Krell KSA 50 PCB

amplifierguru said:
Amplifiers that REQUIRE more than reasonable PS capacitance to function at all well are also crappy designs. Ones that NEED their ripple reduced below input signal level or beyond 'VA need' reg are also poorly designed needing a cover up. To concoct absurd drive requirement is folly. Best to improve the design for cents than salve it with absurd and monstrous overbuilding. Then celebrate the monstrosity of it all.

Hey Greg,

I can use mine as a heater, it works rather well. 600W of class A heating - can't get better than that

😀 😀 😀

PS I do cheat in summer and reduce idle current to ~1.25A
 
awpagan said:
oops,
i hope i didn't say something controversal 🙂

now we have the general idea, my daddies better than yours?

The only reason i asked was because of the price of the caps.
and no not for commercial use, i can see the reasoning.

At present i have a couple of reg'd psu's dual mono, that i hope to
try.
when possible i'll see about large caps.

could be interesting to hear any difference or not?

You are right, large caps are expensive. But only if you buy them from places like Farnell and RS. There's plenty of places that sell surplus stuff, both in Oz and US. I bought my large cans for 5$ each !
No matter what front end is supplied with (regulated or not), it still pays to feed the output stage with large caps.

Bratislav
 
AndrewT said:
Hi,

In general classA amps DO NOT have constant current draw.
The current on the rails follows the sinusoidal waveform of the signal.
Of all the many and various topologies adopted for ClassA I have seen only one that is genuinely of constant current draw. That particular single ended topology is very rare indeed. and certainly not in the KSA Klone.

Allowing for that misconception, it might be easy to fall for the notion that speed is not required.
But if you want the rails to be apparently stiff with respect to the highest frequecies that the amp is asked to amplify, I would think that the regulator needs to be faster than the amplified signal.

Hi Andrew,

I still don't understand the mechanisms involved. I don't have inline current meters with my amp, but I've seen Pass amps and needle on the front definitely sits half way dead still until you REALLY crank them up. By then, my understanding is they are not in class A anymore.

Let's for a sake of experiment, say I'm using 2A bias with KSA 50 clone. If I run this amp with theoretical 8 ohm speaker, and never go over 30W rms (quite possible in most domestic situations), why would current supply vary ?

With no signal 2A flows from rail + to rail - , and power is dissipated as heat by output transistors (let's forget those pesky emitor resistors for a moment).
{We now see Greg pulling the remains of his hair 😉 }
Signal appears at input, output stage swings the voltage from virtual 0 and current starts to be diverted - instead through the output stage, part of it goes through the speaker. But total sum is still less than 2A (Kirckhoff law), and I see no need for more current to be drawn from the rails. This will be the situation until the signal becomes so big that idle current cannot supply the speaker anymore. Roughly at this point (if you biased your amp properly) you leave class A, half of your output TRs get starved of current completely, and output stage starts to draw more current from one rail than 2A. You are now in class B.
What am I missing ?

Bratislav
 
awpagan said:
Bratislav

There's plenty of places that sell surplus stuff, both in Oz and US.

Where/Who in Oz?

Would like to try both ideas, just havn't found (electronic component) surplus stores in Oz.

allan

There's a few stores in Melbourne that sell surplus stuff. No web pages unfortunately. One is in the city (top end of I think Lonsdale St.) - haven't been there for a while as I don't work in the city anymore. You have to keep visiting them, they won't always have what you need.
My caps are surplus telephony exchange power supply. 105 deg, low ESR, long life etc.

Bratislav
 
Bratislav

As for your reply to AndrewT on class A

Thats how I understood Class A.

But as for the higher frequencies?

I thought the power (current) was needed to control the Low frequencies?
To drive the bass speakers and control cone movement.

mids and high drivers don't need as much power as less mechanical
movement is needed.
mids and more so tweeters are more sensitive dB per watt.

played with biamp and triamp even with same dB drivers the bass
still sound better with more power.

allan

ps maybe i missed something?
wouldn't be the first time.

🙂
 
Bratislav

There's a few stores in Melbourne that sell surplus stuff. No web pages unfortunately. One is in the city (top end of I think Lonsdale St.)

Thanks Bratislav
Unfortunately i am in sydney
There has to be places up there, just need to find someone (in the know).
No web pages, is a pain.

allan
 
Bratislav said:
why would current supply vary ?
With no signal 2A flows from rail + to rail -
Signal appears at input, and current starts to be diverted - instead through the output stage, part of it goes through the speaker. But total sum is still less than 2A

What you are missing is that the current on 1 rail increases and the current on the other decreases with a signal on the input.
With a 2 amp bias the output stage is on the edge of leaving class A when 4 amps goes through the output node, that 4 amps comes from 1 rail, current on the other rail is then as good as zero.

Like Andrew says, you have a magnified sinus current from the input imposed on the steady state rail current.
Consequently, with 4 amps flowing in the collector of the output devices of one signature the voltage ripple on that rail will become twice as high.

