Last night I managed to fry one of the audio channels of my smartphone when I accidentally shorted the diaphragm of an electrostatic headphone prototype to the the stator. I suspect this caused the high voltage bias supply to briefly conduct and send a spike back through the step-up transformer and into the output of the headphone amplifier chip. As much as fifty volts could've made it back through the transformer and I suspect the chip only has about a five volt supply. Would a pair of (anti-parallel?) clipping diodes (or LEDs) at the amplifier output be a good start? I only need to supply a volt or two of audio signal to the transformer primary.
Parallel zeners would be a good idea (add a series diode to each so that they only clamp in the forward direction). A possible problem could be very fast rise times of a transient so it might be an idea to add a small series resistance.
I think you can get specific "zeners" for this... high speed... but I couldn't just turn up a part number.
I think you can get specific "zeners" for this... high speed... but I couldn't just turn up a part number.
Hi Guys
For symmetric clipping, zeners are usually placed in series not parallel.
If the signal is really low, then using series diodes in two antiparallel strings would be "aster", especially since you could use 1N914 or similar. These diodes will add distortion unless there is enough of them in series to keep their turn on threshold far enough away from the normal signal that would be across them.
The bias supply for the ESL is supposed to be limited by at least 5M-ohms, which reduces shocks one might get to just a tingle - much less painful than many static discharges you get from carpets. If this is some supply you have built yourself add this resistor feeding the bias voltage to the headphone socket; it could save your life.
Many head phone outputs in small devices are bridged to provide maximum output from the low supply voltage. This means that twice the peak voltage might be available compared to single-rail operation as a faux split rail. Were you able to get inside the phone, protection diodes could be added from rail to output and ground to output (assuming a single rail).
Have fun
Kevin O'Connor
For symmetric clipping, zeners are usually placed in series not parallel.
If the signal is really low, then using series diodes in two antiparallel strings would be "aster", especially since you could use 1N914 or similar. These diodes will add distortion unless there is enough of them in series to keep their turn on threshold far enough away from the normal signal that would be across them.
The bias supply for the ESL is supposed to be limited by at least 5M-ohms, which reduces shocks one might get to just a tingle - much less painful than many static discharges you get from carpets. If this is some supply you have built yourself add this resistor feeding the bias voltage to the headphone socket; it could save your life.
Many head phone outputs in small devices are bridged to provide maximum output from the low supply voltage. This means that twice the peak voltage might be available compared to single-rail operation as a faux split rail. Were you able to get inside the phone, protection diodes could be added from rail to output and ground to output (assuming a single rail).
Have fun
Kevin O'Connor
Thanks for both replies. Good call on the 1n914 - I didn't realize they were so much faster than the 1n400x types. I think I'll try anti-parallel strings of them approximating the output voltage I need and use my scope to measure the discharge transients across the diodes with the source disconnected. As a general rule of thumb, will damage be mitigated if the transient voltage is lower than the power supply voltage? Securing the housing so these discharges aren't possible is obviously the best solution, but I'd like to have a second line of defense. I'm just lucky that the phone is 22 months into a 24 month upgrade plan. 😀
The ESL headphones ought to be driven via a transformer for galvanic isolation, especially where there is a situation like the one you have, driving them off a low voltage solid state thingy like a "smartphone".
Since you did not say how you had configured these prototype ESL headphones, it's hard to know what happened or how to possibly protect the source device.
You said there is a step up transformer, I presume that the primary is connected directly to the input from the smartphone?
I'd give serious consideration to another transformer, like a 1:1 or similar before the step up, and maybe a buffer amp between the two. The buffer amp is a good idea since the current available from the smartphone is severely limited to begin with. The "buffer" amp might want to have some gain, or not.
Better to smoke some opamps or transistors than the source.
_-_-
Since you did not say how you had configured these prototype ESL headphones, it's hard to know what happened or how to possibly protect the source device.
You said there is a step up transformer, I presume that the primary is connected directly to the input from the smartphone?
I'd give serious consideration to another transformer, like a 1:1 or similar before the step up, and maybe a buffer amp between the two. The buffer amp is a good idea since the current available from the smartphone is severely limited to begin with. The "buffer" amp might want to have some gain, or not.
Better to smoke some opamps or transistors than the source.
_-_-
The circuit consists of a bias supply (easiest just to think of it as a 600v battery), transformer secondary, and the two plates of the condenser that comprise the actual driver all connected in series. I have always used a small amp to drive the primary of the step-up transformer before - I was just seeing if the phone could drive it directly. While I agree that it is better to blow an opamp than a source, I'm wondering why either is necessary. Now that I know that these transients are damaging to the primary end of things, I might as well apply protection to any device on the front end.
