The DIY Electrostatic amp thread

[livewire]

FWIW ebay's got 50ca10 chinese tubes and sockets.

[spritzer]

Yup, 50C10's are in current production in China (or at least recent production). Those amps do look pretty cool but 1150V is just insane. This is 211/845 territory...

[luvdunhill]

Just build the modified Singlepower circuit. I think it will sound better.

[Antistar]

My KGSS doesn't play nice yet again. Bias voltage has dropped to 235V but the rail voltages all seem to be ok, What does it mean?

[morphsci]

It means

[livewire]

My KGSS doesn't play nice yet again. Bias voltage has dropped to 235V but the rail voltages all seem to be ok, What does it mean?

It means that one needs to troubleshoot the bias part of the power supply.

Depending on who built it, there should be two diodes, two capacitors and two resistors in that sub-circuit.(or not)

Very easy to check if you know how to work around high voltage. If you dont, dont bother going inside it.

The high voltage can kill you. That's why they put them labels on things saying "no user servicable parts inside".

Also, where did you get that reading of 235 volts? The true bias voltage cannot be read from the stax jack

with a regular multimeter. You will always get a lower reading depending on the meter.

[DouglasQuaid]

My KGSS doesn't play nice yet again. Bias voltage has dropped to 235V but the rail voltages all seem to be ok, What does it mean?

It means you touch yourself at night.

A. Is that the highest voltage you can get while adjusting the potentiometer?

B. Are you sure you're taking the reading before the resistor and not afterwards (still won't make your bias reading go from 235 to 580 though)?

C. If you didn't understand either of these questions, stop what you're doing and pay someone else to fix it before you hurt yourself.

As a side to my previous question about cooldown time on the KGBH supply, it took just over five minutes for my supply to get from ~350V to under 1V with a DMM acting as a bleeder and the IEC ground still connected.

[Antistar]

I apologize if i hurt anyones feelings with my lackluster wording... obviosly i meant what might cause the bias to drop.

It means that one needs to troubleshoot the bias part of the power supply.

Depending on who built it, there should be two diodes, two capacitors and two resistors in that sub-circuit.(or not)

Very easy to check if you know how to work around high voltage. If you dont, dont bother going inside it.

The high voltage can kill you. That's why they put them labels on things saying "no user servicable parts inside".

Also, where did you get that reading of 235 volts? The true bias voltage cannot be read from the stax jack

with a regular multimeter. You will always get a lower reading depending on the meter.

I took some measurements for what probably is the bias supply part.

The resitor i couldn't get a reading from.

The Wima cap reads 0V and i suppose that is the problem. it's MKS 4 model rated 0.15uF, 630VDC. It replaced the second TB100n that was damaged when i got the amp.

I measure the bias reading at the bias terminal point, It's located after the 4.7M resistor but it did read 580V before IIRC. If it's wrong please tell me where i should measure.

@douglasquaid:

A, Yeah it's the higest adjustment on the trimmer before it stars clicking.

B, Answered above.

Thanks for the help and pateince guys.

[nikongod]

I measure the bias reading at the bias terminal point, It's located after the 4.7M resistor but it did read 580V before IIRC. If it's wrong please tell me where i should measure.

The input impedance of your volt-meter is loading the output of the bias supply.

Check the datasheet for your meter, betcha the input impedance is somewhere in the 4-5Mohms range.

Measure the voltage BEFORE the 4.7Mohm resistor, or account for the drop your meter will cause due to the various impedances interacting. Do both if your feeling paranoid.

[spritzer]

You really need a 1000V cap there since the voltage doubler easily reaches 700V...

[Pars]

^ Well, they sure as hell will after that masterful post...

[DouglasQuaid]

blah blah blah

_________________

[urlremoved] watch online movies[]

First post and already pimping your own website? Spammer no spammy.

That being said, my kgss psu is up and running. Added a zener across the fet, as KG mentioned, along with a few strategically placed resistors.

Amp board, not so lucky. One channel works beautifully, the other puts out nearly 200V DC upon startup and adjusting the bias pots does nothing useful. DC drops down to 0 over time, slowly tracks back into the negatives, but I haven't left it on long enough to stabilize long term as the third stage 2sc3675s get rather hot. All the LEDs light up, but the one near the -15V supply is definitely dimmer than the other two. No transistors fail the diode test and all resistances check out. Fun stuff.

[guzziguy]

Spambot posting?

[DouglasQuaid]

Spambot posting?

No self-respecting human would name himself/herself Celin Dion.

