Krell KSA 100mkII Clone

[pinkmouse]

Could you not allow space for a discrete CCS as well as the monolithic one? That would solve availability issues.

[Mark A. Gulbrandsen]

I would'nt have any problem shipping the current sources with the boards..... A single current source is so easy to implement. But if a CCS built up out of seperate parts is easy enough I'd be open to it. Would make things all the more flexible....

The original fets were Supertex VN and VP0210N5. I think these can be replaced by Zetex ZVP3310 and ZVN3310 and these are cheap! See if these will fit your spice model.....? In the KSA-80 these devices weren't set up to draw very much current.

Also see THIS THREAD on klonning the KSA-80 in which the Zetex devices were subbed in for the Supertex.

[AndrewT]

Hi,
the 610/9610 alternative should be low capacitance.
At the current and voltages used, I think capacitance is the much more important issue.

[PWatts]

OK here are some of the results:
The best results for the LTP were with BC546/556 transistors. Good news since they are easy to get and cheap.

For the FET's, indeed the ZVN3310 and ZVP3310 are a viable option, and they improved measured specs considerably. The only dissipate 32mW each @ 1.5mA, so it should be OK without a heatsink.

Drivers and pre-drivers, MJE15034/5 seems slightly better than 15030_1.

Concerning the current sources, the 1N5309 or J509 yields much cleaner current than basic discrete CCS's. It will have to be a more complex one that would be able to improve on these diodes, that is quite frankly not worth it. In this application the diodes have a ripple of only 0.7uA, or -73dB.

So it seems like the component cost and variation will be fairly low and easily obtainable.

[Mark A. Gulbrandsen]

Thats good news and pretty much as I thought it would be. Have any of you gotten a board design far enough along that can be posted for us all to pick and pull at and then agree on some basic design? I think thats the next step here..... I was going to build mine on ceramic Tektronix terminal strips until this thread was dredged back up.

Mark

[VT67]

This project looks very tempting but I'm still strugling to get my KSA50 finished.

Regards

[PWatts]

The only issue I have with the ZVN/ZVP's are their rather high Rds of 10ohm and 20ohm respectively. Doesn't seem to be a problem and shouldn't be in this particular application but still. Plus in a TO92 package they may get too hot - the original KSA used TO220's with a small heatsink and if I recall correctly they got a little hot still. I think I'll provide support for both TO92 and TO220 so people can experiment with both types, either for sonic evaluation or if heatsinking becomes a problem. Alternatively the DPAK F-package may be good if a proper pad with lots of vias is used for heatsinking.

I was thinking of doing an exact copy of the original MK2 for the layout (as a matter of fact I'm restoring a genuine MK2 within the next few weeks so it'll be easy to jot down the basic layout), and just add the few refinements as Al has made to his KSA50 wrt grounding, PSU bypass caps, separate power supply inputs etc. Any suggestions you felt that Al left out on the KSA50 that should be present on the KSA100? If not, it will be a fairly straightforward design.

I also have to admit that I'm seriously toying with the idea of a completely separate regulated supply for the input and driver stage; it draws only about 120mA. Anybody ever tried it? Comments? It will not be part of the board but would be interesting.

The part that bugs me is the heatsink though. The predrivers can use small heatsinks as in the genuine one, but the four drivers and thermal feedback transistor need a substantial one. I'll try to place the heatsink at the far end so that people can use whatever they want, since a bolt-down type will be impractical for most people to try and accommodate.

In addition, has anybody looked at the MK2 photo? All the components are easy to identify, but close to the feedback electrolytics there are two caps - one silver mica and the other an axial type, presumably polystyrene or related. The latter is the 100nF bypass for the electro's, but what is the silver mica? It's not the 680pF input shunt (that is the blue one close to the input) or the 62pF Miller caps (they're next to the drivers), so what on earth could it be? I'll investigate closer when I strip the one to be restored but it will be a few weeks.

[pinkmouse]

Originally posted by PWatts
Any suggestions you felt that Al left out on the KSA50 that should be present on the KSA100?

Use both sides of the copper better to increase the thickness of the input and output tracks on the output boards, something I can't believe no-one picked up on. Try and fiddle the input pair closer together so they can be coupled so they see the same temp. increase. Allow a greater spacing for resistors so bigger boutique types can be used if required, and allow for different sizes for the output resistors as well. That's all I can think of at the moment, but I'm sure more will come to me!

[Mark A. Gulbrandsen]

I agree with what Al said. You may just want to follow his general design for the KSA-50 MK-2 boards as they seemed to work out very well for all that have built them. If you use the resistor size/spacing on the original KSA-100 board then we should be fine there. They used Dale/Vishay 1/2 watters everywhere and they are quite large. If I can match input pairs for these amps I don't think distance will be a factor. In a closed in chassis everything seems to equalize anyway and I've had zero problems with DC drift on my KSA-50. I will check with the guy that did the KSA-80 about how the Zetex fets have worked out for him. He used them in a very similar place on the 80 dissipating just under 200 mw with good results. It is very easy to heatsink the Zetex's if needed.

Mark

[jwb]

I haven't had any trouble with the Zetex parts I substituted for the obsolete Supertex. Yeah, the temperature is pretty high, but the whole amp is hot in the steady state. If you make sure there is plenty of copper on the PCB then you'll get satisfactory heat dissipation through the leads.

Watch out for the input grounds in your new design. I had trouble with buzzing using the original Krell ground layout and wound up drilling new holes in the back panel to move the inputs away from the bulk power supply.