I would also be concerned about the normal signal voltage, although such a failure would make matters worse.
Yes, as a very worst-case approximation.
Edit: if both the amplifier and the listener are grounded and the insulation of both sides of the headphone doesn't work, you might get twice the current I calculated.
Last edited:
yes, basic insulation is part of the normal function. But accounting for abnormal things like faults or damage often requires additional measures. Like 'double insulation' or an earthed shell, for instance.
That's exactly my point
After a large cup of coffee I looked at the schematics posted
Boy am I scared!
"All" of them use differential +/- 300-400 V supplies 😱
No fear of shorting capacitors since none are used; we can say they come "pre-shorted"🥴
No or very small DC at output only because output tubes are balanced which as we know will work fine ... until it fails.
Considering average consumer product non-failure lifespan, amps can practically be guaranteed to fail sometime during User life.
Specially in Tube equipment, where tubes constantly degrade just by being ON, no need for actual "failures" for that.
There's a reason tubes are easily removable without a soldering iron and plug into a socket instead.
Meaning even without "a failure", tubes can easily drift and apply significant DC voltage to output connectors.
Given the deadly voltages involved, I would NOT avoid galvanic isolation, either transformers or capacitors.
Even a 5k series resistor as mentioned will not make output safe.
DC current can be, say, 300V/5k=60mA, way above the established safe level.
Twice as much if failure makes one anode shoot up, the other shoot down.
No practical fuse will protect user.
You like electrostatic Earphones?
So be it, I guess sound must be superb.
Will fuses make them safe by themselves?
Not in my book.
Too crude devices for that delicate and vital job.
As I mentioned above: you CAN design a circuit which detects DC, also too high AC (saving you from deafness) and trigger a relay.
IF relay coil is triggered by an external circuit, applying enough current to coil when a problem is detected by a dedicated circuit, yes, it can open in milliseconds.
But to expect it to be triggered by audio signal itself, with no "help", is marginal at best.
And as was also mentioned above, coil can not differentiate between Audio signal or DC by itself.
Electrostatic headphone are the audiophile counterpart to eating Fugu pufferfish. The greatest treat is when it starts to tickle...
How about this as an overall concept to protect users from high-voltage failures:
1. Use pull-down resistors as in the more receng estat amp designs. The resistors will work instantaneously to limit the current in case of a failure.
2. Add a current sensing that triggers a relay to disconnect the output if there is too much (DC) current flowing for too long (say 30 mA for 1 second?).
(1) is done all the time with current estat amps, and it seems good enough for Stax to get a CE certificate.
(2) could be achieved with a sensing coil wound around the wire carrying the output. Or an optocoupler sensing the voltage across the pull-down resistors. Or something else? It should be simple and reliable.
1. Use pull-down resistors as in the more receng estat amp designs. The resistors will work instantaneously to limit the current in case of a failure.
2. Add a current sensing that triggers a relay to disconnect the output if there is too much (DC) current flowing for too long (say 30 mA for 1 second?).
(1) is done all the time with current estat amps, and it seems good enough for Stax to get a CE certificate.
(2) could be achieved with a sensing coil wound around the wire carrying the output. Or an optocoupler sensing the voltage across the pull-down resistors. Or something else? It should be simple and reliable.
I believe that if there is a real risk of dying (just as it's seems) then it's seems we actually are in a cul de sac here.
Mind you, I would love it if some good designer could accommodate the OP's legitimate request with an original and dedicated project, but until now...
Anyway, since nothing is unbreakable not even the components of an eventual circuits to detect anomalies and therefore any solution of this type remains unacceptable. IMHO
Furthermore, even assuming that these hypothetical circuits to detect anomalies were infallible (and they are not) then they would affect the SQ.
This way they would nullify the main reason why you buy an electrostatic headset, the unparalleled SQ.
Hence the already mentioned cul de sac.
Mind you, I would love it if some good designer could accommodate the OP's legitimate request with an original and dedicated project, but until now...
Anyway, since nothing is unbreakable not even the components of an eventual circuits to detect anomalies and therefore any solution of this type remains unacceptable. IMHO
Furthermore, even assuming that these hypothetical circuits to detect anomalies were infallible (and they are not) then they would affect the SQ.
This way they would nullify the main reason why you buy an electrostatic headset, the unparalleled SQ.
Hence the already mentioned cul de sac.
And to add to this, the detection of DC requires some low pass filtering to distinguish from ultra deep bass notes of great amplitude - not uncommon in actual music. As a consequence, response time will at least be some 10msec - too long for a secure interrupter.
