Low distortion current driven Class-B output stage

[LKA]

some measurments (global feedback was not applied)
spectrum 1W/4R 1kHz and 10kHz

THDvF 1W 10W 50W /4R

 

[LKA]

When global feedback is applied, even harmonics are less reduced than odd ones. What could be the explanation?

1W/4R 1kHz, no global feedback
2H 0.0028% -91dB
3H 0.00077% -102dB
4H 0.00062% -104dB
5H 0.00051% -106dB


1W/4R 1kHz, global feedback
2H 0.00053% -106dB
3H 0.000026% -132dB
4H 0.00022% -113dB
5H 0.000030% -131dB


Distortion reduction by global feedback
H2 15dB
H3 30dB
H4 9dB
H5 25dB

 

[LKA]

Next step >>>
I'm probably going to make this buffer, Class B+G.
Many thanks to David (dave zan) and Toni (astx) for the Class G part.

 

[astx]

 

[LKA]

thx, I'll finalize the pcb soon

 

[LKA]

The dual PSU is done, arranged in parallel.

 

[LKA]

First measurments of the Class B+G prototype.

Outer PSU +-48V
Inner PSU +-24V
Yellow trace = output into 4R
Blue trace = upper/inner transistor collector voltage
Magenta trace = lower/inner transistor collector voltage

 

[astx]

Congratulations!
Looks promising!


You have used IRFPs 240/9140 for higher rail voltages?
what gate resistor values are you using here in your test sample?

 

[LKA]

thx
First I tried IRFP240/9140. 9140 was stable with 120R gate stopper, 240 only with 220R. Now I have IRFP140/9140 and 150R stoppers.

 

[Boden]

Where did you buy the IRFP140/9140?

 

[LKA]

I don't remember, it was some time ago, probably mouser.com

 

[LKA]

The distortion is bad when high voltage rails are involved. The transition is somewhere between 11-12 Vrms. The 100Hz modulation is also very strong above the transition line, but acceptable at higher outputs like 18Vrms. Tony, what do you think ? Where is the problem ?

 

[astx]

r60 and r61: the original design uses 20mA idle current.

have you used the same fast schottky types as our test sample?

 

[LKA]

"have you used the same fast schottky types as our test sample?"

no, I have used MBR20100CT

what about this one ?
https://www.tme.eu/Document/10a0166a026625918c85178c0a76cca8/DSA20C100PN.pdf

 

[astx]

During the transition the 100 Hz ripple from the low voltage power supply trigger the schottkys on/off. maybe the LV PS elko ground routing is too near the feedback line.

 

[LKA]

I solved the problem by increasing the gate stoppers from 150R to 200R. The oscillation was not visible on the oscilloscope. Now the distortion is also excellent above the transition line.

 

[astx]

Did this also solve the 100Hz problem?


BR, Toni

 

[LKA]

Yes, the 100Hz problem also disappeared. I soldered 150R directly to the mosfet gate and 51R is on the pcb (200R total). Besides that I also put 100R+47pF zobel between D and G, see picture.

 

[astx]

🙂
Congratulations! The IRFPs do not like longer legs!
Have you tried to short circuit the 0.1R resistors R37 and R38 (schematic post #63).
I have used 150R gate resistors as close to gate as possible. (see picture in post #1408).

AFAICS you are using only one MOSFET per rail so current sharing is not a problem. Should better the THD a little bit.

BR, Toni

 

[LKA]

R37,38 are not used. The THD is quite good 😀
(the transition is around 30W/4R)