Electrostatic Amp THD Specs/Measurments for the Full Audio Band ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Electrostatic Amp THD Specs/Measurements for the Full Audio Band ?

I can't find a single electrostatic amp characterized with THD specs up to 20Khz or measured by a reviewer. All THD specs I've seen, are only for 1Khz -which was fine in the 50's, but not today, especially when we are talking about High End. Do you know of any such specs / measurements?

The reason is, I'm trying to develop an "impossible" electrostatic headphone amp. Initial goal: ultra low distortion <0.01% at the full audio spectrum, ultra low noise, high output voltages -about 1400Vpp (500Vrms) -solid state, DC, direct drive. Complementary output (which would be the easy solution) is out of the question, as there are no PNP bipolar or P-channel mosfets capable to withstand such voltages. Simply using two N-channel mosfets is an option but highly unlikely (or too difficult) to achieve such low levels of distortion at such a high voltage span, and at least up to 20khz. So I'm researching other topologies, even some completely exotic ones, that I'm currently trying.

Which is why I would like to know what has been achieved so far in terms of THD levels for the full spectrum -as a reference.
 
Last edited:
Why would you need 500Vrms for headphones?
Two reasons:
1) I want it to be able to drive anything on the market (there are a couple of amps with a similar output)
2) I want to drive my current DIY electrostatic headphones to the maximum and make another version with an increased space between the diaphragm and the stators in order to get a lower frequency response.

And can't you use 900V PNP in cascode to get 1800V max ?
Nope. You can cascade more than one transistors to make a high voltage one, but you can't have more output voltage sweep than the voltage rating of the transistors, because then each transistor will "see" the full supply voltage difference which will be much higher than their ratings and all of them will get fried.


save some time by reading Gilmore's ES amp threads Headphone Amplification - www.Head-Case.org
Thanks. BTW, I just saw a post of his on head-fi.org about electrostatic amps' voltages and he mentions the ratings as "peak to peak stator to stator" by adding each output peak-to-peak and thus reaching double the actual voltages -which doesn't make sense. For example for +-750volt supplies he mentions 3000V "stator to stator" which is wrong. Even stator-to-stator the pp value is the same.
Electrostatic Amplifiers Voltage ratings | Head-Fi.org
 
Last edited:
You were right, I stand corrected, thanks for the link. I misunderstood the concept. If all transistors on one side are on, the voltage difference of 1600 volts will be divided to the transistors of the other side based on the resistor ladder.

Still, I wouldn't consider this solution as elegant -it's ugly :D
And it requires triple the active parts.
Not to mention they play a dangerous game. Humidity or other factors could shorten the life of the output stage and if one fails, then most likely half the output transistors will fail.
So I will leave it as a last resort in case a more elegant solution fails to provide the "ultra" specs.

P.S.
That topology would be nice to drive large ESL speakers of 10,000 volts or so -although I've thought of something completely different without the risks of a frying avalanche.
 
Last edited:
If you use 2x 2SA1968 etc., it is still very simple and elegant.
Say you use +/-800V rails, you can put a 900V ZD across the C-E of the cascode device as added protection.
Under normal operation, the BJTs will not see full output swing.
So the operating voltage is a fraction of its max. rating.


Patrick
 
If you use 2x 2SA1968 etc., it is still very simple and elegant.
I prefer not to use abandoned parts from an ex-company (Sanyo) that was bought by ONsemi but they are no longer produced -except from a ghost company that I still can't find after searching for half an hour. Unless I'm missing something.

Aside that, its PNP version -2SA1967 at 900V can only withstand about 5ma max continuous current which is too little for my purposes.

Powerful PNP/P channel transistors and mosfets simply do not exist in voltages higher than about 300 volts for bipolar and 500 for mosfets.
 
Last edited:
IXTH10P60 is 600V and stock at Mouser.
But do what please you.


Cheers,
Patrick
I've seen them a few days ago, but somehow I missed them in the last search -too many searches in companies and distributors, for bipolar, mosfets, bmos, etc in a short time. :hypno1:

Still, for 1400Vpp three of those 600V mosfets are needed x2, =6 per stator x 4 stators =24 transistors total for the output stage! Too many in comparison to 8 normally.

It's not just a matter of personal preference, it's the principle of trying first to achieve the best and more efficient, although most difficult, before accepting a compromise.
 