A pushpull output stage with a 2 amps bias leaves class A when output current exceeds 4 amps.
In 8 Ohms that would be at 64 watts(4*4*8=64).
That is peak output, continuous 8 ohm output is half of that, 32 watts.
So, an amplifier that does 32 watts class A continuous in 8 ohms will also do 32 watts class A peak in 4 Ohms.
 
ryssen said:
If I where to use for example "+40V and -40V regulators for the front end of the amplifier" how large should the trafo be,volt/amp

For a regulator to function properly input voltage needs to be some 10 volts higher.
Say you'll use the 40V regulator circuit i posted, input voltage is 60 volts.
If the regulator is supplying voltage to the input stage and Vas only, it only needs to deliver a couple of milli-amps
Normal currents in the input stage is in the 1-2 mA range, a bit higher in the VAS, some current is lost through zener diodes.
Say a max total of 25 mA per side, times 60 volts is 1.5 watt.
If 1 transformer delivers current to positive and negative rail, the figure is 3 watts.
As with a regular powersupply it needs to be overdimensioned, and small block transformers are less efficient than toroids, 7.5 to 10VA would do.

BJT drivers for BJT output stages need plenty current.
If the output stage of a KSA would be able to deliver 20 amps, the drivers would need to be able to push near to 1/2 amp in the base of the output devices.
Times 40 volts makes a minimum transformer of 40VA, times 60 volts entrance makes a minimum of 60VA.
Reason why you hardly see separate powersupplies for bipolar driver and output stages, MOSFET output stages only need voltage and can be driven from small transformers.
 
Hi Bratislav,
does Jacco's answer in post5189 explain clearly?

The only classA topology that I have seen having constant current on the supply rails is as follows.

Single ended with constant current tail load. Capacitor coupled output taken to positive supply rail. This also works upside down using a constant current source and output taken to negative supply.

With two pole supplies (bipolar+-), the output must not be taken to ground or you end up with varying currents again.

Now you can contrive some other bridged topologies using single ended outputs that draw constant current when both sides of the bridge are algebraically added, but I have not seen any example schematics yet.
 
Hi all,
now I have a question for you.

Taking my example post5193, ccs single ended capacitor coupled, load returned to supply line.
If the load is resistive the constant current holds true.

Q. does the constant current rule still hold when the load is reactive?
 
true, but the principle is the same.
Bratislav was speaking about 2 amps going from + rail, 2 amps going in the - rail, max output current would be 4 amps.


On the input node of the amplifier there is not just a voltage sinus, also a small AC current.
The amplifier not only enhances the voltage signal to desired output voltage, it is also a current amplifier.

The max output current of the output may be 2 amps, it is always the product of current flowing out of the NPN device and current flowing in the PNP device.
 
jacco
yes

AndrewT
Hi all,
now I have a question for you.

Taking my example post5193, ccs single ended capacitor coupled, load returned to supply line.
If the load is resistive the constant current holds true.

Q. does the constant current rule still hold when the load is reactive?

But in this case we only have the 2 amps. (or 4 amps depends on the way he measured it?)
Single ended 2 amps (or 4 amps depends on the way he measured it?)
(bipolar+-) 2 amps (or 4 amps depends on the way he measured it?)

allan
 
jacco vermeulen said:


What you are missing is that the current on 1 rail increases and the current on the other decreases with a signal on the input.

I get it now.

<I have to see how is current meter connected in X amps>

The 64K$ question : is PSRR of a front end like KSA high enough for ripple residual not to get magnified enough to matter at output ?
<I know for a fact that unlike KSA 50, KMA 200 had more than a dozen large-ish (100nF - 330nF) polypropilenes throughout the board. Could this be to decouple power supply even more ?>

And even bigger question - does it really matter if some of the ripple gets through ? We have truckloads of extremely successful amps that don't use regulation (including of course KSA50/100). Would power supply regulation make things better ? Of course. Distortion will become even lower.
Can we hear it ?
And why do we still see designs without the regulation, if it really matters ?

Bratislav
 
Bratislav said:
And why do we still see designs without the regulation, if it really matters ?

Example:

your Costas Metaxas designed an amplifier long time ago that received much credit.
His friend Fred Gassman in Switserland developed a battery fed powersupply for the amplifier, selling it under his Avantgarde company name.

Sound difference between the 2: absolutely.
Price difference: staggering.

A battery fed power supply is like an infinitely large regular PS, no ripple and very high current capability.
Now that was 20 years ago, why did it take so long before others like Jeff Rowland started doing the same ?
Market mechanism, imo, no demand, no supply.

A regulated rail is inbetween common power supplies and a battery operated one. An improvement, for a smaller investment.
 
Bratislav
we seem to have come back to my questions on post 5140

i do have some questions though.

1-what is the ppsr like in the Ksa50?
2-How much ripple or noise on an uregulated line?
3-How much ripple or noise on a LM317/37?
4-Unregulated or reg'ed, easiest way of dealing with ripple/noise?


ripple... Post #5151
Here's my 3c worth...

I constructed a CLC filter section for an aleph drawing 2.5 amps, each C was 68k, the L was 2.7mH. The ripple on the back end was simulated to be <5mv, and measured to be 5-10mv. The current consumption of the Krell is largely constant, unless you push it into class B, when it does exactly what you would expect...

You really don't need great PSRR when the ripple is already in the single digit millivolts, but it definitely can't hurt, and if you decide to regulate it's a great place to start from...

Stuart



my point was
Is 5-10mv enought to cause noise on the front end of this amp?

next was
4 amps current
How about current fluctuations in the front end

remember we are talking about miili volts and milliamps?

allan