Hi,
the wrong way to run an ESL is to run it by current.
Thats why any tube concept fails as long there is a stepup transformer involved. In order to magnetize the coils you need current and tubes are not capable to deliver current.
So if you use a stepup, a transistor amp is the better choice. But most people use amps with low supply voltage ranges (e.g. +/- 30 Volts). If you want to transform those values up to kilovolts, you need high stepup ratio. The performance of a stepup transformer suffers by a square factor with increasing stepup ratio. That means, that a ratio of 1:120 instead of 1:60, will increase the lower bandwith, decrease the upper bandwith and more important increase the phase shift by factor 4 !!!
I build myself a transistor amp using +/- 120 Volts powersupply. The only expensive parts are the capacitors in such voltage range.
Having such an amp you can reduce stepup ratio down to 1:30, which reduces the negative impact of the stepup drastically.
Capaciti
the wrong way to run an ESL is to run it by current.
Thats why any tube concept fails as long there is a stepup transformer involved. In order to magnetize the coils you need current and tubes are not capable to deliver current.
So if you use a stepup, a transistor amp is the better choice. But most people use amps with low supply voltage ranges (e.g. +/- 30 Volts). If you want to transform those values up to kilovolts, you need high stepup ratio. The performance of a stepup transformer suffers by a square factor with increasing stepup ratio. That means, that a ratio of 1:120 instead of 1:60, will increase the lower bandwith, decrease the upper bandwith and more important increase the phase shift by factor 4 !!!
I build myself a transistor amp using +/- 120 Volts powersupply. The only expensive parts are the capacitors in such voltage range.
Having such an amp you can reduce stepup ratio down to 1:30, which reduces the negative impact of the stepup drastically.
Capaciti
Hi,
or take a KR 845 with simple resistive load and drive it up to 2kV anode voltage
We had a thread here about such low-U trannies driven by rather conventional high-voltage amps. The problem remained that such a tranny must be wound on specification, because there´s no commercial offer and the driving amp should be at minimum 100V capable. One should also keep in mind that a amp with supply voltages of op to 400V-500V (like lots of tube amps) need to be designed for higher impedances. This means it must be able to put out voltages of a couple of hundreds of V. It doesn´t need to put out so large currents though and the output impedance may be higher, so tubes would be fine for this job or HV-MOSFETs. The advantages of a HV-amp plus Tranny would be galvanic isolation (output signal would be free of DC), cheaper standard parts and ´conventional´ circuitry.
Still though there´s a (costly) coupling element between amp and ESL with all it´s restrictions and limits.
jauu
Calvin
or take a KR 845 with simple resistive load and drive it up to 2kV anode voltage
We had a thread here about such low-U trannies driven by rather conventional high-voltage amps. The problem remained that such a tranny must be wound on specification, because there´s no commercial offer and the driving amp should be at minimum 100V capable. One should also keep in mind that a amp with supply voltages of op to 400V-500V (like lots of tube amps) need to be designed for higher impedances. This means it must be able to put out voltages of a couple of hundreds of V. It doesn´t need to put out so large currents though and the output impedance may be higher, so tubes would be fine for this job or HV-MOSFETs. The advantages of a HV-amp plus Tranny would be galvanic isolation (output signal would be free of DC), cheaper standard parts and ´conventional´ circuitry.
Still though there´s a (costly) coupling element between amp and ESL with all it´s restrictions and limits.
jauu
Calvin
I looked into the SRPP output stage which has the advantage that you can optimize it for the capacitive load, or it says on John Broskie's site. This is a rather appealing approach because of the extend current capability of the circuit. I prefer tubes over semiconductors. The other logical approach is a conventional push-pull output stage but have it floating as in the circuits posted in this thread. Instead of choke load this task could be performed by the low ratio step-up transformer for the bass.
We have a pretty good transformer winder in the Netherlands who is willing to do single pieces and experiment and has plenty experience with high voltage and audio etc. The only downside is long waiting times (couple of months usually). Automatic Electric
We have a pretty good transformer winder in the Netherlands who is willing to do single pieces and experiment and has plenty experience with high voltage and audio etc. The only downside is long waiting times (couple of months usually). Automatic Electric
We are getting into something interesting now.
