That also appears to have distributed inductors along the transmission line, (except inexplicably shorted out, maybe those are Ferrite beads). That would be the way to make a LF transmission line (slow wave propagation speed ). The termination at audio should work then.
Well, not a commercially made ESL. Have you measured the acoustic amplitude put out by the panel versus frequency? Are the impedance peaks and valleys affecting that? If so, then some DIY ESL may be in order.
The shorting turns are there to make it a damped (lumped approximation of a) transmission line, especially at high audio frequencies. It has something to do with preventing edge effects due to the finite size of an ESL 63. The on-axis frequency response is quite good, as far as I know. They certainly sound good!
How about cascading a row of 1kV MOSFETs? But if it has to be tube a few of those colour tv 25Kv/30W ballast triodes in paralell should do. If that is not to your liking, may be the transmitting triode TB4/1500 is, if i remeber correctly, it has a thoriated tungsten heater that could be run underpowered when only low current is needed. Anyway, any transmitting tube should be able to cope with at least double Va at the peaks.
Last edited:
Ihave not seen an answer to this.
These are RF output tubes and would be tank loaded, so output would swing near twice B+.
No?
Decades ago, a late colleague of mine made a direct-drive amplifier for DIY electrostatic loudspeakers using stacked MOSFETs. I don't know much about his design, but I believe it was something with optocouplers cascoded with stacks of MOSFETs. He could get it to work in class-AB by driving the upper and lower optocouplers properly.
It is not straightforward, because you have to ensure a proper voltage distribution between the MOSFETs at all times. The MOSFETs also have to be free of Spirito instability (thermal second breakdown) - no problem back then, but it is nowadays with switching MOSFETs optimized for very low on resistances rather than for (non-class-D) amplifier applications.
Long time ago, before the time of switchers, I bougth 100 pcs (100V "high voltage") 4994 junction fets. At the time, 15 and 25V was the norm 
. After selection for similar Idss I stacked a bunch of them without problem. The only downside was the very low Idss, around 2mA. So, capacitive loading was restricted too that what you can get from a ecc83, but i got the higher voltage swing i needed.
Those switchers are propable trickier, but should be doable.
. After selection for similar Idss I stacked a bunch of them without problem. The only downside was the very low Idss, around 2mA. So, capacitive loading was restricted too that what you can get from a ecc83, but i got the higher voltage swing i needed.Those switchers are propable trickier, but should be doable.
Last edited:
Vertical MOSFETs have always had those second breakdown limits. It wasn’t as pronounced on older types like the IRF240. They handled full power up to about 80 or 100 volts which was enough for practical linear amplifier use. Data sheet didnt indicate any voltage limitations, but the truth is they we’re never tested for it because they beat the tar out of bipolars in general. Handling full power up to about 20 V is all you get today. Comparable to 1970’s switching bipolars, actually.
Even a JFET is going to have some voltage limits on power handling, but it is a different animal from a mosfet so it will fare better than todays switchers. Maybe look into silicon carbide or even GaN, as it will tolerate voltage well. Costly, though.
Even a JFET is going to have some voltage limits on power handling, but it is a different animal from a mosfet so it will fare better than todays switchers. Maybe look into silicon carbide or even GaN, as it will tolerate voltage well. Costly, though.
The USSR made a range of planar external heatsink triodes that might be worth looking into e.g GI-7B, indirectly heated oxide filament so fairly power efficient, rated for a few KV, few hundred watt dissipated anode.
Eimac made some big planar's with big tungsten heaters that had very high ratings for the anode.
Howabout an autoformer step up to get the required stator voltage.
Eimac made some big planar's with big tungsten heaters that had very high ratings for the anode.
Howabout an autoformer step up to get the required stator voltage.
I originally worked on a direct drive design using 4500V IXYZ MOSFETS and IGBTs.
Basically it worked but these devices have such large interelectrode capacitances that I basically was driving the devices rather than the speakers!
One day I yanked out the HV MOSFETs, connected the tubes, added heater supplies and bingo! all worked as expected. Never looked back.
I also knew a guy who worked on a direct drive amp with stacked devices, but never got them to survive more than a few hours of playing time. At some point, a combi of V and I unbalanced the stack and the weakest guy blew.
