Alpha Nirvana 39w 8ohm Class A Amp

Not sure where you get those part numbers? There was a lengthy discussion about V131 around post #1550. In the end it looks like either KSA992 or BD140 can be used here, if I understand correctly. And it is important for me to understand too, since I am about to solder V131 myself any day soon.
Ksa992 is almost equivalent to bc556 and you don't need the dissipation of a BD140 in the position of v313
 
I know Fab from his work; I will be very interested in his review!!
The single ingredient with my amps is always 'engagement'. I like my music to be involving......
I continue developing the AN in fact. Maybe I will do one for commercial; but a Class A is always very challenging because of the high dissipation and low power; a combination which makes it expensive to build, as you know well.

Hugh
I’d encourage you to do that Hugh, I think you can tell there is still a lot of interest in Class A, and not everywhere is as warm as Oz ! 😉
 
The BD140 has a lower ft than the KSA992 which makes it more desireable.
This relates to a tendency to set up a local oscillation between the BD140 and the pmos, since the pmos is a very high frequency component.
Since the pmos is operated by the BD140, the latter is a master device and if a very fast device it can oscillate, which can destroy the output stage.
This was pointed out to me by Haik, a very smart guy who has unfortunately left the forum. He has done a lot of work with the output stage of the AN, predated it for some years in fact, and I realised he was absolutely right. I have also put in a small cap between base and collector to snub out any oscillation, so you could either the A992 OR the BD140 as XRK has suggested and it will be fine either way.

The A992 (50MHz) is a similar speed device, but the BD140 is just a bit lower (20MHz). These figures are a bit difficult to find, and vary. But given the BD140 can pass 1.5A, I am assuming it is a much larger die, so it will be appreciably slower.

HD
 
I was just playing with Alpha Nirvana schematic, and decided to try to use JFET op-amp as the input stage.
I actually have never seen class A amp driven by an op-amp, but here it is 🙂
Cheap TL071 op-amp can be used here. This versions seems to be doing fine at 1kHz and 20kHz, and shows
much better sim results than discrete amp.
This is just a proof of concept. Speed, and stability have not been tested, but quick testing in spice so far looks promising.
Hugh, do you think this may work? Besides the fact that some "purists" may be put off by the op-amp...
The VAS transistor seems to be much more stable temperature-wise, than our previous attempt..

Edit: Actually I've seen class A amp driven by op-amp: Calor amp by Bora Jagodic.
 

Attachments

  • beta_nirvana.20v.asc
    beta_nirvana.20v.asc
    7.7 KB · Views: 114
  • beta_nirvana_1khz_fft.png
    beta_nirvana_1khz_fft.png
    9.2 KB · Views: 195
  • beta_nirvana_1khz_thd.png
    beta_nirvana_1khz_thd.png
    10.9 KB · Views: 194
  • beta_nirvana_schematic.1.png
    beta_nirvana_schematic.1.png
    18.7 KB · Views: 632
Last edited:
Minek,
I certainly looks like it would work well. Very good topology!
It adds a high loop gain in the opamp to the high loop gain from the VAS.
This gives huge levels of nfb, which should deliver very low THD but does it maintain the harmonic profile?
I thank X is right because the sonic overlay is dominated by the output stage, which could be taken out of the fb loop in fact.

HD
 
Hi Minek,
I used an OPA454 to drive the LuFo (40w) and the SuSyLu (100w). It’s capable of 90Vpp. Sounds quite good and distortion profile is dominated by output stage, which was SE Class A here.
https://www.diyaudio.com/community/threads/lufo-amp-39w-se-class-a-from-28v-rail.372679/
View attachment 1027532
X, this profile here is the profile of the Output Stage only, right?
In your Lufo setup, pre-amp and OS have separated feedback loops, and each of these 2 components have it's own "profile".
 
Minek,
I certainly looks like it would work well. Very good topology!
It adds a high loop gain in the opamp to the high loop gain from the VAS.
This gives huge levels of nfb, which should deliver very low THD but does it maintain the harmonic profile?
I thank X is right because the sonic overlay is dominated by the output stage, which could be taken out of the fb loop in fact.

HD
"Should deliver very low THD but does it maintain the harmonic profile?"

Of course not - it doesn't maintain the same profile, as I skipped all your magic profile-shaping RC tricks 🙂
Now we can play with the slope of harmonics.
I bet you can come up with something to shape the profile the way you like it.
The "bottom" of the new profile is at -170dB, which is exactly where I like it. This level is below measuring capabilities of QA401
analyzer, so in real life we gonna see -120dB when measuring it.
In my sims of AN (at full power), bottom of the "profile" started at -50dB and reached -75dB at 20kHz.
Now it's below -170dB all the way to 100kHz (at all power levels). Peaks of harmonics are all below -80dB, and sloping downwards
from -80dB at 2kHz to -100dB at 10kHz, while Thd stays below 0.002%.
I'm happy with this "profile", but I understand some listeners will prefer higher levels.
The discussion on the subject of perceived advantages of different "profiles" and the content of harmonics is philosophical, not technical
in nature. My position is that real music instruments produce huge amounts of harmonics, on many different frequencies, and as such -
the amount of harmonics added by a semi-decent amp, is like a drop in the ocean. But this is just an opinion.
I think I'm going to build this amp. Of course little bit more work with simulator is needed to make sure it's stable.
 
Last edited:
Maybe this should have it's own thread?
If you have an op-amp that can do it, why not use it as VAS too? Not sure what the output impedance is, but maybe it might even provide better drive (lower impedance) to the mosfet gate? Reduced number of components, -maybe it would increase the OLBW too?