The output transformers can be easily made here in Thailand for any specifications you want. Normally I can get mine made within 1-2 weeks. How about we start looking for a suitable circuit? What circuit do you have in mind, Arend-jan?
Wachara C.
The output transformers can be easily made here in Thailand for any specifications you want. Normally I can get mine made within 1-2 weeks. How about we start looking for a suitable circuit? What circuit do you have in mind, Arend-jan?
Wachara C.
Hi,
SRPP´s features sound seductive at first glance. High gain, high output current capability, low output impedance, very low parts count and low distortion without global feedback. But as it is usually implemented -one resisitor spanning between anode of the lower and cathode of the upper tube- it works correctly only into one specific resistive load. Off of the optimu, value it ceases beeing a symmetrical pushpull stage. Now practise isn´t as unforgiving as theory, so within a certain range of load resistor values the SRPP will function somehow sufficiently. But this doesn´t hold true for such an impedance variation a capacitor represents over the complete audio range. So all commercial amps that use such a simple stage rely on massive global feedback to linearize the amp (the beauty of a properly implemented SRPP is that You could omit with global feedback). You know that inherently nonlinear stages corrected by massive global feedback are not what our ears like to listen to.
The resistor between the tubes could be replaced by more complex circuitry to allow for wide bandwidth drive into a complex load like a cap/ESL as J.Brosky suggested. I haven´t seen this implemented yet.
As amp circuitry I´d suggest a circlotron (sometimes implemented as OTL-amp). The low output impedance would help solve some problems with the tranny. The elevated impedance level of a low-U tranny design works in favour with output tubes. ´Safe´ for the load even when an output tube fails. It works stable into complex loads. The cons are :You´d need two floating powersupplies and a driver stage that can supply the full output voltage range to the output tubes.
jauu
Calvin
SRPP´s features sound seductive at first glance. High gain, high output current capability, low output impedance, very low parts count and low distortion without global feedback. But as it is usually implemented -one resisitor spanning between anode of the lower and cathode of the upper tube- it works correctly only into one specific resistive load. Off of the optimu, value it ceases beeing a symmetrical pushpull stage. Now practise isn´t as unforgiving as theory, so within a certain range of load resistor values the SRPP will function somehow sufficiently. But this doesn´t hold true for such an impedance variation a capacitor represents over the complete audio range. So all commercial amps that use such a simple stage rely on massive global feedback to linearize the amp (the beauty of a properly implemented SRPP is that You could omit with global feedback). You know that inherently nonlinear stages corrected by massive global feedback are not what our ears like to listen to.
The resistor between the tubes could be replaced by more complex circuitry to allow for wide bandwidth drive into a complex load like a cap/ESL as J.Brosky suggested. I haven´t seen this implemented yet.
As amp circuitry I´d suggest a circlotron (sometimes implemented as OTL-amp). The low output impedance would help solve some problems with the tranny. The elevated impedance level of a low-U tranny design works in favour with output tubes. ´Safe´ for the load even when an output tube fails. It works stable into complex loads. The cons are :You´d need two floating powersupplies and a driver stage that can supply the full output voltage range to the output tubes.
jauu
Calvin
Just thought I would mention, I run Acoustat's with the Direct Drive Acoustat tube amps.
The amps have had slight upgrades. I have been running them for over a year with no problems at all, and on average they are run 6-8 hrs a day.
I have had 1 tube failure, which was strange because it happened when I was shutting down my system. I am using 6GE5 tubes as a sub for the 6HB5 tube. From what I have found from other owners, is that the brand of the tube ( Sylvania 6HB5) tubes are the ones that are a problem.
The amps have had slight upgrades. I have been running them for over a year with no problems at all, and on average they are run 6-8 hrs a day.
I have had 1 tube failure, which was strange because it happened when I was shutting down my system. I am using 6GE5 tubes as a sub for the 6HB5 tube. From what I have found from other owners, is that the brand of the tube ( Sylvania 6HB5) tubes are the ones that are a problem.
