistors from 2.2kohm to 220ohm or 100ohm. Nothing changes. In these days I will assemble one on a perforated card.

 

[Rick PA Stadel]

OK -- then we (really, mostly you) still have some work to do.

Looking forward to the result.

Best Regards,
Rick