Hi Guys
Many power amps have protection diodes at the output going to each rail. A head phone output is treated more like any opamp stage, so these diodes are left out.
The fault that caused the transient is unlikely to occur again, but using circuits that are known to be protected as above is the best assurance you can have that they will survive. No need for intermediate transformers beyond the step-up as that will not provide any protection. neither will a buffer be protected unless it has the diodes added - but any circuit you build can have these added for just pennies.
Most products built to directly drive ESL head phones do not have these diodes either.
Have fun
Kevin O'Connor
Many power amps have protection diodes at the output going to each rail. A head phone output is treated more like any opamp stage, so these diodes are left out.
The fault that caused the transient is unlikely to occur again, but using circuits that are known to be protected as above is the best assurance you can have that they will survive. No need for intermediate transformers beyond the step-up as that will not provide any protection. neither will a buffer be protected unless it has the diodes added - but any circuit you build can have these added for just pennies.
Most products built to directly drive ESL head phones do not have these diodes either.
Have fun
Kevin O'Connor
Hi,
I assume that a flashover from the charged membrane to the stator caused the problem and not the bias-supply itself.
A protective or current limiting resistor between the HV-Supply and diaphragm cannot help here, as the membrane´s charge will discharged into the stator by arcing.
Especially metallized membranes will completely discharge quickly and at high current rates due to their low impedance.
Diaphragms with a high resistive coating discharge much slower and with much lower current rates.
Its the high dI/dt that causes spikes to travel through the transformer to the amp output.
The best measures against such conditions are high-Z coated membranes and very fast protection devices like TVS (Transient Voltag Suppressors) or fast surge suppressors, which are similar to blinking lamps, in parallel to the ESL-element.
The devices need to be bipolar, or connected to allow for bipolar protection.
You must carefully choose, for low capacitance devices, low leakage and Avalanche voltages in dependance of the frequency of the overvoltage.
A arc will typically contain HF.
Also, You don´t want the device to limit and restrict too early.
As a start, a symmetrical ESL stays within its linear range as long as the peak signal voltage remains <0.5xBias voltage.
See Microsemi (MicroNote125) and Diotec (Notes on TVS Diodes), Littlefuse (AN9768), HighEnergyDevices (AN-UI-10) et al for App-notes.
jauu
Calvin
I assume that a flashover from the charged membrane to the stator caused the problem and not the bias-supply itself.
A protective or current limiting resistor between the HV-Supply and diaphragm cannot help here, as the membrane´s charge will discharged into the stator by arcing.
Especially metallized membranes will completely discharge quickly and at high current rates due to their low impedance.
Diaphragms with a high resistive coating discharge much slower and with much lower current rates.
Its the high dI/dt that causes spikes to travel through the transformer to the amp output.
The best measures against such conditions are high-Z coated membranes and very fast protection devices like TVS (Transient Voltag Suppressors) or fast surge suppressors, which are similar to blinking lamps, in parallel to the ESL-element.
The devices need to be bipolar, or connected to allow for bipolar protection.
You must carefully choose, for low capacitance devices, low leakage and Avalanche voltages in dependance of the frequency of the overvoltage.
A arc will typically contain HF.
Also, You don´t want the device to limit and restrict too early.
As a start, a symmetrical ESL stays within its linear range as long as the peak signal voltage remains <0.5xBias voltage.
See Microsemi (MicroNote125) and Diotec (Notes on TVS Diodes), Littlefuse (AN9768), HighEnergyDevices (AN-UI-10) et al for App-notes.
jauu
Calvin
Hi Guys
MOVs and similar transient absorbers deteriorate every time they absorb a pulse of energy. At some point they are dead and do nothing. They may add THD in this type of application, so overall are not recommended. fast diodes or a spark gap would be better, but these will only protect the transformer.
Under normal conditions, the stators have no voltage on them. Charge and fault currents are limited by the usual 5M+ resistance between the bias supply and the plates. If this R is not present then you have a potentially fatal shock hazard.
Have fun
Kevin O'Connor
MOVs and similar transient absorbers deteriorate every time they absorb a pulse of energy. At some point they are dead and do nothing. They may add THD in this type of application, so overall are not recommended. fast diodes or a spark gap would be better, but these will only protect the transformer.
Under normal conditions, the stators have no voltage on them. Charge and fault currents are limited by the usual 5M+ resistance between the bias supply and the plates. If this R is not present then you have a potentially fatal shock hazard.
Have fun
Kevin O'Connor
- Status
- Not open for further replies.