Hi,
I don´t know what prob Mr. v.d.Veen found, but to my taste it is just such an amp I described above. SRPP, very low bias current and ~80dB of openloop gain. Most of which is used up by the massive global feedback that is needed to make the thing roughly do what it is intended to.
It is a very clever design regarding power efficiency and basically a very low part count amplifier. Afaik the original circuit (Schmitt, later called SRPP) was first used by Beveridge commercially to drive ESLs (patent 3,773,976 and 3,668,335). Beveridge used a feedback loop that only encompassed the tubes of the output stage.
Mr. Strickland added a transistorized difference amplifier stage ahead of the tubes and included those in the feedback loop (patent 4,324,950). The additional gain of the transistor stage was used to raise the value of feedback.
The AudioExclusiv amplifier later used a very similar approach to Mr. Strickland´s amp (6HB5 output tubes but different transistor pairs) and the Shackmann amplifier (as Tweeter amp from 400Hz upwards) used a pair of 6BQ5/EL84 Output tubes and a 12AX7/ECC83 tubed input stage (feedback coupled to the cathode of the input tube).
I´d prefer more linear stages (already under open loop conditions) even if that means to ´waste´ more power. Every of the well known audio amp-designers strives to improve circuit behaviour under open loop conditions for good reasoning. Circuits that ´mask´ inherent misbehaviour with excessive global feedback don´t sound well.
jauu
Calvin
I don´t know what prob Mr. v.d.Veen found, but to my taste it is just such an amp I described above. SRPP, very low bias current and ~80dB of openloop gain. Most of which is used up by the massive global feedback that is needed to make the thing roughly do what it is intended to.
It is a very clever design regarding power efficiency and basically a very low part count amplifier. Afaik the original circuit (Schmitt, later called SRPP) was first used by Beveridge commercially to drive ESLs (patent 3,773,976 and 3,668,335). Beveridge used a feedback loop that only encompassed the tubes of the output stage.
Mr. Strickland added a transistorized difference amplifier stage ahead of the tubes and included those in the feedback loop (patent 4,324,950). The additional gain of the transistor stage was used to raise the value of feedback.
The AudioExclusiv amplifier later used a very similar approach to Mr. Strickland´s amp (6HB5 output tubes but different transistor pairs) and the Shackmann amplifier (as Tweeter amp from 400Hz upwards) used a pair of 6BQ5/EL84 Output tubes and a 12AX7/ECC83 tubed input stage (feedback coupled to the cathode of the input tube).
I´d prefer more linear stages (already under open loop conditions) even if that means to ´waste´ more power. Every of the well known audio amp-designers strives to improve circuit behaviour under open loop conditions for good reasoning. Circuits that ´mask´ inherent misbehaviour with excessive global feedback don´t sound well.
jauu
Calvin
Last edited:
Hi,
well each of the two stators has its own amplifier, both beeing equal. They get out of phase input signals, so their outputs are similarly out of phase. The load is connected between the outputs and as such constitutes a pushpull-drive. The single amplifier itself can be of SE or pushpull type. When we talk of pushpull amps it is usually meant that there is current symmetry between upper and lower output device. This is not the case with the Acoustat amp. SRPP can only be symmetrical into one fixed load impedance, which an ESL most definitely isn´t.
jauu
Calvin
well each of the two stators has its own amplifier, both beeing equal. They get out of phase input signals, so their outputs are similarly out of phase. The load is connected between the outputs and as such constitutes a pushpull-drive. The single amplifier itself can be of SE or pushpull type. When we talk of pushpull amps it is usually meant that there is current symmetry between upper and lower output device. This is not the case with the Acoustat amp. SRPP can only be symmetrical into one fixed load impedance, which an ESL most definitely isn´t.
jauu
Calvin
What about using rather more exotic valves? There are both forced air cooled and water cooled types available for use in transmitters that can work happily with 10-20KV or more of anode supply..... a pair of these could produce several KW of output power!
Ok it would be a brave man that tries to tame those voltages... and the PSU could be "fun".... then there is the heater supplies of maybe 500W per valve... but it should solve all the problems of high voltage drive
I'll get my coat......
