Krell KSA 100mkII Clone

[Jozua]

Gentlemen

I think Pierre Watts needs to be congratulated for his excellent PCB layout. I have great confidence in this fellow South African expertise who lives a stone's throw from me.

I am convinced that this project is going to be a greater success than the KSA50 clone.

Apart from being a true audiophile he is also a serious home theatre enthusiast.


Jozua

[jacco vermeulen]

Home theater is not my thing, yet.

But anyone who offers to do a PCB layout, especially without the intention of using it himself, i take my hat off for.
If the gent is a pro designer as well, i'll do the deepest Japanese bow i still manage.
I'll go along with the first two lines of Jozua's post, i have spare KSA50 boards but stage 3 looks even better.
Thanks, Pierre.

btw: what does the finishing stage of the kitchen table KSA50 look like ?

[AndrewT]

Hi,
I went back to post11 for the mk2 schematic. Have I found the right one?
Is c7 the cap across the Vbe multiplier?

The bias current through the drivers is about 1.1v controlled by the 75r emitter resistors. Have I got the right values? This comes out at about 14mA?

Driver dissipation is about 51V * 14mA = 750mW not 3.3W.

Can someone confirm the numbers or the calculation?

[Flodstroem]

Your calculating is right but you forget the milliamps for to drive 1.3Amps thru the CE in the output pair (a pair to each driver). If counting with worth Hfe value in output (ca 25) the driver (each of them) must also supply the output base with ca 52mA (0.052A). Total thru each driver is: 14+52mA=66mA
Total dissipation will then be of the order: 0.066x51V=3.37W(each driver)

hope this make sense.

Regards

[PWatts]

Thanks for the flattering comments :-)

I'll check the simulations tonight and post some disspation figures according to supply rails, bias and load. If memory serves, at 52V rails and 100W bias into 8ohms, the drivers dissipate somewhere in the 2-3W region, the predrivers 800mW and 30mW for the two sections and the others almost nothing. All the resistors can be 0.25W except the 1k feeding the LTP's and 12k1's that should ideally be 0.5W since they dissipate over 100mW each. Simulated THD at 1kHz under these conditions is about 0.01%.

[AndrewT]

Hi,
I see what you're getting at. I had overlooked the quiescent bias requirement. Thanks.
I had better go back and check my 50 Klone calculations.

What if one uses modern output devices with a minimum gain of 80? That would reduce the driver dissipation to about 1.5W.
Selected devices have hFE>=100

Now, do we design the layout for low gain (guaranteed to be 25 or better) or selected higher gain at Iq=430ma, or 50 or 80?

[AndrewT]

Hi,
ideas for extra dissipation capability for the drivers.

could a large aluminium angle be bolted across all 5 devices and then a computer CPU cooler or two be bolted on top and run at low revs (low voltage) and thus low noise?

Is there sufficient head height/room?

What if the drivers were soldered from the underside and bolted tight to the PCB but with the hot face away from the PCB.
CPU coolers could then be bolted direct to the drivers?
Would this release more room?

[Stuart Easson]

...will diminish as more current is delivered to low impedance loads, so even if the quiescent gain can be guaranteed to be high enough, we are unlikely to be able to predict the hFE under more extreme conditions...

More output transistors will help their collective gain remain at the upper end of the range, but IMHO it would be unwise to design heatsinking for drivers that is sized for anything but the lowest gain...and of course in the spirit of excess capacity the heatsinks should be at least twice the required size...

Just my 2c

Stuart

[Flodstroem]

but IMHO it would be unwise to design heatsinking for drivers that is sized for anything but the lowest gain...and of course in the spirit of excess capacity the heatsinks should be at least twice the required size...

Very wise Stuart, I agree to that 100%

Regards

[AndrewT]

Hi Stuart,
I hear you, but let's take your numbers on a bit farther.
Many modern devices maintain excellent gain upto 5A and some beyond 7A.

In the KSA100 Klone we will be dealing with two sets of 3pair of output devices (from the 2pairs of drivers) or even more for a few builders who like to go that bit further.

Assuming a gain of 80 upto 5A per device, that equates to 30Apk of output current into 1r5 and requires 45Vpk of output voltage (=675W into1r5). This must be pretty near the maximum that the Klone (on 6pair) can manage into that kind of load. The base current will be about 62.5mA and the emitter current of each driver half will be about 188mA. The drivers have long ago left ClassA bias. The drivers will now be dissipating about (52-45) * 0.188 = 1.3W into a resistive load. Their dissipation into a reactive load could easily exceed many times that, but only for very short periods. During these short periods a massive driver sink should absorb a considerable amount of heat for a small rise in sink temperature.
It transpires that the worst condition is high quiescent current and low gain. Just as Flod identified before he needed to correct my error.

ps. I have now committed to twin drivers into 6pair for my 50 Klone (just as the 100 does), but using hFE>=80 for all output currents and driver gain>=120 (2sa649/c669)for all output currents and loads down to 3ohm but with heatsink designed to match the required dissipation and projected temperatures (not doubled in size).