current feedback electrostatic amp

[kevin gilmore]

so sorenb really wanted this, so here it is

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

the outputs are 300v darlington transistors  just need to find a suitable npn match for the stn9360, and this can be done all surface mount

tiny, fully complementary, fully dc coupled, decent input impedance (none of this 75 ohm nonsense) push pull output...

edit:  PBHV2160ZX looks great for the npn.

so you can add one more set of outputs and run up to 500v power supplies

balanced input only, and you need 4 of them

 

Edited November 24, 2016 by kevin gilmore

[nopants]

am i the only one imagining kevin drawing this up with one hand and basting a turkey with the other?

this needs the plug in card form factor to get the full msb experience

[kevin gilmore]

actually 3 hands.

(updated schematic with great respect to you know who)

1 hand for turkey baster  (actually making casserole at the moment)

short hand for mouse

1 hand holding 50 year old single malt

we will definitely be building this one

 

 

Edited November 24, 2016 by kevin gilmore

[JoaMat]

don't hold it for to long, it might get old

[Sechtdamon]

Do you accept preorders Dr. Gilmore?

[JimL]

See, this is the real idea behind the Stax mafia.  They pat you on the back for building one of their designs while they're sucking money out of your pocket by posting a new one.

[nopants]

No LV supplies and no sinking, seems like this was what the exstata should have been.

[joehpj]

can't catch up with the mafia...

looks like a nice, low cost, all sand design.

[spritzer]

7 hours ago, nopants said:

No LV supplies and no sinking, seems like this was what the exstata should have been.

It will need some heatsinking if we are to run it at reasonable current levels aka. mafia style.  That said, nothing too excessive. 

[chinsettawong]

If everything can be surface mount, how small a footprint can it be?  It would be nice if it can be a small and portable amp.

[nopants]

It will need some heatsinking if we are to run it at reasonable current levels aka. mafia style.  That said, nothing too excessive. 

what are the typical ways of sinking smd parts? i saw kerry do it with the transportable dynalo and it seemed a bit unwieldy, hard to wrap my head around. are you just pressing the board against a surface?

[kevin gilmore]

So we actually can do it as the schematic is drawn and make darlingtons out of 2 different transistors

So ksa1156 and ksc2752 for the output transistors, as the schematic is drawn they are 750mw each the driver transistors are 8mw each so definitely surface mount and the input transistors are 50mw each

I think there might have been a better match for the ksa1156, someone go find it.

also remember that Fairchild was bought by onsemi, so lets make sure the parts are going to be there for a while

[Kerry]

That's the correct compliment per the original NEC datasheet for 2sa1156.  I don't know about future availability.

[spritzer]

Yup that's the correct complimentary part but who knows if they'll continue to make them.  That said, they aren't extremely high voltage so they might be of use to the normal people... 

[kevin gilmore]

single sided even, can add a ground plane on top if necessary, current size 3.45 x 3.7 inches, i'm sure birgir will be able to shrink it in the one direction

you need 2 per channel, stack on top of each other, or next to each other.

I should probably add some mounting holes.

[PretentiousFood]

Just popped it into a Spice. I don't have models for the output devices, so used BC337/327s everywhere (they would blow up, but Spice doesn't care. )

Bandwidth doesn't look too hot, even with the outputs unloaded. May have made a mistake, but the input stage looks like it's starved for current? Replacing the emitter degen Rs with 1k and bumping the ouput stage Rs accordingly (~3k) helps a lot.

Not sure how to attach the .asc, here are some screenshots.

[kevin gilmore]

I am sure there will be value changes once I build the thing, and since the outputs will be heatsinked it's easier to turn up the bias a bit an decrease the values of the voltage divider resistors

[JimL]

deleted

Edited November 25, 2016 by JimL

[gepardcv]

Should the output devices be matched?

[kevin gilmore]

matching the output devices not a bad idea

board file posted, new software may not be compatible with some gerber viewers, I did it in rs274x although I can do the new gerber x2 version

changed a number of resistors, its less than 1db down at 20khz.  about 7.5 watts per board, 15 watts per channel, 30 watts total plus the power supply

joamat i'm expecting you should be able to have this done by the end of the day

 

 

Edited November 26, 2016 by kevin gilmore

[JoaMat]

not possible, I am several hours ahead of you Americans

[wink]

So, before breakfast tomorrow is a go...?

[spritzer]

It will surely be done by 6pm on Monday!! 

I'll look into updating the board a bit.  Might change it all SMD and into stereo channels.

[JimL]

So, if I understand this, it looks like this design really likes to be driven balanced.  If you have an unbalanced input (e.g. RCA) the negative input is grounded and the stator gets no signal, unlike the differential input designs which become a long-tailed phase splitter if you ground the negative input, thus providing a balanced output.  Correct?

Edited November 27, 2016 by JimL

[kevin gilmore]

no this is the krell/Marantz style of doing unbalanced to balanced via the low impedance - cfa input.

Marantz still does this in products today.  differential gain way less than .1db, look at the graph. schiit has trademarked this as "schiit pivot point"

my unbalanced/balanced/cast input board does the same thing.

if cavalli was not such an idiot, he could have done the same thing with his liquid cabon instead of a cheap front end opamp. By the way, liquid gold is the same fucked up mess.

Edited November 27, 2016 by kevin gilmore