KRELL KSA50 power supply - kind request of info.

[beppe61]

Dear Sirs,

I would like to ask kindly to anyone who knows the following data about the original KRELL ksa50 (first series) power supply:

1) number of transformers;
2) VA rating of the transformer/s;
3) number of filter capacitor and their value (uF/V)
4) brand of the transformer and capacitors.

As you can well see I am quite interested in the PS of this little wonderful beast.
Moreover from what I have gathered the first series was the better performing of all.
Comments on this last statement would be very welcome and much appreciated.

Thank you so much.

Kind regards,

beppe61

[AndrewT]

Hi Beppe,
I think there were two 400VA transformers inside the stereo KSA50 amp wired as true monoblock.
Have a look at the KSA Wiki for more info.

[beppe61]

Originally posted by AndrewT
Hi Beppe,
I think there were two 400VA transformers inside the stereo KSA50 amp wired as true monoblock.
Have a look at the KSA Wiki for more info.

Dear Mr. Andrew,

thank you so much for the extremely interesting link.
I have found all the info I needed.

Thank you sincerely.

Kind regards,

beppe61

[beppe61]

Originally posted by AndrewT
Hi Beppe,
I think there were two 400VA transformers inside the stereo KSA50 amp wired as true monoblock.
Have a look at the KSA Wiki for more info.

Dear Mr. Andrew,

please excuse me if I take the liberty of asking your opinion.
I have listened to a KRELL KSA50 recently and its powerful bass impressed me much.
I see that it had 2 400VA toroidals and four 40.000uF high-grade caps.
Do you think that its prodigious bass can be related more to the work of the transformer or of the capacitors?
I ask you this because I am always thinking of upgarde an amp of mine and I do not know if it is better to start with the transformer or with the capacitors.

Thank you very much indeed.

Kind regards,

beppe61

[AndrewT]

Hi Beppe,
I have not started building my KSA Klone yet.
I have a few thoughts, in decreasing order of merit, on why it produces adequate bass:-

1. good current capability.
2. only two RC time constants controlling bass roll off, NFB and PSU.
3. a carefully balanced set of design compromises to maximise use of mechanical and electrical resources, ensuring no weak points.
4. multiple output devices and low output impedance.
5. ClassA
6. 4 & 5 combine to ensure current delivery from 1.

BTW 40mF per rail is about 6mF per Apk of output current into 4ohms. This is about twice what I recommend for a classAB amp.

[pinkmouse]

Just as a quick addendum to Andrew's comments, that I agree completely with, the bass gets significantly better in my clone as I turn the bias up.

[AndrewT]

Hi pink...,
your comment confirms Still4given re bias and bass.

WHY?

Did I really say Krell and design compromise in the same sentence. I should be excomunicated. Oh I didn't actually commit that sin, just implied it. Phew!

[K-amps]

Agree with Al and Andrew, just to elaborate on point 4.

By virtue of being class-A, the voltage rails are very stiff (or pseudo regulated by virtue of hi/ constant loading).

In lay terms, This means as opposed to a class AB amplifier which has unloaded rails of 45vdc which drops to 37vdc underload, this reduction of vRails as the amplifier is trying to squeeze out a kick drum can make it appear, "soggy".

On the other hand a KSA-50 with unloaded rails of 40vdc (no bias) and fully loaded (full class-A bias) rails fo 37vdc will not droop rails by more than 1-1.5vdc hence the bass will be more consistent and substantial relatively.

[amplifierguru]

Building Better Amplifiers

Scanning through the volume of promotional literature from many 'high end' manufacturers, one could be easily forgiven for concluding that sonic perfection in a power amplifier is only approachable at huge expense in the form of massively overbuilt 50kg behemoths with enormous power supplies, Class A, 20 KHz + open loop bandwidth, and minimal or no global negative feedback (NFB) - and that anything less would severely compromise either sonic performance or available power or both! However, it has subsequently been shown (Ref 1), that the premise on which these designs were conceived is flawed and that far more elegant solutions are possible with clearly audible, and measured, performance benefits (Refs 1,2,3) as well size, weight, cost and, far from the least important, environmental advantages.

Evolution of the Modern-Day Dinosaur

Amplifier design changed direction in the mid 1970's after a new form of distortion was 'discovered' (Ref 4) supposedly resulting from the application of large amounts of global loop NFB. Dubbed transient intermodulation distortion (TIM) this was subsequently displayed as being not stimulated by conventional measurement techniques and thus previously unassessed in the design process. Many designers and manufacturers responded to this by producing low TIM designs with little or no loop NFB and 20KHz+ open-loop bandwidths to eliminate the internal overshoots of conventional designs, unaware that these overshoots are a normal function of NFB and only result in TIM when an internal stage of the amplifier begins to severely distort or limit a signal approaching it's slewing rate limit. (See Fig 1). Loop NFB was singled out as the culprit to be alienated - akin to banning motor vehicles to eliminate road accidents! As a direct consequence of the preoccupation with this one trendy distortion, and the assumed inacceptibility of loop NFB as a design tool, these designs suffered in many performance aspects, primarily poor power supply rejection and high levels of steady state wideband distortions (THD ansd IMD). Further development necessarily centred around reducing power supply impedance and linearising the output stage, with massive high regulation transformers and capacitor banks, and Class A multiple bipolar output stages at enormous size, weight, cost and environmental penalty! However, these designs still suffer from poor IMD performance in the aurally sensitive mid-range, while crosstalk and a high sensitivity of the bass performance to power supply capacitance result in a narrowing of the sound stage and an unnatural, often punchy bass particularly in a.c. coupled designs. Some designers even went to fully regulated individual channel power supplies to reduce crosstalk and retrieve detail and soundstage rather than redesign the amplifier for better power supply rejection. MOSFETs which have substantial fundamental advantages over bipolar transistors, could not be used as their one big shortcoming, transconductance variation, could not be adequately controlled without loop NFB.

Ref 1 Cordell R "Another View of TIM" Audio F/M 1980
Ref 2 Leach W. "An Amplifier Input Stage Design Criterion for the Suppression of Dynamic Distortions" JAES Apr 1981.
Ref 3 Holman T. "New Factors in Power Amp Design" JAES July 1981.
Ref 4 Otala M. "Transient Distortion in Transistorised........ " IEEE Transactions on Audio Sept 1970.

End Part 1

[lgreen]

Wow, interesting.............

Hey, back to the original question, I think bigger caps would be the first thing to try and improve, perhaps with snubbers. Reason for this is that I'm thinking that your ripple would be less and not very much affected by the XFO. Because you are drawing a lot of steady state power, I don't think a xformer change would be as effective, plus larger xfos are much more $$$ than caps.

My krell clone is dual mono with a 700VA transformer and a pair of 64,000 uF caps per channel, bass is simply incredible. Even at "low" bias of about 20WPC bass is very very impressive. I don't have the noticable change in bass performance between low and high bias that others have reported. Then again, I'm using way overspec'd powersupply, and 4 pairs of output devices per channel rather than the original 2. Kind of a krell mutant actually.