Krell KSA 100mkII Clone

[daylne52]

Hi mpmarino: I repaced your name in the buy list with mine. Thanks! daylne52

[PWatts]

57V @ 2.9A is a bit higher than the original one, which is 52V @ 2.5A = 260W.

However, since 63V is the smallest capacitor value >50V, 57V rails should be OK.

[rolandong]

to really get the most out of this amp,
it's necessary to regulate LTP/pre-driver.
i'm thinking of all discrete parts
(developed by Bob and jens in the scalable regulator thread, ~60-65VDC).
while waiting for Mark's board,
can we now discuss ps for this amp ?

[AndrewT]

to really get the most out of this amp, it's necessary to regulate LTP/pre-driver.
i'm thinking of all discrete parts

I agree.
What part/s get the brute force cap input supply and what gets the regulated?
How effective is the dual supply compared to stock?
What current works best? (not what current does the section draw?)
Is there any benefit in rough regulating the driver pairs? to separate from the gross ripple on the main rails?
How much capacitance is beneficial for the main supply?
How should the caps be arranged? single pair per channel or twin pairs or multiple pairs?
Worth adding snubbers? or just some film caps? how big?
Should the rectifiers be RC snubbered or just C or not at all?
or high speed rectifiers? or soft recovery?
Do RF filters at the primary side of the transformer kill the transient audio at the output?

[jacco vermeulen]

Do RF filters at the primary side of the transformer kill the transient audio at the output?

Nope.

[PWatts]

AndrewT raises a number of important questions. As there are a plethora of options that can be simulated, measured and evaluated it's unlikely that all will be adressed.

However, a good starting point would be to just separate the front-end from the drivers, and letting the drivers run from the main supply. Then to run the drivers and front-end together and the main supply on its own, and lastly all three sections separately.

The boards do not provide for separate power for just the LTP, only for everything upto the drivers. To regulate the LTP separately, the 1k resistor should be removed and power applied there, or remove the 39V zener as well and use a properly regulated supply.

It should be mentioned that active regulation will not necessarily guarantee better performance, especially at higher current levels. The drivers in particular should be happy with a separate but unregulated supply. If regulation is used, it must be designed for high enough bandwidth. Snubbers should only be necessary if the supply is located far from the amp, but they won't do harm either.

However, due to the low power levels involved it may be worth investigating on-IC switching regulators that requires minimal discrete parts.

[rolandong]

Hi PWatts,

Thanks for sharing with us excellent mkII board layout.

only for everything upto the drivers

IMO, the next logical thing is to build the regulator on top of existing boards we have.

cheers,
roland

[K-amps]

Sorry to be so late in jumping into this...

Originally posted by Mark A. Gulbrandsen


K-Amps got them all!

.....

Mark

Yeah!! But they were like 6 or 8 of them, not like 20-30

Originally posted by Stuart Easson
...is a strange thing in an output stage...

According to most everything I've read, and so far tried, the more transistors there are in the output stage, the less current is needed from the drivers, all else being equal...ie using 3 pairs of the plastic power transistors will stress the drivers less than using 2 pairs of the to3s...

Stuart

I know you have been a proponent of this concept Stuart and I tend to agree. However at some point in a class-A amplifier where the driver needs to load up (bias) the OP device, the class-A loading will require some DC current to be fed from the driver. More the OP devices you have, more the loading, so from this aspect, we have an inverse relationship of number of OP devices to drivers needed as far as driving the signal goes. If you plot both current requirements for different number of OP devices, I am guessing the curves would intersect at some point (break-even point) below which you are better off using more OP devices for one driver, and above which you need to add another driver.

I guess this variable is less of an issue in class-AB amps...

[Stuart Easson]

[QUOTE]Originally posted by K-amps

"...However at some point in a class-A amplifier where the driver needs to load up (bias) the OP device, the class-A loading will require some DC current to be fed from the driver. More the OP devices you have, more the loading, so from this aspect, we have an inverse relationship of number of OP devices to drivers needed as far as driving the signal goes. If you plot both current requirements for different number of OP devices, I am guessing the curves would intersect at some point (break-even point) below which you are better off using more OP devices for one driver, and above which you need to add another driver..."

If all else is equal, more output devices will each need less bias to achieve their fraction of the standing current, ie if 2 pairs are used with 2 amps total idle current they get 1A each while 3 pairs will have 660mA each, and will actually use LESS total current from the drivers, this is easily demonstrable. If you plot the curves as you suggest, you'll find you will never need more drivers, assuming everything else is equal.

Don't misunderstand me, I'm not saying there aren't scenarios where more drivers are necessary, just that arbitrarily adding output transistors while keeping everything else the same isn't one of them.

I've done a little research on this now, (as did Doug Self) and it is clear the appropriate 'size' of the driver stage is determined by a fistful of factors, including output stage Hfe and the load current. More transistors (within reason) have a higher HFe, so I am confident that more output transistors driving the same load will NEVER need more drivers...OTOH different load, all bets are off, higher rails another story...but since we are all making the 'same' amp, +/- 20%...and Dan wasn't a penny pinching designer...

While a lot of the time I am a big advocate of overbuilding for it's own sake, sometimes you just have to go with reasoning...

Stuart

[K-amps]

[QUOTE]Originally posted by Stuart Easson

Originally posted by K-amps

...

While a lot of the time I am a big advocate of overbuilding for it's own sake, sometimes you just have to go with reasoning...

Stuart

I have experienced BJT amps with one pair outputs (& drivers) sounding better in the HF and mids, but having flabby bass.

Adding more outputs gave me better bass/midbass... for the same power supply. So yes there might be situations where going optimal may be a good thing, point in case, I love the sonics of the 669/649. not so the older 15030/31 (haven't tried the 34/35 though).

As for the biasing requirements, the example works out well till you get to a minimal bias required to sustain linear operation per OP device i.e. you don't want to go below (pick a number?) 40mA perhaps. Granted these low numbers will happen only if you use an unGodly number of OP devices or a weaker driver.

Glad to have you back in the fray.