kevin gilmore

So any otl tube amp with a capacitor coupled output is going to have one very large turn on thump. except for Mikhail's disasters that left the output cap charged forever (black gates stay charged for months) That turn on thump can be 50 volts or more. Not a good idea. And the original cap coupled square wave probably had a 10 volt or more turn on thump to charge the 2 x blackgates on the output.

servo is taken from the output. can't possibly stabilize when the output is shorted to ground. in fact, probably makes it much worse.

1) of course not. and it is usually a bad idea to short out any amp to ground 2) idss 3) both of those should work.

charge them up for a week discharge them discharge them again measure

whether it has a loop output makes no difference unless you have installed extra switching relays etc. So if the pot is 50K and turned to zero, then the impedance is 50k from the headphone amp. If the speaker amp and sub amp are also 50k then the dac see's about 16.6k. But some amps may short the input, or open the input when not powered. build a set of loop thru cables.

http://www.jogis-roehrenbude.de/Verstaerker/PCL805-KH-Amp/Schaltung.jpg by the way there is a cap in series with the grid that is missing on the bottom output tube its a futterman design with a lf351 solid state front end. would be much better if you build it with 6as7

so take a look at the 2 scope graphs here http://www.head-fi.org/t/439657/headamp-blue-hawaii-special-edition/5685#post_10741753 First one has no vertical scale. My guess is that it is no more than 20 volts peak to peak. And look at the very rounded edges of the triangle wave. If you did the first one at 100vpp it would look absolutely grim with lots of ringing. And really horrible slew rates. I pulled some lundhall transformers, and a few other things this morning, drove them with a krell ksa50 and got nothing even close to acceptable. Same with a srd7 and srd5. Serious amounts of deliberate fudging of the specs here. But the worst one is the Ri of .01 ohm. Not even the biggest of the atmaspheres can come close to this number. Off by at least a factor of 20. Adding transformers just fucks things up. Same with the electra. Now if you were gaku-on and were able to wind the transformers correctly (like quad does for its esl63) then things would be different. And the price would be more than building a solid state amp. No free lunch.

5pf is 3db down at 278Khz 3pf is 3db down at 500khz

10k input resistor, 55k feedback resistor should work fine. 4 volt peak to peak balanced input results in 22 volts peak to peak balanced output. 110k feedback resistor results in 44 volts peak to peak balanced output

this thing has a real problem. self powered bias. so you get to listen to music for maybe 30 minutes before the bias caps charge up, and before the headphones charge up. major stupid. anyone with a koss esp6 would know what i'm talking about. and yes eric65 is a fucking shill, but mainly an incompetent moron that cannot think and believes anything on the web. like the fact that the rkv has .01 ohm output impedance. see BHSE thread over there.

this is the cap you should use for the servo, fits the smaller holes 667-ECQ-E2105KF

yes I think that is the right part. too busy making these http://gilmore.chem.northwestern.edu/dsc_0783.jpg birgir had found another part that definitely works too

a lot of people were having trouble getting the original parts. you can still use the other parts in that board, just have to bend the transistors a bit.

for the servo cap you really want a film cap. which is why it is that big. ceramics as I have found out have real trouble as integration caps.

anything over 20v is fine for the servo

on that version of the board, the A970 and mpsw56 are labeled backwards. The correct position is the a970's on the left and the mpsw56 on the right. latest board had the silkscreen right, consult that for more info.

I think you have the mosfets and bipolars backwards. all the Motorola parts are current production. all of the ixys parts are current production

Do not jumper the 270V zeners! either install them, or leave the slots empty.

that is how I build mine.

the added zeners for the mosfets and changes in the labeling also the dual transistor option

http://gilmore.chem.northwestern.edu/kgsshvproductiondualmot.pdf you replace Q1,Q2,Q9,Q10 with ixtp05n100m I don't think lil-knight has the latest dual transistor layout. the Motorola transistors are a different pinout on the older boards. most of us like the bipolar transistors better.

nope, you need 2 high voltage windings

yes it looks real. I have not seen 2sc4686a parts as fake yet. I bought extra parts from many ebay vendors recently and all test ok. 2sa1968 are a different story, most are fake.

lots of 2sc4686a on ebay still for cheap. suggest you buy a bunch.

if you have the latest boards with the zener protection you can replace the 3rd stage bottom and top of the 4th stage with 05n100m (latest on board version) or solder the zeners on the back schematic was never updated I have never actually seen 2sc4686. only 2sc4686a which is a 1200v part