Ok it would be a brave man that tries to tame those voltages... and the PSU could be "fun".... then there is the heater supplies of maybe 500W per valve... but it should solve all the problems of high voltage drive
I'll get my coat......
Hi,
I think that it would work, but it´ll be a hellot of effort and tha amp wold be a dangerous beast.
Lately some sand manufacturers came up with some very intersting new devices. IXYS introduced some HV-IGBTs. The first were the IXEL40N400 4KV/40A and the IXLF19N250 2.5kV/32A. Now there are severel series starting at 2.5kV with the IXGFxxN250-series, followed by 3kV with the IXGFxxN300-series and last the 4kV-series with the IXGFxxN400-series. Series and parallel connection of devices is quite simple to gain even higher ratings.
Alternatively YXIS offers MOSFETs under the label BIMOS. For example the IXBH2N250 a 2.5kV/2A device with remarkable low capacitance values.
The linear MOSFETs, designed for linear applications like current regultors are up to 1.5kV (IXTN8L150L).
YXIS manufactures a range of N-depletion mode MOSFETs which could be used as active CCS loads (IXCP10M90S, IXTH20N50, IXTH10N100D). At last they manufacture some P-channel MOSFETs up to -600V (IXTH10P60).
These devices allow for new amp designs, with less part number count circuitry that was not possible or very difficult to implement.
They even supply Spice data for some of the parts.
Other very interesting devices could be the upcoming SIC power JFETs.
For example the 1200V enhancement mode or ´normally-off JFET (enhancement or normally-off sounds weird with JFETs but they call it so) of Semisouth (SJEP120R063).
Another example are the 1200V depletion mode or normally-on JFETs of Siced.
It can be expected that the JFETs hit the market soon and that higher voltage rated types will follow.
These JFETs could then easily replace HV-Triodes. At least simulations run very well already ;-)
jauu
Calvin
I think that it would work, but it´ll be a hellot of effort and tha amp wold be a dangerous beast.
Lately some sand manufacturers came up with some very intersting new devices. IXYS introduced some HV-IGBTs. The first were the IXEL40N400 4KV/40A and the IXLF19N250 2.5kV/32A. Now there are severel series starting at 2.5kV with the IXGFxxN250-series, followed by 3kV with the IXGFxxN300-series and last the 4kV-series with the IXGFxxN400-series. Series and parallel connection of devices is quite simple to gain even higher ratings.
Alternatively YXIS offers MOSFETs under the label BIMOS. For example the IXBH2N250 a 2.5kV/2A device with remarkable low capacitance values.
The linear MOSFETs, designed for linear applications like current regultors are up to 1.5kV (IXTN8L150L).
YXIS manufactures a range of N-depletion mode MOSFETs which could be used as active CCS loads (IXCP10M90S, IXTH20N50, IXTH10N100D). At last they manufacture some P-channel MOSFETs up to -600V (IXTH10P60).
These devices allow for new amp designs, with less part number count circuitry that was not possible or very difficult to implement.
They even supply Spice data for some of the parts.
Other very interesting devices could be the upcoming SIC power JFETs.
For example the 1200V enhancement mode or ´normally-off JFET (enhancement or normally-off sounds weird with JFETs but they call it so) of Semisouth (SJEP120R063).
Another example are the 1200V depletion mode or normally-on JFETs of Siced.
It can be expected that the JFETs hit the market soon and that higher voltage rated types will follow.
These JFETs could then easily replace HV-Triodes. At least simulations run very well already ;-)
jauu
Calvin
I do agree on new IXYS devices.
If I were to use those JFET I would buy lifelong stock of them.
One can still find soviet made SIT (static induction transistors) similar to those of Toshiba. As well as BSIT with threshold voltage equal to zero.
Low voltage are still in production 20V 500mA.
At the moment they were better than MOSFETs, justifing their existance, but were really awkward to drive, especially normally open ones.
Same will happen to SIC jfet - from what I know (from the manufacturers).
I've been told that as soon as they can produce REAL MOSFET Jfets will be dropped.
Sincerely
Hi,
there has been a japanese study that claims, that MOSFET or IGBT that could be replaced by a SIC-device in switching appliances, could reduce power losses by 30%. If that holds true, the SIC would be THE future power devices.
