Transformer coupled electrostatic amps

[spritzer]

I'd say go all out, balls to the wall and pit it against the T2.

[jcx]

very curious about es headphone xfmr design - when I looked at the numbers inter and intra-winding capacitances were several times the headphone's load C - and the Iron cross section is stupidly huge to get down to 20 Hz

[kevin gilmore]

very curious about es headphone xfmr design - when I looked at the numbers inter and intra-winding capacitances were several times the headphone's load C - and the Iron cross section is stupidly huge to get down to 20 Hz

And herein lies the problem. Correctly made transformers that don't ring are a real problem.

Only way i know of to do it really right is to wind the things yourself.

[jp11801]

Frank OMG that amp looks stunning can't wait till Mayberry to hear what you've cooked up. I'd love a 211/845 based tube design. the BH sounds great with push pull configured tubes but normally I do not like 34/kt88s as they can sound hard and give me listening fatigue over time. The BH did not have the ill effects that I am normally used to in these tubes. Love them for Marshall stacks though

[FrankCooter]

very curious about es headphone xfmr design - when I looked at the numbers inter and intra-winding capacitances were several times the headphone's load C - and the Iron cross section is stupidly huge to get down to 20 Hz

Maybe you could detail your numbers a bit. I'm curious about how you arrived at these conclusions without any specific design information. The "stupidly huge" core is 2.5 x 3.0" with a 1.5" stack. It tests at -1db at 27hz.

I will test for wave form anomolies, capacitive and otherwise, when the amp is operational. Based on many years experiance with Electra-Print designs, I don't expect to find anything signifigant. Preliminary listening tests were very positive. The amp is a test platform. If I'm missing anything, I'd appreciate enlightenment.

[jcx]

my comments were general - how hard it appears to me to make an appropriate xfmr for for sub 100 pF es headphone loads for the full audio bandwidth

I was playing with the numbers for Koss esp950 - since the K90 uses +/-600V H-bridge drive a similar drive V from the xmfr will require ~800 Vrms

stuff 800 Vrms, 20 Hz, 10,000 gauss Bmax into the Faraday relation and I get ~900,000 N*Ac

at 10,000 turns that's 9 cm^2

9 cm^2 effective core area looks very big to this engineer with very little xfmr design experience - when I only need a (few?) W of circulating pwr max in the headphone Cload

the parasitic capacitance is just a "guesstimate" from some tube power output xfmr resonance peak #, a old Magnetics catalog, and measurements of toroid pwr xfmr similar to those suggested as "poor-man's" push-pull tube amp output xfmr

[jcx]

err - that N*Ac number would be 90,000, not 900,000

[FrankCooter]

My understanding of transformer design is strictly technician level, but I think your design parameters are unrealistic. No electrostatic amp that I'm aware of swings 800Vrms. Most Stax amps swing about 350Vrms. Even the Blue Hawaii "only" swings 1500p-p, which is a little more than 500rms. If you're looking for ruler flat 20Hz response, you probably shouldn't be using a transformer at all. 25Hz is probably about the practical limit.

[spritzer]

It really isn't a great idea to go past 1200V on the stators (bias*2) though as bad things will happen. We need to design our own drivers with 1100V bias...

[FrankCooter]

[ATTACH=CONFIG]3320[/ATTACH]

Up and running today! Here's a "beauty" shot of the interior. Lots of iron in a 6"x13" space. The large black transformers are the Electra-Print outputs. The silver center pair are 1:1 3k interstages made by a now defunct Florida company for an old "Sound Practices" article. I'll neaten things up later. If the amp is a "keeper" I'll convert the screw type terminal strips to soldered turret terminals. At this stage I like to be able to change things around. Already thinking about a "big brother" 845 version. More later.

[Voltron]

I got wood looking at the beauty shot of the latest Cooter Creation. Really want to hear this one because I missed Frank's last two offerings.

[spritzer]

Glad to hear it is a keeper.

[FrankCooter]

[ATTACH=CONFIG]3334[/ATTACH]

Here's a picture where you can actually see something. I'm still in the stumbling around part of the learning process with my first digital camera.

Got about 15 minutes of listening before going to the bathroom and letting my son have a turn. His idea of a test was the "Gladiator" soundtrack at maxed volume. I could hear it all the way across the house (briefly) and then a huge screetching sound. On paper, this amp puts out 2kv p-p. Both sides of my Koss ESP-950's got blown out. Sent them back to Koss. Told them the truth. Going to put that "unconditional" to the test!

[spritzer]

If the bias on the Koss really is 300V (as the circuitry inside the E90 points to) then no wonder they blew up. You should talk to your son about safe listening levels though...

[kevin gilmore]

Frank's son is evidently practiced in the ways of pabbi1.

Not a good thing. Not even for a few minutes.

[FrankCooter]

Frank's son is evidently practiced in the ways of pabbi1.

Not a good thing. Not even for a few minutes.

We had some "words". Actually he felt pretty bad. Just glad it wasn't a pair of O2's. I actually thought about your Chicago "experiance".

If you don't mind, I'd appreciate a little lesson in lab testing electrostatic amps. The problem is of course the very high voltage swings involved in these amps. I've got a 100x, 2kv rated scope probe on the way, so I should be able to do some basic sweeps. I own an old HP 331a distortion analyzer, but it's only good up to 300v. I guess I could try it with some sort of resistor divider network. Any suggestions would be appreciated. New territory for me.

