kevin gilmore

Not sure since i've never seen an audio note up close, but the voltages are about half that of electrostatics. Its unlikely to actually be black teflon. I did test black delrin, and it starts conducting at about 200 volts. All of the connex octal sockets are white teflon. Probably for a reason. If its bakelite, then that is a different story. Not sure what the real stax jacks or the amphenol jacks are made of. I can probably test an amphenol jack, but i'm likely not to have the time to do that for a few months.

just have not had the time to test the final version. i can give you the board layout, have a few made and test. Biggest problem is that i'm moving my lab beginning of august, and i'm working 10 hour days in prep. Plus the stax mafia seem to be doing every possible electrostatic tube amp at the moment, all at the same time.

so its been over a month since andy has built an amplifier, and the stax mafia needs to keep the guy busy. So here is the bae. I changed the front end with a current source, and its dc coupled input and dc coupled output. 2 caps in the signal path. http://gilmore.chem....ern.edu/bea.pdf http://gilmore.chem....ern.edu/bea.jpg http://gilmore.chem....du/beapower.jpg The power supply has a + and - 15 supply, because later i will probably add the servo but at the moment with virtually all of the good opto isolator parts no longer in production, the ones i would use are way over priced. And the test version oscillates very nicely at about 2 volts on the output of 2 hz. So that needs more work. Next up what we are now calling the 2B10. only one cap in the signal path. Paralleled 5687's and DC coupled outputs with the above power supply set for +/-300 A total of 8 5687's. 6 seperate filament supplies... I'll also be doing a new version of the kgsshv power supply with the added 60 volt supply to make a high end, and universal power supply.

I have been thinking of adding a resistor string to the -HV line. But really i would like to add yet another power supply, say -30 volts, and do a current source instead. No reason to have more capacitors in the signal path if they are not necessary. Could use the input filament supply doubled and inverted. More on the srx circuit later. Also thinking of adding an opto coupled servo to the output stage. That gives me the excuse to add 2 more power supplies.

updated Megatron schematic. If you manage to figure out what the stax mafia is doing, please let us know. We are having a hard time keeping track of it all

100 ohm 2 to 5 watts

http://gilmore.chem.northwestern.edu/bae.pdf

you wanna build one of these! http://gilmore.chem.northwestern.edu/beveridge.jpg

It is certainly not compact. Unless you are comparing it to the DIY T2. It's about half size. Amp Board is 10.5 x 6 maybe it will end up A bit smaller. The power supply board is bigger. 2 power transformers are required if you Don't want to trash the output tubes.

Yep, just like it should be. 6sn7 is max gain of 20 (10 is reasonable for low distortion) 12ax7 is max gain of 100 (50 is reasonable for low distortion)

So the stax mafia has been hard at work on many versions of simple tube amps. Here is my favorite. http://gilmore.chem....ern.edu/bae.jpg birgir did not want to use the 6sl7 because of availibility issues, so this uses the most common in production tubes. 6SN7 just do not make sense, they do not have enough gain. Mikhail learned that first hand, someone else is about to learn the same lesson soon. With not enough open loop gain, all sorts of issues including much higher distortion, and channel balance issues... Of course you really should do 10m90s current sources, lowers the distortion a factor of 10. And i don't think this can be crammed into one box. DC coupled outputs. i'm sure some people will want to use 6L6GC fatboys instead.

A cat scanner. The whole inside section rotates about 2 rps. MRI are just giant superconducting magnet. Nothing inside but a coil of wire and lots of liquid helium. There is a reason these things are over $2M Upper left with the big high voltage lead is the xray tube. Lower right with the 5 fans is the detector module. upper right with the warning sign is the xray power supply, typically 60kv at about 500ma.

I have three sets of pins here, the originals I took from the conex sockets, The luvdunhill pins and the newest ones from Justin. All are slightly Different in diameter and the bigger ones require a slightly bigger drill Than my original print. Machining Teflon is a pain, I used water cooling to prevent creep. What I will eventually do is make some from macor and then there won't Be any dimensional change ever. And they will be $ 300 each

http://gilmore.chem....ern.edu/srx.jpg uses cheap current source.. Just 2 power supplies. 12 volt filaments for input and output stage. Need 2 seperated windings. Same holes and size board as the bate. power supply is the bate power supply with the C- supply not stuffed. About as simple as one can get. the current source really takes the distortion down a fair amount. To about .03% thd. edit: birgir wants to do dc filament supplies. That is next.

Sure there is a schematic. Its in my head somewhere. There will be an official schematic at some point. There is a bunch of different ways to build it including single ended only, and balanced/single ended The on board buffer which you don't have to use is an lme49710 driving an lme49600 closed loop with a tl071 as the servo, optionally the lt1150. The onboard single ended to balanced is an opa1632 with a tl071 servo. you don't have to use that either. multiamp works for both. Single ended there are x1 buffer opamps. Clear as mud right? There are a couple of people on amb's forum that think i have sold out to ic's... But they should hear the thing. Just about impossible to tell between the chips and the multiamp.

a relatively infinite supply for the future would be a good thing.

I'll second that one. Never figured it out either, i don't know how they can continue to sell those things, every time i try one even in a simple circuit i have trouble.

stax has always come up with unique circuits. The srx circuit is all about differential gain usually from a single ended input. You have to model it in spice to see what is going on, but there is voltage gain from both sections of the input amplifier. Synthesis later when i have some time. Front end is good for 45db of gain. If the output stage has 20 db of gain or more and the closed loop is 60db, that leaves 5db for feedback. Something mikhail clearly did not understand.

If some company would just make transformers as good as the ones in the quads, things would be much better.

But But... A Bentely is a volkswagon. So is the Lambo. and the Bugatti.

bankers lamp with one of the new GE 100 watt equivalent A19 LED bulbs. Plenty of light.

Mobile and 30 watts per channel ??

Due to the required balanced output for electrostatics, the balanced is better thing ends up coming along pretty much for free. Also the fact that you need a voltage gain of 500 minimum, and usually a more reasonable 800 to 1000 forces an open loop gain of at least 2000+. Something that is virtually impossible with any 2 stage tube amplifier, without massive amounts of distortion. This was the main reason why most of mikhail's esx things had volume/balance problems, because even with some feedback there was not enough open loop gain to level out the amplification. You certainly can do the input stage with a transformer, but if you want 10 to 20db of gain from it, the input impedance of the transformer is going to be under 1k ohms, otherwise there will not be enough output current to correctly drive the input impedance of the first tube stage. The hev90 was unbalanced input only, and then followed by a phase splitter anyway. So replacing the exact same 2 triodes with a differential amplifier results in more gain, less distortion, and better balance. For the nelson pass supersymmetry thing to work, you have to have at least balanced outputs.

Opa1632 with a servo is a better and lower distortion way of doing that Lower input impedance however

http://gilmore.chem.northwestern.edu/kgst.pdf not 100% sure if that was the final revision.