But who knows what development MOSFETs will take?
Anyway, do You have any info as to why the JFETs are awkward to drive??
Calvin
there has been a japanese study that claims, that MOSFET or IGBT that could be replaced by a SIC-device in switching appliances, could reduce power losses by 30%. If that holds true, the SIC would be THE future power devices.
But who knows what development MOSFETs will take?
Anyway, do You have any info as to why the JFETs are awkward to drive??
Calvin
Power is ON as well as ALL switches in your converter
Power electronics engineer's point of view.
Knowing that medium voltage (up to 600v) converter cell usually consists of
two series connected switches, namely half-bridge, how would you prevent initial shoot-through
(another name for short circuit)? So comes additional POWER start-up circuitry,
so the control could be applied beforehand.
If that is not awkward than what is?
I did work with those devices.
Normally OFF devices behave like BJTs - here comes minority carriers - as soon as the gate junction becomes forward biased...
then desaturation circuitry comes then reasonably high gate (base) current - reasonably sized transformers, due to "high" power in driver circuit...
Back to thread - those normally ON JFETs are really fast, rugged and can be driven as "normal tube" - no awkwardness here then...
In regard to IXYS devices - what about SOA and would not one need high speed overcurrent protection.
And the very last - fet has penthode-like characteristics, not triode ones.
Sincerely
Alex
BTW There were a lot of non-Japanese studies of higher efficiency, especially at higher frequencies and high blocking voltage areas, (MIT, CREE etc.)
I do apologize for off-topic...
Power electronics engineer's point of view.
Knowing that medium voltage (up to 600v) converter cell usually consists of
two series connected switches, namely half-bridge, how would you prevent initial shoot-through
(another name for short circuit)? So comes additional POWER start-up circuitry,
so the control could be applied beforehand.
If that is not awkward than what is?
I did work with those devices.
Normally OFF devices behave like BJTs - here comes minority carriers - as soon as the gate junction becomes forward biased...
then desaturation circuitry comes then reasonably high gate (base) current - reasonably sized transformers, due to "high" power in driver circuit...
Back to thread - those normally ON JFETs are really fast, rugged and can be driven as "normal tube" - no awkwardness here then...
In regard to IXYS devices - what about SOA and would not one need high speed overcurrent protection.
And the very last - fet has penthode-like characteristics, not triode ones.
Sincerely
Alex
BTW There were a lot of non-Japanese studies of higher efficiency, especially at higher frequencies and high blocking voltage areas, (MIT, CREE etc.)
Hi,
I´m not going to argue with You about switching devices since this is not my field of expertise I just cited that in that report SIC-based technology was claimed as very promising. And some big names in device business are introducing such parts to the market (infineon for example, SIC-diodes, SICED is a daugther company of infineon).
For our usage -linear amplification- shoot through represents no issue and there are many audio enthusiasts who dare to get their fingers on some good power JFETs. So lets hope that their tiny light is not immideately blown out by the next generation of MOSFETs.
Calvin
I´m not going to argue with You about switching devices since this is not my field of expertise
I just cited that in that report SIC-based technology was claimed as very promising. And some big names in device business are introducing such parts to the market (infineon for example, SIC-diodes, SICED is a daugther company of infineon).For our usage -linear amplification- shoot through represents no issue and there are many audio enthusiasts who dare to get their fingers on some good power JFETs.
So lets hope that their tiny light is not immideately blown out by the next generation of MOSFETs. Calvin
Sure,
nevertheless I do have quite a few SITs and BSITs.
Silicon ones, up to 550V/15A
If you need any info or curves - please do not hesistate to ask.
I've got HV tubes and trying to figure out how to drive the upper
tube in bridge configuration... the best idea so far is the interstage transformer.
Got some toriods - measuring BH curves...
Alex
nevertheless I do have quite a few SITs and BSITs.
Silicon ones, up to 550V/15A
If you need any info or curves - please do not hesistate to ask.
I've got HV tubes and trying to figure out how to drive the upper
tube in bridge configuration... the best idea so far is the interstage transformer.
Got some toriods - measuring BH curves...
Alex