Krell KSA5

If you enhance the output devices on the picture of the amp's inside of the KSA-5V brochure it reads the big M logo on two of them, definitely MJEs.
=> http://www.xs4all.nl/~hansvt/pdf/brochures/ksa5.pdf

Enough of the MJE150** in parallel makes an output stage of a decent size power amp, as the Continuum Duhh group did with their S5 model, 50 Watts of Class A.

Wayne, thanks for posting the schematic.
 
Figured i would add a bit on this. As a speaker amp, i did not think it
sounded very good. But as a headphone amp it is absolutely fantastic.
Just for fun, and because after years of trying to buy one, i'm finally doing
a diy version.

Real reverse engineered schematic of the amplifier
http://gilmore.chem.northwestern.edu/ksa5.pdf

virtually identical layout klone board single channel with
much more ground plane.
http://gilmore.chem.northwestern.edu/ksa5.jpg

2 stacked for balanced much more fun for HIFIMAN HE6 headphones
 
2 stacked for balanced much more fun for HIFIMAN HE6 headphones

Does the KSA5 use fully regulated power supplies? The pictures show enough parts for full regulation.

For headphones, has the brain trust agreed upon any favorable reasons for using a balanced amp topology which puts the headspeaker between two output stages, and hence not ground referenced?
 
any thread over at the other place for this mate? i'm interested in the bridged version.

Figured i would add a bit on this. As a speaker amp, i did not think it
sounded very good. But as a headphone amp it is absolutely fantastic.
Just for fun, and because after years of trying to buy one, i'm finally doing
a diy version.

Real reverse engineered schematic of the amplifier
http://gilmore.chem.northwestern.edu/ksa5.pdf

virtually identical layout klone board single channel with
much more ground plane.
http://gilmore.chem.northwestern.edu/ksa5.jpg

2 stacked for balanced much more fun for HIFIMAN HE6 headphones
 
thanks Spritzer, thought there might be, i'm used to calling it 'the other place' when on head-fi, force of habit. really great diy section that over there, much better than head-fi, far fewer nonsense tweaks and ********, but I try to stay away a bit since I have enough 'problems' as it is without adding more stuff I want to build....

nattonrice is a terrible influence, we tend to be a little competitive =) cant beat that man on casework though! hell, I cant beat anyone on casework
 
Hi Kevin,

Voltage reference (12V zener diode) has current bias of ~33mA...way to high. For low noise and low output impedance 1/3 (11mA) is sufficient. Opamp output sits at ~+/-19V7, not a single opamp would like it ... series 12V zener (or 6 x green LEDs) and bypass capacitor at opamp output is an efficient solution.

And one more thing (colleagues of mine with a lot more experience warned me abot that) ... although it does not seem so at first glance, the error amplifier - opamp operates open loop, resistor divider from regulator output to ground does not close the feedback loop... they only polarizes negative input opamp to the desired voltage.
One will need to close the feedback loop from opamp output (or one of the emitters in darlington pass transistors) to negative opamp input.

Otherwise, error opamp will work open loop and whole system will not be unconditionally stable (it can become sporadically unstable), and will largely depend on of the parasitic components that sees opamp negative input (stray capacitance, inductance etc.)

Edit: There aro no AC bypass caps over R8/R9 resistors...also RCRC filter in front of regulator would be nice.
 
Last edited:
correction, the voltage reference is a lm4040-2.5 which results in
16ma.

I'm pretty sure the 2 diode drops of the darlingtons close the loop
and the schematic as drawn matches the pictures of the krell power
supply exactly.

op27 with the supplies as drawn can work up to a 44v pre-regulated voltage.
 
Last edited:
All,
I just joined this site today, which explains why I haven't commented earlier on this thread. That, and of course the NDA I signed in order to get and keep my job at Krell.
To answer a few questions that would be easy for someone to figure out if they had the actual unit, yes, it uses 2 pairs of MJF15030-31 output devices per channel. The supply rails were fully regulated (outboard supply with cascaded Zener regulators). The servo op amps were SSM2131s (now obsolete) or OP27s, either of which could live with +/-20~21V rails. (Can't remember which, though.)
The unit did legitimately make more than 5 watts into 8 ohms, but its output impedance was higher than I'd want for driving a complex loudspeaker load. That said, back in the day, Joe Sammut (now of Pass Labs, along with Wayne and, of course, Nelson--a nicer bunch of guys you will not meet) had a KSA-5 at home driving some horns, and he told me it was his favorite Krell amp at the time.