[luvdunhill]

I guess I could try it with some sort of resistor divider network. Any suggestions would be appreciated. New territory for me.

actually a capacitor divider might be more appropriate. I'm at the same place you are, in that I find my tools lacking for testing high voltage amps.

[kevin gilmore]

Lots of problem with the capacitive dividers too... matching over frequency very miserable.

What you want is the old tektronix 1000x probe thats filled with freon. Good for 30kv. And a frequency

response up to about 10mhz. Its about 1.5 inches in diameter, and two different colors of blue.

I have one, i keep it locked away so that no one drops it which would trash it.

The 100x scope probe should do it if you are careful.

Here is what absolutely everyone including me does...when they don't want to blow up something expensive.

Little pomona box with a male bnc on one end, and a female on the other.

Add two series back to back zeners from center to ground. 100 volts.

Insert in the line between the scope probe and the scope.

Ignore the readout pin if you have a scope that uses that.

I'll verify the koss bias voltage tomorrow, but mine certainly does and has worked at 600 volts...

For a very long time.

Need to put the L'espresso on the bench and test for distortion and rise times...

Edited July 5, 2010 by kevin gilmore

[kevin gilmore]

my bad... 75 mhz

these are the new ones with silicone inside them instead.

TEKTRONIX|P6015A|1000X High-Voltage Passive Test | Newark.com

Tektronix P6015A/2R 25 MHz High Voltage Probe - eBay (item 320511121758 end time Jul-30-10 16:09:08 PDT)

eventually you will probably find it on ebay for about $500

[cetoole]

Both sides of my Koss ESP-950's got blown out. Sent them back to Koss. Told them the truth. Going to put that "unconditional" to the test!

Ach, my old baby!

[FrankCooter]

Thankyou, Kevin!

I'll go with diode protection box. Sounds like what you used to see in the crossovers of Klipsch K-horns and LaScalas. The 100x probe cost $30.. My scope (an old tube Techtronix) costs $150.. No great loss if I blow it up. Very thin operation here. Aside from the Electra-Print transformers, if it doesn't come from the surplus store, university cast offs, Ebay, or some other scrounged souce, it doesn't happen. It's actually part of the fun to try to make something from next to nothing. In this case I'll admit to a liitle jelousy, your scope probe is worth more than all my test equipment combined!

[kevin gilmore]

Do not assume that i paid anything even close to retail for the stack of test equipment i personally own.

I've been collecting stuff for 35 years. 4 oscilliscopes at home, all solid state, one portable and one transportable.

signal generators, tdr's, spectrum analyzers,... The pile is probably worth more than the house that contains the stuff.

4 years at zenith during the beginning of the solidstate tv days gave me a lot of knowledge in messing

with high voltage stuff without killing myself. An art that is clearly long gone.

Edited July 5, 2010 by kevin gilmore

[kevin gilmore]

OK, absolutely positively i should never rely on anyone who thinks they know what they are talking about.

The bias on the koss headphones is absolutely close to 600 volts.

Here is the proof

http://gilmore.chem.northwestern.edu/kesp1.jpg

So the tan resistor has absolutely no markings.

The 10 meg resistor is in series with the +630 volt power supply (plus or minus a bit, on batteries, about +600)

The unknown resistor is to ground in parallel with the cap.

You just can't measure that resistor with it in circuit, which is the failings of everyone else.

http://gilmore.chem.northwestern.edu/kesp2.jpg

and this meter measures to 200 meg.

http://gilmore.chem.northwestern.edu/kesp3.jpg

Now usually when i see resistors like this, they are 1 gigohm.

But even if that resistor is as little as 250 meg ohms, the bias is

still (referenced to 600v) a minimum of 575 volts.

I could bring the thing to work where i have meters that measure

into the gigohms, but there is no reason.

THE KOSS ESP950's ARE HIGH BIAS HEADPHONES!

[mypasswordis]

The mystery is solved! Thanks KG. Though that means there is no longer a reason for the squealing issues some people have other than poor QC.

Poor blown-up ESP950s. If I ever have a child I'm locking all my audio stuff up until he/she is old enough to understand not blowing my shit up. All my headphones are vintage, there is no warranty or replacements outside of somehow managing to find another pair.

[FrankCooter]

4 years at zenith during the beginning of the solidstate tv days gave me a lot of knowledge in messing

with high voltage stuff without killing myself. An art that is clearly long gone.

By contemporary standards, I have a fair amount of high voltage experiance. As a teenager in the late '60's I built kilowatt level amateur c.w. transmitters. Components were almost exclusively WWII military surplus. My friend had a 5kw rig that originally came from a marine transmitting station on the California coast. We set up the transmitter in his parents garage and used it on the 80 meter novice band. It took up half the garage and looked like something out of an old Frankenstein movie. It used 813's as driver tubes. The finals were some huge, irregularly shaped cylinders with a sort of central ball shape, perhaps 16" tall by 8" across the central ball. When we'd fire the thing up, strange things woud happen like nearby flourescent bulbs would glow on their own and you'd occasionally see a sort of St. Elmo's fire effect on the backyard closeline. Obviously, our parents (and the FCC) had no idea what we were doing. I look back now and shutter at our crazieness, but it certinly was educational.