This supply is derived from my ion-chamber project:
https://www.diyaudio.com/community/...ve-this-electronic-filter.395321/post-7256296
Since it has little relevance to audio, I am not going to document it in its entirety, but the supply section could be of some use.
It is a low-power bias source, that could be used in an ESL, electrostatic headphones, special condenser microphones, etc. In my case, it is set at 4.7kV, the optimum for the geometry of my chamber, but it is adjustable from ~3 to 9kV, and the circuit can easily be adapted for practically any other voltage range.
Its current consumption is low enough to allow battery operation, and the output is remarkably clean, without using large filter caps thanks to an electronic filter. At a supply voltage of 7.5V, the consumption of the HV section alone is ~10mA, rising to ~16mA with the filter.
Although it is designed for static bias only, it can deliver a non-negligible current too: when I connect a 100 megohm test probe at the output, the regulation absorbs the supplement, and the current consumption rises to ~50mA.
Here is the circuit:
And here is how it looks like physically:
It is essentially a flyback converter, followed by a voltage tripler.
I will discuss the circuit and the components in the next post
https://www.diyaudio.com/community/...ve-this-electronic-filter.395321/post-7256296
Since it has little relevance to audio, I am not going to document it in its entirety, but the supply section could be of some use.
It is a low-power bias source, that could be used in an ESL, electrostatic headphones, special condenser microphones, etc. In my case, it is set at 4.7kV, the optimum for the geometry of my chamber, but it is adjustable from ~3 to 9kV, and the circuit can easily be adapted for practically any other voltage range.
Its current consumption is low enough to allow battery operation, and the output is remarkably clean, without using large filter caps thanks to an electronic filter. At a supply voltage of 7.5V, the consumption of the HV section alone is ~10mA, rising to ~16mA with the filter.
Although it is designed for static bias only, it can deliver a non-negligible current too: when I connect a 100 megohm test probe at the output, the regulation absorbs the supplement, and the current consumption rises to ~50mA.
Here is the circuit:
And here is how it looks like physically:
It is essentially a flyback converter, followed by a voltage tripler.
I will discuss the circuit and the components in the next post
Last edited:
Yes and no: the HV converter core is pretty simple, with a commodity transistor, MOS and a general purpose CMOS IC.
Most of the "complication" resides in the bells and whistle: the electronic filter, its supply and the Lo-Batt detector.
A plain-vanilla version could dispense with all of it, and use a 1 or 2µF 5kV capacitor as a filter, and would result in a further reduction of the power consumption, but it would be much bulkier, costlier and seriously hazardous.
I worked for about two months on this supply, ~1 hour every other day. I am retired, I have all the time that I need, and I have no deadline to respect
Yes: they are carbon-composition resistors, able to withstand very large pulses for a short time. They do not seem to intefere with the operation of the filter, and anyway the 1n5 caps are X7R, thus rather lossy anyway
Here is the corrected schematic:
As I said earlier, the topology is of the flyback variety, and the control is achieved via the repetition rate: the conduction time is fixed, and determined by R1/C1, and the OFF time is controlled by the feedback transistor; R2 limits the maximum frequency.
This variable-frequency mode of control is generally disregarded, but it is the only way to achieve a really low current consumption, and although the output ripple is variable in frequency, its amplitude remains ~constant.
The 5V supply voltage serves as a reference and is compared to the output voltage via suitable resistors. The regulation is reasonably stable and accurate, since the stability of the low-power regulator is good; one Vbe remains uncompensated, but it is 1/10th of the reference, and doesn't degrade much the stability of the voltage.
If required, it would be possible to include a PNP follower to compensate the Vbe.
With this scheme, the output voltage is inherently negative: that's a requirement to properly bias the ion-chamber: the inner electrode needs to be positive wrt. the outer wall, and it would be unpractical to have all the detection circuitry sitting at +5kV.
For most other applications like ESLs, the polarity doesn't matter, and the circuit is usable as it is. In case it matters, it is probably possible to reverse the topology to generate a positive output.
The PWM signal drives a logic-level MOSfet, made necessary by the 5V drive. The MOS is quite hefty for such a low-power circuit, but the peak current is substantial, and the flyback voltage seen on the drain is high too.
The transformer is a critical part of the circuit, and I didn't build it myself, even though I have the means to do it: it is a time-consuming chore if it has to be done properly, something I prefer to avoid.
Instead, I opted for a cheap, ready-made model:
https://www.banggood.com/20Pcs-10KV...-1355350.html?rmmds=category&cur_warehouse=CN
They are rudimentary, and would probably not withstand 10kV for much more than the duration of a science-fair, but I use it at a fifth of its theoretical capacity, and I have improved it somewhat.
