kevin gilmore

the cathode winding of opposite polarity is what mcintosh calls cross coupled cathodes. A patent from the late 1950's. Definitely made a big difference in distortion. And they did bifilar winding too. but otherwise standard steel with alternating E and I pieces. There was a very recent youtube that showed them making transformers for the reissue mc275.

that is close to right, because of the way supersymmetry works. i will when i have time get you the exact values. using this amp with 300 ohm headphones is probably pretty goofy, as the voltage swing will easily fry both the headphones and your ears. put mainly your ears. really designed for low impedance low efficiency phones. he6, that kind of thing. new supersymmetry dynalo is much better for high efficiency high impedance headphones

same as any of the .1uf ceramic caps, 15v or more

you are going to need a bunch of heatsink. use the off board power supply version.

NO. you are missing the point. The point is a fair bit MORE current. one 10m90s is not going to handle that. you would need 3. and you need 3 x fqpf8n80c. other than that, both supplies are about the same in cost and performance ksc5026 works fine as 2sc3675 or 2sc3840 replacement and is a current part ksa1156 works fine as 2sa1486 replacement with the 270V zener and is a current part

depends on if you are pabbi1 or not.

coming soon to a file server near you http://gilmore.chem.northwestern.edu/kgdt.jpg 4.15 x 4.17 inch think of it as a cheap swiss army knife of electrostatic amplifiers works with any standard dual triode suitable for voltage 1 per channel Russian 6n1P, 6cg7,6fq7,ecc99,12bh7.... NOT designed by morons with just a digital voltmeter NOT open loop Not high distortion rationally priced (you get where I'm going here)

free gerber viewer for board files, open zip folder http://bayareacircuits.com/free-gerber-viewer/

I actually did this because it was taking too much time to respond to every single person separately. Now I will just point to the file tree, and keep it updated.

hopefully i cleaned this up and its the current versions of everything see updated links on page 5

I have seen plastic covers for to220 transistors before. Where I have no clue

power supply board http://gilmore.chem.northwestern.edu/kgstpsu21.rar

this should handle the power angle bracket to big heatsink and 3 x fet per side http://gilmore.chem.northwestern.edu/kgsshvpsminiexternal.jpg

final kgst board http://gilmore.chem.northwestern.edu/kgst-miniv03.zip

birgir messed up the lsk389 footprint on the test boards. how, he does not know as this should have been a cut and paste. version .4 of the boards will be production. also the one cap was too close to the ic socket.

final pic before I put the fets in. works just fine. http://gilmore.chem.northwestern.edu/dsc_0591.jpg

300v pulls too much current for the kind of thing used in a srd. Besides which very low noise is rather important.

iwik is talking about the power supply, and technically the resistor was running very close to its voltage limits. So its now 2 resistors in series. Same value. Its on both of the 500v supplies

i really like Teflon http://gilmore.chem.northwestern.edu/staxmale2.jpg http://gilmore.chem.northwestern.edu/staxmale3.jpg http://gilmore.chem.northwestern.edu/staxmale4.jpg will knurl the end of the plug just for fun. setscrews are 0-80 the production versions will be white delrin i think. staxmafia personnel will get the Teflon versions.

a pair of 10m90s in parallel will definitely do it.

that stuff grips the pins very tight. no glue required. what the Teflon does, I do not know. also have to try delrin and nylon. I made the first one just to make sure I got the machine code right and in the right order. turns out I made the first one from kel-f. Not what I was intending on

you really want 100k. with 100k resistor, the servo gets it under 1mv in 30 seconds. otherwise its a few minutes. r7 and r8 can be up to .2 watt. if you short the output, about .5 watt and its a dual opamp in a single package you can get rid of both electrolytics, or the one electrolytic in the latest board by maching the fets and moving the left side of R21 to the junction of R1 and R2 and then shorting the bottom of R17 to ground. Then its fully dc coupled, but you have to go thru a bunch of fets to get the matching right.

following chinsettawong's idea ... http://gilmore.chem.northwestern.edu/staxmale.jpg will make more once I get some real Teflon. Plexiglass does not like to do this.

this is the final and correct schematic http://gilmore.chem.northwestern.edu/squarewaveproduction.pdf if you are careful and match the fets, you can get rid of the 1500uf caps. the original board had the servo inverted, hence the extra opamp to invert the output. latest stereo board http://gilmore.chem.northwestern.edu/sqwave9.jpg pair of boards installed in icariums SS1 http://gilmore.chem.northwestern.edu/sqwave10.jpg the corrected mono board is also available. R16 really should be 100k because the servo is too slow with 1M

so the real reason that 68k resistor was there is because otherwise the ccs transistor which is not heatsinked would get way too hot. And if the ccs transistor ever shorted the 68k resistor prevents the fets from blowing up. This is the problem when you try to make an amp with only the high voltage rails. same issue the one guy over did many years ago when he built a kgss with only high voltage power supplies.