Krell KSA 100mkII Clone

[Mark A. Gulbrandsen]

One last question... I had to increase the value of R-36 to nearly 7.5 K just to get the bias to go low enough. Its low point is now about 50mv and after a half hour warm up it stabilizes really well. I set it to 100mv and still very stable with the drover heatsink getting fairly warm but not so warm that you can't keep your hand on it. I didn't experience this problem with the other channel. It is at 0mv with the pot all the way down. Could the hfe of Q-13 be critical in this amp and causing this problem? I also wonder if Krell selected a resistor to bring the bias within range or would they have selected the device for proper gain?

Thanks!

Mark

[PWatts]

The hfe of Q13 is indeed a large factor regarding bias setting. That's why I mentioned in an earlier post that you should use MJE15030 to keep the original values, or adapt R36/VR2 if using the higher-gain versions.

However, the values you mention are a bit way-out. With simulations using the 15032's I still had to set the trimpot value to below 2k (R36 still 1k5) to achieve 580mA bias per transistor.

That should have brought the same problem in the KSA50 clone, and I'm using 15032/33's thoughout in the MK2 I've rebuilt and did a prelim test yesterday evening with the trimpot maxed out for minimal bias, and it was 0mA.

Note that the drivers' hFE also influences the bias, so perhaps you should try and check all of them (though start with Q13); maybe one's a dud or somehow got damaged.

OK, now that everything seems sorted out, the GB should be ready to progress. I've adressed all the problems and comments Mark had with the design I sent him. We still need concensus over a few things. It looks like we've settled for blue soldermask and gold traces. Concerning covering the traces with soldermask or not: I strongly feel that at least the bottom side shouldn't remain uncoated, it's often very close to metal parts and the close proximity of all the traces will have the risk of thin solder shorts. Besides - once the boards are fixed to the heatsinks or wherever, you'll never see the bottom side again

As a compromise I'd suggest we keep the top layer's traces uncoated. They are the ones in plain sight, and since they're not as densely routed as on the bottom layer the risk is smaller.

[Flodstroem]

Ok, I found the problem...

Now you can relax

Interesting to hear what you think of the anp in Stereo-mode, you must tell us whats your impression.

As a compromise I'd suggest we keep the top layer's traces uncoated. They are the ones in plain sight, and since they're not as densely routed as on the bottom layer the risk is smaller.

Yes PWatts I agree to that. It would be fine.

Regards

[jacco vermeulen]

switched positions

Who better to show that it's the little things when you're pressed for time than a guy in SLC who's built more amplifiers than killed turkeys.

I'll blend right in with the bird, without reading glasses R36 looks like R38 to me.
I haven't been able to read the newspaper for the last 12 months, suppose it's time to get used to wearing jars again.

Personally, i prefer to switch to the brilliant Sanyo 2SC3955 suggested by Pavel for the KSA50 or try a multi-stage version posted by Mike-K recently.
I find the Vbe-multiplier of the KSA100 kinda odd, and 15030/31s should be on display in a museum.

[loek]

I can buye them here for ca € 0,50 and can send them to Mark so he can add it to the collection for the Wiki.
Loek

[jacco vermeulen]

Loeki here,

Now that you're Online:
contrary to my statement i've not had a chance yet to deliver your Sym boxes.
I'm hoping to be able to drop them off at your daycare centre this thursday.

[loek]

thats the way it goes Jacco, i allready have your drinks overhere.
See you, Loek

[AndrewT]

Hi,
I can't understand why r36 has to be raised to 7k5 to get the bias to back off.

The voltage across the Vbe multiplier is about 4 Vbe + 2 Vre.
This is about 2.4V @ zero bias rising to about 3.8V @ maximum bias.
That needs a minimum of 3.4times across the Vbe multiplier.
Now working back from r37=7k5, then (pot+r36) = 7.5/(3.4-1)=3k1, with r36=1k5 then pot =1k6.

At the maximum, the ratio needs to be about 6.3times. This needs the (pot + r36) = 7.5/(6.3-1)=1k4. However the r36 value is 1k5 so we cannot guarantee to get 6.3 times out of the multiplier for all types of transistor (high Vbe).

Personally I would lower the value of r36 to 750r and also drop r37 to 5k1 or maybe lower still.
This is to ensure that sufficient current flows through the resistor string when using a low gain transistor for Q13.

[Jozua]

Mark

I would prefer that we have a soldermask on both sides. It prevents solder runs and having open traces adds to the risks of accidently shorting a trace with probe or getting unwanted solder stains/droppings on an exposed trace. The moment you have unwanted solder stains on the board, the whole object of having a professional look is defeated.

Good looks are nice and can give one some temporary pleasure when looking at them. However I would rather want this group to opt for board that complies to the technical specifications you expect to find in any high end amp.

The above does not mean I wish to negate your efforts in assisting this group in having the boards made up or building the prototypes so quickly. We do appreciate it.

Jozua

[VT67]

I have to agree with Jozua here

Regards