Anyway, the secondary is wound on a proper, sectioned coil-former, and it is impregnated. The rest of the transformer is bare, and held together with ordinary sello-tape.
It has two primary windings: the one I used, and a feedback one having a thinner wire gauge and slightly more turns, intended for self-oscillating converters. I didn't use it, but it could be used to create another, floating auxiliary supply.
The inductance values shown on the schematic were the raw, initial values.
I improved it by heating it whilst applying pressure on the two E's, and I applied an impregnating varnish, keeping the assembly under pressure. When it was cured, I heated it one more time, and applied a heat-shrinking band around the core.
When all was done, the primary inductance had risen to 65µH, and was rock-stable (it wasn't the case before the treatment).
To be continued....
As I said earlier, the topology is of the flyback variety, and the control is achieved via the repetition rate: the conduction time is fixed, and determined by R1/C1, and the OFF time is controlled by the feedback transistor; R2 limits the maximum frequency.
This variable-frequency mode of control is generally disregarded, but it is the only way to achieve a really low current consumption, and although the output ripple is variable in frequency, its amplitude remains ~constant.
The 5V supply voltage serves as a reference and is compared to the output voltage via suitable resistors. The regulation is reasonably stable and accurate, since the stability of the low-power regulator is good; one Vbe remains uncompensated, but it is 1/10th of the reference, and doesn't degrade much the stability of the voltage.
If required, it would be possible to include a PNP follower to compensate the Vbe.
With this scheme, the output voltage is inherently negative: that's a requirement to properly bias the ion-chamber: the inner electrode needs to be positive wrt. the outer wall, and it would be unpractical to have all the detection circuitry sitting at +5kV.
For most other applications like ESLs, the polarity doesn't matter, and the circuit is usable as it is. In case it matters, it is probably possible to reverse the topology to generate a positive output.
The PWM signal drives a logic-level MOSfet, made necessary by the 5V drive. The MOS is quite hefty for such a low-power circuit, but the peak current is substantial, and the flyback voltage seen on the drain is high too.
The transformer is a critical part of the circuit, and I didn't build it myself, even though I have the means to do it: it is a time-consuming chore if it has to be done properly, something I prefer to avoid.
Instead, I opted for a cheap, ready-made model:
https://www.banggood.com/20Pcs-10KV...-1355350.html?rmmds=category&cur_warehouse=CN
They are rudimentary, and would probably not withstand 10kV for much more than the duration of a science-fair, but I use it at a fifth of its theoretical capacity, and I have improved it somewhat.
Anyway, the secondary is wound on a proper, sectioned coil-former, and it is impregnated. The rest of the transformer is bare, and held together with ordinary sello-tape.
It has two primary windings: the one I used, and a feedback one having a thinner wire gauge and slightly more turns, intended for self-oscillating converters. I didn't use it, but it could be used to create another, floating auxiliary supply.
The inductance values shown on the schematic were the raw, initial values.
I improved it by heating it whilst applying pressure on the two E's, and I applied an impregnating varnish, keeping the assembly under pressure. When it was cured, I heated it one more time, and applied a heat-shrinking band around the core.
When all was done, the primary inductance had risen to 65µH, and was rock-stable (it wasn't the case before the treatment).
To be continued....
According to the datasheet I found, the 10N15's on resistance is only specified at 10 V gate-source voltage and the L in 10N15L stands for lead-free. The threshold voltage is 2 V to 4 V, so it should be usable at 5 V if you can allow a higher on resistance.
Is it big enough to survive low supply voltages without heatsinking and without undervoltage lockout?
Is it big enough to survive low supply voltages without heatsinking and without undervoltage lockout?
I have measured its threshold voltage, and it is 1.2V, thus most likely logic-level; there are other letters on the package, but I don't remember exactly which (and they are barely legible).
Even if the battery voltage drops below 5V, the probability that an exhausted 9V battery manages to damage a device of this size is slim.
But your warning is certainly useful: a logic-level device capable of 10A/150V is required there, and not all 10N15 are made equal
Even if the battery voltage drops below 5V, the probability that an exhausted 9V battery manages to damage a device of this size is slim.
But your warning is certainly useful: a logic-level device capable of 10A/150V is required there, and not all 10N15 are made equal
Very Cool Stuff !!
I built something like that a while back found here,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-2531218
to here,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-2848194
and a revision to a mistake in the print,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-5487247
That supply still works to this day as built !!
I am just now actually revisiting these types of circuits again lately myself, I am planning slightly higher voltage one of up to 25 KV ( maybe even bigger for Kicks........ ) , or so and one at a much lower voltage of 2.5kv max but higher current (.200 amps or so) to power some HV amps I have planned for ESL work.
