There is no disagreement between us on this, @EUVL . However, then most of the “projects” here would be banned, as they are in simulation stage only 😀.
The audio experience only comes with doing. And it is only from this experience that audio-suitable concepts, circuits, can be derived. So you can even see who does audio - and who does nothing;-)
You can place small miller caps across the VAS transistors of 22-60 pF. You must remove the 100 pF cap.
BUT you will not be able to optimise the value for best stability vs distortion (note very carefully how I wrote this) without a loop/gain phase plot.
🙂
BUT you will not be able to optimise the value for best stability vs distortion (note very carefully how I wrote this) without a loop/gain phase plot.
🙂
The current feedback is through R20.
C4 is on top of it. C4 doesn’t mean it is stable. Often it makes the stability worse. The close loop gain rolls off when C4 kicks in. The transition frequency is about 7MHz. I don’t think an average dude can build an amp with 7MHz bandwidth with discrete parts. Thus C4 has to be gone.
Traditionally, you can put a pair of Miller caps on top of U5, U6. You splits the C4 in half, that is 50pf. Put 47pf on top of U5 and U6 each. As said, the combination still results 7MHz bandwidth. If it doesn’t work, double the size of those caps.
C4 is on top of it. C4 doesn’t mean it is stable. Often it makes the stability worse. The close loop gain rolls off when C4 kicks in. The transition frequency is about 7MHz. I don’t think an average dude can build an amp with 7MHz bandwidth with discrete parts. Thus C4 has to be gone.
Traditionally, you can put a pair of Miller caps on top of U5, U6. You splits the C4 in half, that is 50pf. Put 47pf on top of U5 and U6 each. As said, the combination still results 7MHz bandwidth. If it doesn’t work, double the size of those caps.
Here is the new version we can discuss.
GainMargin 36 dB
PhaseMargin 68 Deg
R20 must be 2 Watt
GainMargin 36 dB
PhaseMargin 68 Deg
R20 must be 2 Watt
Hi
Where is the "Alessandro cap"? In this amp it would span from the collectors of U5,6 to the feedback node, which is the wiper of the trim pot. I thought all current-mode amps had such a cap?
Where is the "Alessandro cap"? In this amp it would span from the collectors of U5,6 to the feedback node, which is the wiper of the trim pot. I thought all current-mode amps had such a cap?
No, not all CFA’s have this. The nx-Amplifier uses that type of compensation, but it doesn’t work well if the VAS gain is high (as is the case if you use a beta helper for example) because you get a zero in the response at HF causing loop gain peaking. In the nx-Amp, like lineups amp here the VAS gain is low, so it would probably work. I used 68pF in my design. But, if you go down this route, you really need to look at the loop phase/gain characteristics to optimise it.
BTW it was first used in the Alexander Amplifier so I think it’s the ‘Alexander capacitor’ 🙂
https://www.analog.com/media/en/tec...tes/58052492001115525484056221917334an211.pdf
BTW it was first used in the Alexander Amplifier so I think it’s the ‘Alexander capacitor’ 🙂
https://www.analog.com/media/en/tec...tes/58052492001115525484056221917334an211.pdf
Last edited:
Biasing the differential with a diamond makes sense. Where exactly is the current to voltage conversion
typically use mirrors to do so. U3 and U5 current is miles apart and would be mirrored for conversion.
likely just voltage amplifier and that is why the feedback compensation isn't a problem
typically use mirrors to do so. U3 and U5 current is miles apart and would be mirrored for conversion.
likely just voltage amplifier and that is why the feedback compensation isn't a problem