kevin gilmore

pull the opamp make sure that both 10k pots in parallel with the 500 ohm resistors are set at max resistance. just noticed that one of them is going to be full ccw and the other is going to be full cw. make sure the other 2 10k pots are set in the exact middle. or start by not populating any of the 4 pots.

Shorted 1815 or open led or led in backwards.

if the outputs are at +VCC then probably Q20 is shorted. or the led is toast. Need to know the voltages at the top of D11 and D12

gerbers were posted on page 5

those look real in every way. And kerry, that's a great job.

http://gilmore.chem.northwestern.edu/ksa5klone.jpg the fat version of the pot fits perfectly. I figure about 5 hours of build time total if I had all the parts ready to go without having to start and stop waiting for multiple orders.

Inu, nice use of Teflon tubing to get all the small transistors exactly the same height.

my guess is that the output transistors and the 2 ohm resistors are toast.

yep that is going to cause smoke.

ok, so the krell documentation is a little questionable, but when they measure the bias they do it from emitter to emitter. so the number is twice as big as the two test points I put on the board. specifically its 250 millivolts emitter to emitter cold, and 180 to 200mv hot. so at the two test points provided its 125 millivolts emitter to output no load cold a,d 90 to 100 millivolts hot. which sounds about right.

1) I think so, but I don't remember at this point. 2) just make sure the bolt does not touch the top and the bottom of the chassis at the same time. otherwise you get a shorted turn and that is real bad. if you use the Teflon wire and pre-tin, makes it much easier to do the amphenol connectors. at least 1 hour each.

That would work out to 2 watts per transistor, 8 watts, per channel. A bit too much, should be about 5 watts, about 1.25 watts per transistor. so about 125 mv.

its less than that, but I have that info at home. will look later. specific voltage across the emitter resistors.

The one board is clearly working, adjust the offset pot to get the outputs close to zero. On the other one, make sure the differential pot is set to the middle first and the offset pot set for the higher end of the resistance range.

so what we really want is avid part number 513202b02500g which is 2.0 inches high or 513302b02500g which is 2.5 inches high which is the to220 only version neither of which are stocked by digikey or mouser instock at Newark

I have the bigger heatsinks but one of the holes seems to limit The thermal efficiency. Will test... Would be nice if there was a taller heatsink with only one hole

this is the original krell heatsink mouser 532-513102B25

not so sure there are output caps. This may be very similar to the eddie current electra which has a center tapped inductor as the output device. also could be similar to the woo audio wes if they would pull their head out of their ass and use a center tapped inductor instead of a pair of inductors as the plate load. Clearly its push pull, same as the electra. How they control the output tube bias is another issue, as there are no didle knobs. If they actually load the secondary when in electrostatic mode, it may perform more linearly than the electra.

this should be as small as we can get it. please check for errors. http://gilmore.chem.northwestern.edu/dynahibal8d.jpg

sure as long as you use thermal isolation washers and in the one place isolate the two of them on the one bolt.

ha ha translation error. guinea pig.

i'm sure you could probably get it done in a day if you have all the parts and a pre-punched chassis. No high voltages to cause any grief.

That is about the right amount of time. You want the output filaments warm before the high voltage Kicks in. The timing cap is a lyric so the timing is not exact

http://gilmore.chem.northwestern.edu/ksa5ampfinal.zip but you should probably wait till we are finished testing.

so that's what a correctly assembled amb power supply looks like. I sure know what an incorrectly assembled one looks like.