Using a transformer or capacitor in the amplifier output was mentioned as a safe method to isolate the user from a high-voltage fault at the amplifier output. A current flow of 10 ms would correspond to a lower cut-off frequency of 100 Hz for the capacitor or transformer. That can't be right.
Is there any data out there related to the (safety-)effects of the duration of current flow?
There will be some kind of inverse correlation between critical pulse amplitude and pulse duration.
I know that electrostatic discharge for instance may peak to some 10amps - for few nanoseconds.
Making these annoying but harmless.
I know that electrostatic discharge for instance may peak to some 10amps - for few nanoseconds.
Making these annoying but harmless.
In hospital TV series they always set the energy of the defibrillator to some value specified by the doctor when they try to reanimate someone, so chances are that it is something proportional to I2t that matters for short shocks.
There is a graph about shock currents and durations in https://en.m.wikipedia.org/wiki/Residual-current_device
There is a graph about shock currents and durations in https://en.m.wikipedia.org/wiki/Residual-current_device
Last edited:
In Italy the installation of defribillators called DAE (Defribillator Automatic External) in easily accessible places frequented by the public is mandatory by law in various public buildings.
Those types are automatic or semi-automatic and adapt to the situation running a software that makes the diagnosis.
They are also even suitable for children aged 8 and up (however, the related plate must be chosen in advance).
In case of life threatening, the Italian law allows its use even by a person who has not been trained in its use, and therefore he is not even a doctor, nor even a specialized nurse.
Those types are automatic or semi-automatic and adapt to the situation running a software that makes the diagnosis.
They are also even suitable for children aged 8 and up (however, the related plate must be chosen in advance).
In case of life threatening, the Italian law allows its use even by a person who has not been trained in its use, and therefore he is not even a doctor, nor even a specialized nurse.
That IEC reference is quite useful! I attached the diagram to this post. Looking up the Wikipedia reference and translating the Chinese description, I got the following:
IEC 60479-1 specifies the four zones of current-time relationships and describes the pathological physiological effects on the human body within each zone, as shown in Figure 1. The horizontal axis represents the current through the human body, and the vertical axis represents the duration of the current. The meaning of each zone is as follows: AC-1 zone indicates no perception; AC-2 zone indicates perception; AC-3 zone indicates reversible muscular contractions; AC-4 zone indicates possible irreversible effects; AC-4-1 zone indicates up to 5% probability of ventricular fibrillation; AC-4-2 zone indicates up to 50% probability of ventricular fibrillation; AC-4-3 zone indicates more than 50% probability of ventricular fibrillation. ... For example, if the current through the human body exceeds 30 mA, the person is likely to be electrocuted unless the current can be cut off within a very short time.
In other words: according to the IEC 60479-1 diagram, anything longer than 5 s should remain below 5 mA to prevent irreversible effects, and below 30 mA to prevent electrocution. For shorter pulses, the current limits are a bit higher according to the diagram.
5 mA or 30 mA does not see like much, but it's actually just within the needs for electrostatic headphones. My own measurements suggest that about 4 mA peak is enough for very loud music playback (see https://www.diyaudio.com/community/threads/open-source-dht-estat-headphone-amp-osdeha.407679 ). Also, due to the capacitive impedance of estats, the current peaks will appear with high-frequency signals, which should be easy to distinguish from DC.
Attachments
Pickering has very low capacitance 1.5kV reed relays, 131 serie.
For example, the 5V version has a 250ohm coil, off voltage 0.5V = 2mA, on voltage 3.75V = 15mA, that could be useful as crowbar.
The 12V version, needs even less current, it has a 750ohm coil.
For example, the 5V version has a 250ohm coil, off voltage 0.5V = 2mA, on voltage 3.75V = 15mA, that could be useful as crowbar.
The 12V version, needs even less current, it has a 750ohm coil.
Maybe just the standard 5K resistors in series with the outputs are good enough to prevent unnessary killing of humans. There's some discussions about these over at head-case.org.
Or some real life usecases, i.e. I went through 3 sets of Jecklin Float's over half a century, and I'm not dead yet. And these thingies are not even legal, their Hirschmann connectors are only rated for 250V. Any testimonials from the deceased?
Or some real life usecases, i.e. I went through 3 sets of Jecklin Float's over half a century, and I'm not dead yet. And these thingies are not even legal, their Hirschmann connectors are only rated for 250V. Any testimonials from the deceased?
A fuse is unlikely to blow fast enough to protect a human. If you want the fuse to protect the end user you'll have to make the headphones such that they fail with the high voltage shorted to ground. Then the fuse will blow.
Usually semiconductors, such as the output devices in a power amp, do a wonderful job of protecting the fuse by sacrificing themselves before the fuse blows.
Tom