kevin gilmore

in the original kgbh, the 2sj79 can be replaced with a ksa1220A and the 2sk216 can be replaced with a ksc2690a. Not the same pinout. someone would have to test, but synthesis shows it would definitely work. in fact the ksa1220a and ksc2690a are faster and lower capacitance parts than the originals.

t2shrunk2 updated. definitely missing the one land. not sure how that happened.

ignore the tube pin numbers, they do not show up anyway. its how that socket was made when I did the t2 picture above updated

updated file current thru the tube designed to be 20ma, so a bit more than previous driver transistor is ksc2690a only, dissipates a little less than 1W, needs very small heatsink you loose about 45 volts across the tube, so +/-450 is ideal

board file posted I can do the same thing with the nanotube, for the ultimate in gigantic. fets, triodes... its all the same thing...

I can put up jiml's gerber file on my server, or convert over to my software with a decent picture.

have not built one yet. may change the drive transistors to ksc2690A been thinking about this a while, saw no reason not to do the board. 5.6 x 4.7 inches

kgsshv-carbon with the sic-fets turned into 6ca7 was going to be next year

My boards are the original circuit exactly. And sounds like the original, exactly. the crap from China is something else entirely some have half the output transistors, some have half the capacitors, none have the original power supply etc. some don't run at the right voltages etc. cheap crap, krell in name only

since its Christmas, I will be nice. 1) your son is a fucking clueless idiot that has obviously not fallen far from the tree. 2) you obviously did not read the thread 3) The original krell board for which I have exact copies of pictures of the top and bottom of the board supplied from krell directly thru tom... The original krell board was hand layed out with film and tape on .1 inch centers. The left and right channels were not physically identical. My version made the left side identical to the original with each component exactly in the same place on the board. And each circuit land in exactly the same place except electronically aligned. Then the right channel was an exact duplicate of the left channel. So mine is actually a bit better. I just posted a new version of the board with teardrops enabled. 4 or more layer boards can also be done without destroying the board. You just need a large plate medical Xray to do so. And I have one.

a F5 is a much bigger amplifier, 20 watts into 8 ohms. also, not a transconductance amp, and has feedback If you want a transconductance amp, F1 or F1J both of which really like very low impedances and have significantly higher thd.

2 Watts RMS into 50 ohms and you expect it to drive desktop speakers... NO klipsch corner horns, probably ok. not really desktop speakers much more powerful amp coming at some point

no parts changes up to 35V

for planar ortho's with an impedance of 50 ohms,this amp really shines. with 30V supplies, does 2 watts pure class A into 50 ohms with a thd of about .03% which is much better than the bakoon does. for hd800, all it does is accentuate the frequency response issues.

Congo5 has a working board now.

Probably a hacked up version of a Srx with absolutely the wrong output tubes. or something really fucked up

the board is 9.35 x 6.85 inches the heatsinks prevent the board from changing that 6.85 inch number more than maybe .1 or .2 inch the 9.35 I may be able to reduce .5 inch but things are going to be tight. neither of those changes is going to influence the board price much. 373 holes

those are the low voltage regulators. I still think that heatsinks were a good idea for those. the 10m90s are in the middle and do have heatsinks

change the high gain resistor to 75 ohms and the low gain resistor to 450 ohms to atch the gain of the bakoon this is a bit higher than i thought

120 watts should be plenty in other news soren found a problem with the turnoff condition. easiest fix is a pair of back to back 12V zeners on the outputs gate to source in addition because soren wanted the feedback at the driver board, there is this modification which actually reduces distortion even further and allows direct local control of current in the output stage without a "classical" servo. the carbon really is very stable over time without the servo, mine has drifted less than a volt with the servo removed. so there will be another version of the output board. http://gilmore.chem.northwestern.edu/circlotronoutputnew.pdf

like i have time for that at this point.

don't see any difference with an extra diode in the synthesis.

so tyll has been going crazy over the bakoon for planars. well there are a lot of things i don't like about that amp, especially the battery power and distortion. so here is something much better, tested last night at .06%thd, no feedback, output impedance 100k which is solely due to the servo... http://gilmore.chem.northwestern.edu/currentsourceamp2.pdf http://gilmore.chem.northwestern.edu/currentsourceamp2.zip also testing curved traces

kgsshvcarbonv6.zip kgsshvpssicfetsinglenewleftfats.zip kgsshvpssicfetsinglenewrightfats.zip all posted, all are 98mm bottom of board to top of ceramics resistor changed to 120 ohms.

http://gilmore.chem.northwestern.edu/kgsshvcarbonv6.pdf top blue line is where the ceramics end up and is 98mm from the bottom of the board. If you mount the sicfets at the lower hole. 2nd blue line is optional cut point the sicfets would be bent .7 inches from the center of the hole which should be doable. seems to fit, but just barely.