Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Tube and DMOSFET electrostatic headphone amplifiers.
New Posts  All Forums:Forum Nav:

Tube and DMOSFET electrostatic headphone amplifiers.

post #1 of 9
Thread Starter 

Here are schematics and simulations of 2 electrostatic amplifiers. These amplifiers were previously posted both on esldiy and on my website on December 26th. 2013

 

The first uses DMOSFETS throughout.

 

 

 

This second uses both DMOSFETS and 6SN7 tubes. 

 

 

The credit for one unusual feature of these designs goes to John Broskie. By tying the feedback network from the output to the +ve input on the opamp, and thence to the NEGATIVE opamp rail, this network controls the quiescent DC voltage at the output; the ratio of the 2 resistors times the negative rail voltage giving the value of the quiescent voltage: 22*15=330V 

 

w


Edited by wakibaki - 1/24/14 at 4:28pm
post #2 of 9

I've compared (simulated) both your amp and what I suggested here

 

What stands out:

 

- your feedback scheme is much more stable (wrt oscillations) than the one I was using;

- on the other hand, using the top mosfet in srpp makes quite a positive difference in current drive, notably at as frequencies goes up.

 

it's a bit difficult to compare thd directly, since it depends a lot on the opamp models used.

 

So mixing both, we've got this one:

 

At the input, I'd take it easy, a drv134 or a that1646 would be perfect.

post #3 of 9
Thread Starter 
Quote:
Originally Posted by 00940 View Post
 

I started from a mix of idea of Tubecad

 

Where did you credit John Broskie for the use of his idea?

 

I always acknowledge the source of any ideas I incorporate into my designs and I'm not really interested in collaborating with someone who doesn't.

 

w

post #4 of 9

some are so old that it is a major research project to trace true origins - particularly if you credit tube circuits as the inspiration for many fet/mosfet circuits

 

or take an even more abstract view of topology as relatively independent of device type, even polarity/complementary - many complementary circuits were analyzed as 1/2 circuits of one polarity

 

substitutions/reflections/inversions are all really expected tools in a circuit designers box

post #5 of 9

@Wakibaki: First off, I honestly had never seen your schematic earlier. It was posted on websites I don't usually read. Had I seen it, I wouldn't have posted mine since yours is working better. I'd rather have contacted you about it. At the time of posting, I really thought that this particular mix of parts hadn't been done before. Clearly I was wrong.

 

@all: Secondly, I should probably have linked to the RKV which was my main source of inspiration. But since I had just posted a thread about it, with proper references, and that the posting of the schematic was, in my mind, just a continuation of it, I must admit I forgot. That much I'm willing to confess. Please accept my apologies for a moment of carelessness.

 

Then, to answer the last direct question, I didn't credit Broskie because the schematic I posted doesn't make use of his material. It is the biasing scheme of the RKV I used, not his. The one thing Wakibaki gives credit to Broskie for is exactly the one thing that makes our schematics quite different. Ironically, it is also the one thing that makes his superior.

 

When I wrote I started with Tubecad, it is literally that seeing that schematic made me think I could extend the "variations on the rkv theme" to electrostatic amps. Broskie's basic hybrid building block is just the same as the rkv (a tube controlled by an opamp wrt gain but also operating points, especially plate voltage). So I tried to see if I could do it somehow differently.

 

Finally, there aren't that many way to skin a cat. I don't pretend the schematic I posted is something revolutionary or even new; all the elements are well known if not trivial. There aren't many tubes able to sustain the voltages needed and even fewer high voltage depletion mosfet with low capacitance. The controversy between srpp and resistor loading is spread over so many threads, who am I supposed to credit ?

post #6 of 9

I did something just like this back in 2000, as soon as I find the schematic I will post it.

But the output capacitors influence the sound, so no matter what you do, the output caps

are a significant part of the circuit. Mine lifted everything by 300V so no output caps.

 

Various versions of this have been around a long time.

 

sennheiser hev70  opamp and mosfet current source and mosfet drive, capacitor output

http://gilmore.chem.northwestern.edu/he70-1.jpg

 

original stax portable

http://gilmore.chem.northwestern.edu/stxp.gif

opamp driving grounded grid transistor, resistor load, capacitor output

 

stax srm300

http://gilmore.chem.northwestern.edu/srm300.pdf

bigger version of opamp driving discrete stuff

post #7 of 9
Thread Starter 
Quote:
Originally Posted by kevin gilmore View Post
 

...the output capacitors influence the sound, so no matter what you do, the output caps

are a significant part of the circuit...

 

 

If it's good enough for Broskie, it's good enough for me.

 

w

post #8 of 9
Thread Starter 

I think the all-depletion-MOSFET version shown here:-

 

 

...will be cheaper, since there's no necessity for a heater supply. Since I've designed the boards to be 100*100mm, I can get them fabbed at minimum cost, this one is all thru-hole, so easy for diy assembly. I've got the case designed in Autocad for cutting on my CNC machine, all rectangular parts, aluminium and perspex, KISS. 5in*5in*4in deep. There's a digital readout for the bias voltage, that's the righthand pot. The readout uses an off-the-shelf ebay voltmeter. I have nearly all the parts including the FETs and 1250V WIMA caps. I just have to buy some transformers to be absolutely sure about the fit.

 

The layered design means that the PSU and its tracking is separated from the signal board by a ground plane. All the power drops down to where it's required on risers, completely orthogonal to the signal board. Standard PCB stand-offs tie the structure together.

 

Progress on the 'phones is slow, but steady. I bought the earpads made to fit off ebay.

 

 

The whole thing, amp and 'phones, could be supplied as a flatpack kit, including pretensioned Mylar diaphragms. The parts are all simple in the extreme, and could all be cut from sheet perspex with the addition of some nylon studding (threaded rods) and wingnuts. Oh, and I need a metal (aluminium) back for the amp to act as a heatsink. I can send them out for manufacture, it's a simple job on a CNC machine, and a lot of 10 wouldn't break the bank. They'd be the most accessible (cheapest) electrostatic 'phones + amp in existence. If that's what I decide to do. I kind of like not being a MOT, so I might just supply the boards at cost like I did with some of my other designs.

 

w

post #9 of 9
Thread Starter 

THD @ 20kHz for the all-DMOSFET version, 5mA quiescent per side, loaded each side with 1M||120pF. Should lose some even harmonics due to push-pull cancellation. All you can say is that THD will be lower than the 0.88% total shown here with ~500V pk-pk, probably in the region of 0.2%. You can probably get LTSpice to calculate the THD from the subtracted outputs, but I can't see a simple way, and TBH I don't care that much, these numbers are best used to pick between paper designs, not for forecasting an exact performance.

 

 

w

New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Tube and DMOSFET electrostatic headphone amplifiers.