This would make for a much safer unit with smaller valued caps in it for a fast discharge rate when it is shut down,
Since I made the circuit it has performed great but I did not much like using and designing the 555's as the Osc, as only certain types will work at +180khz, since then I have been looking at other better ways of excitation of the FET's.
It packs a walup and can maintain a constant arc while using up every bit of current the 50 watt transformer that is powering it can supply, I have yet to use a bigger one to see just exactly how much power it can produced but as is.
I actually just found that thing yesterday to start forming a new case for it, as it got kinda destroyed but the circuit still works as it is now 10 years old !!!
As Usual, I will be following this project closely, as I do along with the rest of your threads.
Great Stuff and Keep em Coming !!
Cheers !!
jer
I built something like that a while back found here,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-2531218
to here,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-2848194
and a revision to a mistake in the print,
https://www.diyaudio.com/community/...ulation-and-mylar-coating.186530/post-5487247
That supply still works to this day as built !!
I am just now actually revisiting these types of circuits again lately myself, I am planning slightly higher voltage one of up to 25 KV ( maybe even bigger for Kicks........
) , or so and one at a much lower voltage of 2.5kv max but higher current (.200 amps or so) to power some HV amps I have planned for ESL work.This would make for a much safer unit with smaller valued caps in it for a fast discharge rate when it is shut down,
Since I made the circuit it has performed great but I did not much like using and designing the 555's as the Osc, as only certain types will work at +180khz, since then I have been looking at other better ways of excitation of the FET's.
It packs a walup and can maintain a constant arc while using up every bit of current the 50 watt transformer that is powering it can supply, I have yet to use a bigger one to see just exactly how much power it can produced but as is.
I actually just found that thing yesterday to start forming a new case for it, as it got kinda destroyed but the circuit still works as it is now 10 years old !!!
As Usual, I will be following this project closely, as I do along with the rest of your threads.
Great Stuff and Keep em Coming !!
Cheers !!
jer
C2 and C4 need to be low-esr, preferably of the solid polymer type: C2 because it is at the output of the LDO, and C4 because of the high current pulses generated by M1.
The secondary of the transformer drives an ordinary voltage tripler. The voltage ratings of the diodes and capacitors need to be sufficient for the intended application.
The number of stages can be adapted to the voltage to be generated, from 1 to ....n. In the single stage case, you will have to take into account the polarity of the windings, to catch the flyback pulse correctly.
The feedback resistors R7 to R9 can be modified accordingly.
I used relatively low values, because the stability of high-ohm resistors tends to be poor.
The converter also feeds the auxiliary supplies via L1. L1 establishes the share of the power allocated to the auxiliary supply: if it is too large, the current will be insufficient to feed the TLO62, and the voltages will crumble. If it is too small, the excess current will be wasted in the zeners and increase the global consumption.
Most of the filtering after the tripler is done by an electronic filter: large-value, high-voltage caps are bulky, costly and dangerous. Here, all that is required is a pair of 1.5nF caps.
Series resistors of the carbon-composition variety protect the circuits in case of an accidental discharge.
For more details on the filter, see the original thread: https://www.diyaudio.com/community/...ve-this-electronic-filter.395321/post-7256296
I have also placed a spark-gap at the HV output: during the tests, the feedback loop was accidentally interrupted, and the output voltage rose well beyond 10kV. This caused arcing everywhere, but fortunately, the only casualty was the TLO62.
With a deterministic overvoltage path, such a situation shouldn't cause any damage
The secondary of the transformer drives an ordinary voltage tripler. The voltage ratings of the diodes and capacitors need to be sufficient for the intended application.
The number of stages can be adapted to the voltage to be generated, from 1 to ....n. In the single stage case, you will have to take into account the polarity of the windings, to catch the flyback pulse correctly.
The feedback resistors R7 to R9 can be modified accordingly.
I used relatively low values, because the stability of high-ohm resistors tends to be poor.
The converter also feeds the auxiliary supplies via L1. L1 establishes the share of the power allocated to the auxiliary supply: if it is too large, the current will be insufficient to feed the TLO62, and the voltages will crumble. If it is too small, the excess current will be wasted in the zeners and increase the global consumption.
Most of the filtering after the tripler is done by an electronic filter: large-value, high-voltage caps are bulky, costly and dangerous. Here, all that is required is a pair of 1.5nF caps.
Series resistors of the carbon-composition variety protect the circuits in case of an accidental discharge.
For more details on the filter, see the original thread: https://www.diyaudio.com/community/...ve-this-electronic-filter.395321/post-7256296
I have also placed a spark-gap at the HV output: during the tests, the feedback loop was accidentally interrupted, and the output voltage rose well beyond 10kV. This caused arcing everywhere, but fortunately, the only casualty was the TLO62.
With a deterministic overvoltage path, such a situation shouldn't cause any damage
Last edited:
