Abomination! or Painting a Mustache on the Mona Lisa meets the Island of Dr. Moreau

[MJL21193]

My VAS is running ~3mA - same as the Abomination. According to simulation this is as close to ideal as I can get it, distortion wise.

 

[MJL21193]

Speaking of output devices, what do we think of these: 2STA2121 and 2STC5949? These look pretty good to me and the price is roughly half the MJL21193/21194 with similar/better specs.

 

[jacco vermeulen]

Doesn't look all that much better then the A1943/C5200 to me for 70V and Up .
(No problem to buy the PNP ?)

OK, your time's up, my turn for the chickaseat.

 

[roender]

Guys,

With more than five output pairs and 0r1...0r15 Re, correctly biased in classB, as per Self definition, you do not need GNFB

Cheers,
M

 

[MJL21193]

Doesn't look all that much better then the A1943/C5200 to me for 70V and Up .
(No problem to buy the PNP ?)

OK, your time's up, my turn for the chickaseat.

Summers coming...

What?...220 watts as opposed to 150. Higher collector current and peak current. Faster than the MJL21193/94 and higher gain too. Worth looking at, methinks.

The PNP has a 2 month lead time from Digikey. Lots of NPN in stock.

 

[MJL21193]

Guys,

With more than five output pairs and 0r1...0r15 Re, correctly biased in classB, as per Self definition, you do not need GNFB

Cheers,
M

You are on to something... quickly done - feedback shifted to VAS, 3 output pairs. 0.1 ohm re's:

 

[MJL21193]

5 pairs, 200mA idle per device (distortion went down as idle current went up):

 

[MJL21193]

At 20K, same idle current:

 

[MJL21193]

Truly a work in progress...

 

[sajti]

Be careful with the DC offset! Better solution to use 2 pcs of 47kohm feedback resistors parallel, connected bot end of the Vbe multiplier.

Sajti

 

[MJL21193]

Good tip sajti,
This reduces DC offset:

 

[roender]

MJL,

You are a BSG fan, or only number six fan?

 

[MJL21193]

Baltar. In bed with both Tricia Helfer AND Lucy Lawless
Clearly, my hero.

I'm a bit surprised you dropped that NGFB bomb here...Thanks!

 

[AndrewT]

Hi,
post 251 looks nice, but it's going to need an almighty heatsink to get rid of that ~150W of output dissipation. Somewhere around 0.1C/W to 0.16C/W for each channel.
Check reactive load line against SOAR @ 80Vce and 90Vce

 

[roender]

Baltar. In bed with both Tricia Helfer AND Lucy Lawless
Clearly, my hero.

I'm a bit surprised you dropped that NGFB bomb here...Thanks!

Be sure it was not the only ace in my hand
As for Blatar, he was rehabilitated in the series final episode. IMHO, clearly he doesn't deserve that!

 

[MJL21193]

Hi,
post 251 looks nice, but it's going to need an almighty heatsink to get rid of that ~150W of output dissipation.

Yes, it's true, lots of heat to get rid of. Yet, there are degrees - how much distortion is acceptable? Dropping the idle current down to ~80mA/device has the total dissipation at 50 watts and the distortion at 1k is still less than .01%. Choices...
This raises the question: is it worth this to eliminate global feedback? Will the benefits be heard or are they just assumed? Is global feedback that bad?
Surely, for quiet listening the fact that the amp will be effectively operating class A should be an improvement.

Be sure it was not the only ace in my hand

Roender, you're all right.
Have you tried it with your amp? My sim plays it well - stable, low distortion. I'm now very curious how this will be in real life.

 

[MJL21193]

Triple EF version. Performance is no better (in simulation) than the CFP one, but I had good results with it on my big sub amp - stabilized it after I tried the CFP there.

I have reduced the OS to 4 pairs. Distortion looks good at reasonable idle current.

 

[MJL21193]

I have done some more playing with the HAVOC circuit and I've settled on this for now:

Sticking with the no global feedback (for the heck of it ).
I still need to do some fine tuning before I prototype. I've changed the outputs to the MJL4281A/4302A. These are in the schematic for simulation but I will probably use the MJW21193/94.

Here's how the 1k Fourier looks at ~90mA/output. I could live with this...

 

[roender]

You need at least 150mA/output pears, optimally 180mA

 

[MJL21193]

You need at least 150mA/output pears, optimally 180mA

Hi roender,
I have 90mA/device so 180mA per pair...or did you mean per device?

I'm getting good distortion figures in simulation with this idle current. I will need to build it and test it to see how it is in real life. I have the feeling it's going to be good.

I lowered the current through the CFP drivers and I also increased the current through the front end (and lowered the VAS current to ~2.5mA) and this shifted the harmonic content to higher H2 and lower H3.
Made a change to the CFP input - replaced the 2SA1123s with higher power 2SA1381. Too much dissipation for the TO-92.