[FrankCooter]

Here's a simple little tube electrostatic amp that started out as an Egmont. The output stage, formerly a pair of resistor loaded 6SN7's run at 300V, is now a pair of triode connected 7591's with IXYS10M90 CCS plate loads run on 400V rails. The original 12AX7 based "long tailed pair" first stage was unstable with the 7591 CCS loads. I replaced it with a simple 6SL7 based Williamson type circuit and all was fine. Total costs are a bit over $200.00. Sounds surprisingly good!

[chinsettawong]

Hi Frank,

That looks like a nice little amp. Thanks for sharing.

Wachara C.

[FrankCooter]

Screwed up the sizing of the schematic so I pulled it. I'll put it back up tomorrow. Only thing worse than my photography is my internet ability. Please excuse my incompetence.

[FrankCooter]

Here's the resized schematic. Hope the second time's the charm. If anyone's interested, I'll do a detailed list of component values. I like the 6SL7/7591 combination, but a 12AT7/7189 combo would probably work just as well, and if you used the Russian 6P14EV, would be dirt cheap to implement.

Power supply can be anything you want. I started with solid state regulators, but after blowing the pass transistors a couple of times, I replaced the regulators with simple CLC filters. output caps are motor-run PIO, and eventually I'd like to experiment with Schotky rectifiers to see if it makes audible difference. Running straight AC on the filaments, but still sounds dead quiet.

[mrarroyo]

Darn just what I have been looking for, an electrostatic amp. Of course I have no ability to build it.

[chinsettawong]

Hi Frank,

Yes, I'm very interested in the detailed list of the component values, if you don't mind.

Wachara C.

[FrankCooter]

Here's a list of component values for the 7591 electrostatic amp. None of these values are optimized in any way and mostly reflect what I had at hand. As much as possible,

i tried to use I tried to use "bang for the buck" bottom-feeder components.

V1/V2: 6SL7 dual triode. '50's Sylvania are still around at decent prices. Cheaper would be a 12AT7.

V3,V4: 7591 power pentode run in triode mode. 7591's are expensive and rare in NOS. I was lucky to have pulled some out of a garbage bin years ago. J.J. and E.H. make new ones that run about $75.00 a quad. A cheap substitute would be the Russian 6P14P-EV, which is an uprated EL84. Without individual cathode pots, each channel should be a balanced pair. R7 sets the offset, which should be as close as possible to 0V.

CCS loads for 7591: IXYS10m90 current sources. Be sure to use good sized heat sinks. Set R5 current set resistors for 20mA, which is about 150 ohms.

R1: 50k/100k audio potentiometer.

R2:220k, R3,R4: 120k, R5,R6:150 ohm, R7:200ohm 2 w linear pot, R8:470 ohm 5 or 10w. R9,R10: 500k, R11:16k 2w. Stopper resistors on CCS gates are 1k. Stopper resistors on 7591 grids are 330 ohms. Triode connection resistors on 7591 screens are 100 ohms.

C1: 450V electrolytic, value not critical

C2,C3: .22@600V. I used the excellent but cheap Russian K-40 PIO .

Power supply was based on an Antek dual 325V toroid with cheap "Radiodaze" chokes. Filter caps were 220@450V Panasonic electrolytics off Ebay.

Chassis is whatever you want. Mine was leftover materials from another project. Surplus Sales of Nebraska has something called a "Nabu" (sp?) that sells for about $20.00.

Total costs, not counting chassis work, run about $225. if you use a 12AT7/6P14P tube component. Add another $100. for a 6SL7/7591 tube complement.

My long suffering wife calls this the "Everyman's Amp" because it's the cheapest thing I've built in years. Definitely a fun and worthwhile project. Because of the voltage levels, it's not a beginner project. High voltage safety procedures should be understood and utilized when constructing any project of this type.

[dsavitsk]

Here's the resized schematic. Hope the second time's the charm.

Should the grid of V4 be connected to that side of R3?

[FrankCooter]

Should the grid of V4 be connected to that side of R3?

The drawing is incorrect. The grid of V4 should be connected to the plate side of R3. Thanks for spotting the mistake!

[dsavitsk]

I think R7 is out of place, too.

Have you compared the driver + cathodyne to just using the first tube as a LTP splitter? I don't have enough experience with either topology to form much of an opinion so I am curious if you tried both. One thought, though, is that LTP front end might enable DC coupling and eliminating the negative rail (to keep the cathode load high (though, does it really matter?), use a choke, or use a single CCS bypassed by a high value resistor to give some slop)?

Edited April 20, 2011 by dsavitsk

[dsavitsk]

As someone who has never built a stat amp, here's my contribution

Basically, LTP input and output stages. Choke loads on the output, and while chokes are not free, I think the savings of not needing a bipolar supply would make up for it.

Edited April 20, 2011 by dsavitsk