With those stacks you carefully need to divide not only the supply voltage resistively but also capacitively, taking into account the differences in capactances between parts. It probably can be done, just like you can fly to the moon. But it's not something you can do on a rainy afternoon at the kitchen table.
Jan
Basically it worked but these devices have such large interelectrode capacitances that I basically was driving the devices rather than the speakers!
One day I yanked out the HV MOSFETs, connected the tubes, added heater supplies and bingo! all worked as expected. Never looked back.
I also knew a guy who worked on a direct drive amp with stacked devices, but never got them to survive more than a few hours of playing time. At some point, a combi of V and I unbalanced the stack and the weakest guy blew.
With those stacks you carefully need to divide not only the supply voltage resistively but also capacitively, taking into account the differences in capactances between parts. It probably can be done, just like you can fly to the moon. But it's not something you can do on a rainy afternoon at the kitchen table.
Jan
You're playing with fire. Someone will next come along and insist on bypassing the delay line and driving each ring with its own digitally delayed signal. And, "It will all come to tears."
But on a more serious note, thanks for this very interesting project.
All good fortune,
Chris
ps: semi-cons and KiloJolts don't mix well
But on a more serious note, thanks for this very interesting project.
All good fortune,
Chris
ps: semi-cons and KiloJolts don't mix well
Last edited:
I have worked on high voltage stuff. One fellow didn’t believe me when I suggest the problem was cause by the table the gear was placed on. To prove it was not he disconnected the high voltage probe ground strap. As he did so sparks started flying from his fingers. The sparks stopped after a bit more than half a meter of his arm jerking the strap. I believe it was only around 150,000 volts. (1,000 megohm probe, so he survived.)
Now there are devices designed to very quickly switch high voltages. But you are really in uncharted (read classified) territory for linear control of high voltages.
The folks I am familiar with use lower voltages through a transformer and capacitively couple that to a high voltage DC supply. The only uncommon part required is the step up transformer with high voltage insulation.
The transformer is made easier if it runs at a high frequency and the audio is amplitude modulating the carrier.
Also consider that the AC load is a high impedance so low uF cazapatators will work as long as they are rated for at least double the DC voltage.
ES
(When you use capacitors at high voltages they can store a lethal charge. If you short the capacitor you get a nice spark. However left open circuit many capacitors will have a high voltage grow back. So folks who do this stuff leave the capacitors always shorted when not used. This when not shorted they really are cazapatators!)
Now there are devices designed to very quickly switch high voltages. But you are really in uncharted (read classified) territory for linear control of high voltages.
The folks I am familiar with use lower voltages through a transformer and capacitively couple that to a high voltage DC supply. The only uncommon part required is the step up transformer with high voltage insulation.
The transformer is made easier if it runs at a high frequency and the audio is amplitude modulating the carrier.
Also consider that the AC load is a high impedance so low uF cazapatators will work as long as they are rated for at least double the DC voltage.
ES
(When you use capacitors at high voltages they can store a lethal charge. If you short the capacitor you get a nice spark. However left open circuit many capacitors will have a high voltage grow back. So folks who do this stuff leave the capacitors always shorted when not used. This when not shorted they really are cazapatators!)
Last edited:
https://www.diyaudio.com/community/threads/350-kw-valve-amplifier-from-1934.399411/post-7353275
No, not meant as a practical suggestion for Jan.
No, not meant as a practical suggestion for Jan.
There are more than a few ways to stack transistors to switch high voltage. Some of these will work in the sub microsecond range. Thus you could pulse width modulate the high voltage to get a reasonable high voltage AC signal.
Now if you compare the cost of any tube under consideration with a stack of transistors….
Now if you compare the cost of any tube under consideration with a stack of transistors….
Do I understand correctly that the requirement is to operate at about 3KVDC idle and maybe 15mA minimum? Easy to see why you'd want a valve choice less dramatic than transmitter valves, but I'd still wonder if anything with the anode connection in a socket could meet the goals, long term. Maybe a compactron, or anything with an anode top-cap, but transmitter valves are purpose-built for high voltages (and arrive with their own baggage - usually no separate cathode).
