Krell KSA 50 PCB

size...

Troy,

From reading the specs, the heatsink extrusion you are looking is rated at 1c/w per inch. Assuming you are building everything per factory specs, you will have 36Vx2railsx1.9Ax2channels or 275watts of class A power to dissipate...if you plan on extended use in class B add more heatsinks...

If you have a max target temp in mind of say 50c, subtract ambient, normally assumed to be 20c, giving a rise of 30c. The heatsink you chose will dissipate 30w per inch, at a temp rise of 30c, so if you divide 275watts by 30w/inch you get the length you need for 2 channels...or 9.2 inches.

There are all sorts of other issues with getting the best out of your heatsink, the distribution of the devices over the surface, the mounting of the devices themselves, that sort of thing, but the basic principal is to use more if in doubt, so rounding up to an even foot or more couldn't hurt. I just fixed a huge denon receiver that improved the heat distribution of the internal heatsink by putting a 1/8" sheet of copper over the entire flat surface of the heatsink and bolting the transistors to that.

Hope that helps

Stuart
 
Got my caps today!! *YAY* Got the boards, waiting on other parts, they are still coming, got most of the semi's ready to go (well, in a pile!) the heatsinks are on their way, etc, etc etc 🙂
So - what drivers need heatsinks? I'm assuming only the last set of drivers need it? what c/w? i am lost to find out what the drivers will dissapait, something about 10w?? no idea! 🙂
Anyone got to comment on the cap's choice and pf of the caps on the board? Looks to me like it should be 39pf, but i'm really not sure...
Aaron
 
Oh, sorry, previously a post noted about removing the SOA protection, if one wishes to do this, simply do not install any parts with *3** in the number, like R301, D301, Q301, etc, this will remove any protection circuit within the amp... Of course, depending on your outputs you may/maynot wish to do this, but i'm leaving it up to the individuals involved to install/remove as they request!
Aaron
 
"Anyone got to comment on the cap's choice and pf of the caps on the board? Looks to me like it should be 39pf, but i'm really not sure..."
____________________________________________________


I posted a message about this a page or two ago and everyone ignored it.

Yea, I agree that they seem too high of value. If they are it won't even pass a clean 1khz square wave without rounding things off drastically and thats an easy test for some of us. You may end up with more of a triangle wave than a square wave. If so then lower the capacitance to what is on the other schematics and see how things look. We will probably have to wait till someone actually fires their boards up to see what happens. Keep in mind that it is also possible to just fire up the board itself and see how it performs into a higher impedance load of say 600 ohms or so. That driver board is also a small amplifier by itself albiet a high output impedance one of very low power. Judging fomr other amps I've built that 39pf value seems more realistic.

Mark
 
Re: size...

Stuart Easson said:
Troy,
The heatsink you chose will dissipate 30w per inch, at a temp rise of 30c, so if you divide 275watts by 30w/inch you get the length you need for 2 channels...or 9.2 inches.
Hope that helps
Stuart

Thanks Stuart-

Actually I am planning on using an 8" wide x 12" long piece for a SINGLE channel. So I guess that would be adequate.

I will be building 3 mono-block amps with dimensions roughly the size of a sheet of paper but 4-6" thick.

Thanks again Stuart for your time,

Troy
 
Sorry Mark, didn't mean to repost what you said, but i thought i'd put it up a bit differently just in case it invokes a response from someone in the know!! (not me!)
But if you look at the schematic i put up, the capacitor is in a different location, not connected to the collector (i think, going from memory as it's late and i don't have time to check) on the KSA-100, but in the clone it's on the emitter? (i think that's right, i'm tired, someone correct me!)??? Does that seem right? that may be the reason for the variance with the value?
No idea to be honest, i'm lost now...
 
Sorry NUTTTR but the caps are on the same place. But I think that is one mistake on the early ksa-50 schematics from McLean. Resistors from the rails to the diferential transistors are not equal and they should be (or I am wrong?). So maybe the 39pF is mistake too?! What You think? Or anybody?!
 
mistake?

Hi,

The 390pf caps are almost certainly meant to be 39, they crop up at the lower value in too many independently obtained schematics for it to be a coincidence.

The resisters in the differentials don't have to match, it's a design choice. Some people consider a completely matched differential setup to be more elegant, luckily we have schematics that show both arrangements have been used over the lifetime of the amp, so everybody has opportunity to be happy. Probably important to be consistent in the selection of component values, ie choose one schematic and stick with it, don't try and mix and match. I think most people are building the fully symmetric version, so thats where the most 'help' will ultimately manifest.

The caps being too big or small will fortunately not be too big a financial burden, they are cheap, and as soon as someone with an oscilloscope finishes, we'll have an answer. I'm working on it, along with a handful of other projects...

Stuart
 
more info...

As an addendum, anyone wanting to build a truly welding worthy amp should either leave out the base resistors on the output stage (my choice, see below), or make them 3+W devices. In the event the amp is called on to deliver prodigious current at elevated temperature these resistors could be required to deliver an amp or more, so choose the wattage appropriately...

My assumptions here are 2 pairs of to3 devices, the more devices, the less this is an issue...If the output is 50A, and the hFE of the output stage has decreased to 25 (and it could go lower), the output transistors will need base current of 1A each...they are rated at 5A, so they are not going to have a problem, but those resistors, well lets just say, choose wisely.

At these sort of currents, the base resistors are going to have noticeable voltage drop, this will of course decrease the available voltage driving the output stage and the peak power output will drop...

As a less desirable side effect a 2.2v drop across the base resistor (2.2ohm, 1A) here with 50amps of total output will make the output stage dissipate an extra 110w at clipping...hence I am going to do without them...

Stuart
 
Hello Stuart, is there a chart or spread sheet to show the resistors and thier values that effect the output power and mode?

I am using a 30-0-30 VAC transformer and I would like the first 30% output class A and the slip to A/B above that.

I have found the bouncing around of what seems like misinformation on this thread very confusing. Maybe I just have too many amps under construction! 🙂

Regards

Anthony
 
not sure I understand the question...

The output stage base resistors don't really affect any of the basic modes of operation of the amp.

If they are used here they will affect the peak voltage the output transistors see and therefore can deliver. Apart from anything else this is effectively lowering the ultimate output stage efficiency. Since the resistors eat some of the drive voltage, the outputs which are emitter followers can't achieve saturation, the voltage drop across the base resistor is always present across the output transistor as an unusable part of the rail voltage.

In essence the following is true for class A amps in general, though some of those don't have an AB mode to transition into. In the case of the Krell KSA50, the amp has rails that are a little higher than needed, and gives more AB power than spec'ed...

The peak class A output is given by (the square of twice the idle current) times the impedance of the load. RMS is half that. Hence the factory spec of 1.9A gives (1.9Ax2)^2x8ohms, or 115W, so the RMS is 57. Thus the factory settings only give 28W class A into 4 ohms, or 14W into 2 ohms...

The AB peak power is given by (rail voltage squared) divided by the load, or ~(34.5x34.5)/8, or 148W, so the RMS is 74. Assuming you have a truly studly power supply, and we already know many of you do, the RMS AB power into 4 is ~140W, into 2 it's ~270, into 1 it's ~500...you get the idea.

To get the 100w of class A power into 4 ohms that the halving of the load implies would need a minimum idle current of about 3.5 amps, krell's original margin would suggest 3.8A. This change doesn't produce more peak power into 8 ohms, there is a small increase in the class A output into 8 ohms, but ultimately the rail voltage doesn't allow the 8ohm load to be driven any harder. 200w class A into 2ohms, no problem, just have it idle at 7.6A, but don't forget the 550w per channel you need to dissipate...I'd add a couple more outputs too...

The voltage buys you the output power, the idle current sets how much of it is class A, and of course they both buy you a lot more heat...

Hope that helps

Stuart
 
UK is sooner...

My trip to the UK is at the end of next week, I'm there for a week, back for 2 then off to canada, montreal first then toronto. The friends in Toronto are about to deliver a baby, so I may be required to entertain and occupy myself for some of my time there...

To add a general answer to you question about an amp that does 1/3 rated power as class A then the rest as AB, it will depend on your load impedance. The KSA50 default is about 70% power is class A into 8 ohms, but it's only 25% of the maximum power being class A into 4...

If you want to give me more guidelines as to your load, rails etc I'll calculate and explain the numbers...But here's some defaults...with 30v AC, and normal rectifier with capacitive smoothing you'll have ~42v rails, so your amp will push out ~40v peak, which corresponds to 200w/8ohms, so 100w RMS. If you wanted the first 30watts as class A, you'd need ~1.3A idle current, with a total per channel dissipation of 88x1.3, approx 120watts.

Stuart
 
Thanks for the informative posts (as always!) Stuart...
I was considering leaving out the base resistors as the originals never had them... Is there any reason they are there? I assume it would give a "margin" between the rails and the output voltages, but i wouldn't have thought that would be necessary!
I'm going to order 390pf caps and 39pf caps, just in case, for the extra $$ it's not expensive and i'd like to see how they go
Aaron