Last edited:
if you want full power response at 20khz into a stax standard load you will need an output stage current of at least 18ma.

current kgsshv-carbons run at +/-450 volt power supplies and can swing at least 800 volts peak to peak stator to ground. 1600v stator to stator

the cfa electrostatic amp can be built with stacks of 4 transistors and up to +/-600v
power supplies.
 
if you want full power response at 20khz into a stax standard load you will need an output stage current of at least 18ma.
The output stage of my current self-designed DIY amp draws in total 44ma which is 22ma per channel (11ma per output), and a bit less on the transistors for a DC-250Khz freq response. Which is why I said previously that 5ma is not enough. I'm going for a higher voltage and more powerful design though.

current kgsshv-carbons run at +/-450 volt power supplies and can swing at least 800 volts peak to peak stator to ground. 1600v stator to stator
As I mentioned previously, that's not right. If each stator swings between +400V and -400V, then measuring stator-to-stator you will get the maximum voltage difference when one stator has +400V and the other -400V (as they have 180deg difference) which will be equal to 800V peak-to-peak, exactly the same with a single stator peak-to-peak measurement.

the cfa electrostatic amp can be built with stacks of 4 transistors and up to +/-600v
power supplies.
Thanks, but first I have to try to avoid stacking transistors on my design.

BTW, how come NO ONE has measured THD on their electrostatic amps, or other reviewed ones along the full audio band, even in websites dedicated to measurements that have measured several amps for dynamic headphones (like innerfidelity.com)? Is it that bad? :D
 
Last edited:
I'd guess its more that its not easy than done and hiding - I don't expect the hi V divider probes for the o'scope to be good enough

you need a (polystyrene?) C divider load that is better than what you want as a measurement floor

and then the ES headphone cable connectors are all custom so you have to hack - possibly sacrifice a extension cable...
 
It's a lot easier to send the output of a typical headphones amplifier into a soundcard input. You just need a cable splitting the signal and a pair of resistors.

To measure the electrostatic amp, you need quite a setup. You may need to open the amp and be comfortable with high voltages.

It seems to me quite obvious why there aren't more measurements. Reviewers usually don't have a degree in electrical engineering.
 
I'd guess its more that its not easy than done and hiding - I don't expect the hi V divider probes for the o'scope to be good enough
you need a (polystyrene?) C divider load that is better than what you want as a measurement floor and then the ES headphone cable connectors are all custom so you have to hack - possibly sacrifice a extension cable...
It's a lot easier to send the output of a typical headphones amplifier into a soundcard input. You just need a cable splitting the signal and a pair of resistors.

To measure the electrostatic amp, you need quite a setup. You may need to open the amp and be comfortable with high voltages.

It seems to me quite obvious why there aren't more measurements. Reviewers usually don't have a degree in electrical engineering.
How about:
1. The manufacturers,
2. The DIY hobbyists who build and design electrostatic amps.

I see DIY schematics here and there, what's the point on designing another different amp, without aiming at better specs?
Because if you don't measure it, you don't care and you cannot optimize it.
If you have measured it and you don't show the results, then it sucks. Simple as that.


As for the rest, it's not that hard to locate each amp's output.
You only need one output and a ground connection -close to the output.

And you only need a resistor divider, a high upper resistor in the range of several hundreds of KΩ up to MΩ to avoid loading the output, split in three or more smaller ones in order to minimize the parasitic capacitance, and a low second resistor in the range of a few hundreds Ω in order to provide a low-z output that will not be affected by the load (probe + instrument) -especially by its capacitance, and will ensure an unaltered high freq. response.

Since a ground connection is not usually provided on the output jack, you have to open the case and do the connection internally,
preferably directly on one output before a capacitor (if any) and a ground connection close to that output.
Also, you should have different dividers for different output voltages in order to keep the current and the output voltage the same.

example (adjust if necessary):
for a +-500V output, 3 x 330K in series (=990K) + a 430 Ω will draw 0.5ma peak (less than 0.2 W rms) and will provide 430mVpp output.

I think that since there is no standard, the actual divider used for the measurement should be mentioned.
Each divider should be placed inside a probe, so there would be different probes for different output voltages.

Of course, the amp should be measured with the headphones connected.
 
Last edited:
MrMagic said:
current kgsshv-carbons run at +/-450 volt power supplies and can swing at least 800 volts peak to peak stator to ground. 1600v stator to stator
As I mentioned previously, that's not right. If each stator swings between +400V and -400V, then measuring stator-to-stator you will get the maximum voltage difference when one stator has +400V and the other -400V (as they have 180deg difference) which will be equal to 800V peak-to-peak, exactly the same with a single stator peak-to-peak measurement.
OK, that's the worst thing I've said -and insisted too, the last few days -embarrassing and my apologies. Probably because I just started with hardware again after many years break. I confused the peak measurement between the two stators (800V) with the peak-to-peak measurement (800V) of a single stator. Unbelievable.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.