Will the valves' anodes be supplied from big resistors or from a split primary inductor? Doesn't matter to the choice of valves, except in the issue of providing a suitably low driving impedance to the panels.
All good fortune,
Chris
Will the valves' anodes be supplied from big resistors or from a split primary inductor? Doesn't matter to the choice of valves, except in the issue of providing a suitably low driving impedance to the panels.
All good fortune,
Chris
Would help to see the schematic
What will be the maximum potential between the elements.
For EHT you'll need a tube with an excellent vacuum to avoid gas ionization leading to arching. I' m not aware of any low/medium power tube that was designed for 6.5KV in class A, and even if such a tube existed, the vacuum may have deteriated with time in dry storage. However, getters work within a temperate range, so you may wish to consider a safe degassing rig (induction heating or low HT / high current to get the appropriate temperature) to weed out the duds and gradually de-gas the okay ones before going EHT in your chosen forced air cooled circuit. Always keep the glass seals cool.
I have some nos Eimac 4-125 tetrodes from the 1980's. They only cost me £15 each just a few years ago. By this point Eimac had ditched the dubious green uranium glass and were using the pryovac coating on the anode. Older isn't always better and I suspect Eimac peaked around this era. Sadly their glass triodes seem to be older in vintage with the green glass, but you might get lucky. Eimac made a 15E small glass triode that might be worth looking at. I wouldn't risk using a tetrode/pentode at such high voltage due to the closer proximity of the elements and the potential for screens to run away and melt.
What will be the maximum potential between the elements.
For EHT you'll need a tube with an excellent vacuum to avoid gas ionization leading to arching. I' m not aware of any low/medium power tube that was designed for 6.5KV in class A, and even if such a tube existed, the vacuum may have deteriated with time in dry storage. However, getters work within a temperate range, so you may wish to consider a safe degassing rig (induction heating or low HT / high current to get the appropriate temperature) to weed out the duds and gradually de-gas the okay ones before going EHT in your chosen forced air cooled circuit. Always keep the glass seals cool.
I have some nos Eimac 4-125 tetrodes from the 1980's. They only cost me £15 each just a few years ago. By this point Eimac had ditched the dubious green uranium glass and were using the pryovac coating on the anode. Older isn't always better and I suspect Eimac peaked around this era. Sadly their glass triodes seem to be older in vintage with the green glass, but you might get lucky. Eimac made a 15E small glass triode that might be worth looking at. I wouldn't risk using a tetrode/pentode at such high voltage due to the closer proximity of the elements and the potential for screens to run away and melt.
Uranium glass, usually looks like a yellow/green tint, was used in the parts of the glass envelope around the glass-to-metal seals, maybe a cm diameter, in the glass era valves, for its thermal expansion characteristics. Couldn't be done today, most likely.
The Eimacs and their predecessor Heintz & Kaufmann's were made for (often military) use, and are only perversely useful for modern audio*. Just by chance, they can have very nice curves in a (whacko, extreme DIY) linear amplifier context. For Jan's application the problem is to find one small enough to not be too wasteful (needs four minimum). Sweep tubes are too small, transmitting triodes are from the 1930s, modern ceramic oxide cathode transmitting valves aren't very linear, etc. No clear winner because it's a unique application.
* Perversely, the curves of some of the H&K/EM war era triodes (also the triode-connected 4E27 family below 700VDC) are amazingly linear in a modern extreme DIY context. 300B? 2A3? Second rate posers.
Way, way, OT, and only posted because I respect Jan's perspective on the world,
Chris
The Eimacs and their predecessor Heintz & Kaufmann's were made for (often military) use, and are only perversely useful for modern audio*. Just by chance, they can have very nice curves in a (whacko, extreme DIY) linear amplifier context. For Jan's application the problem is to find one small enough to not be too wasteful (needs four minimum). Sweep tubes are too small, transmitting triodes are from the 1930s, modern ceramic oxide cathode transmitting valves aren't very linear, etc. No clear winner because it's a unique application.
* Perversely, the curves of some of the H&K/EM war era triodes (also the triode-connected 4E27 family below 700VDC) are amazingly linear in a modern extreme DIY context. 300B? 2A3? Second rate posers.
Way, way, OT, and only posted because I respect Jan's perspective on the world,
Chris
