Here it comes SJDT170R1400 - SEMISOUTH - JFET, SIC, N-ON, 1700V, 3A, D2PAK | Farnell United KingdomBack on topic
Ciss almost 1000pf, that is a bit high. One of the most important parameters to look for is low gate capacitance (Ciss). You have to stack multiple mosfets to handle the voltage and to do so you need a resistor divider driving the gates. Those resistors together with Ciss limit your bandwidth. Lowering the value of those resistors makes your dissipation skyrocket.
The 1000V IR IRFBG20 is such a nice mosfet for this application because it has only 500 pf Ciss.
The BFC40 from Semelab has 550 pf and will even handle 1500V. Semelab also has the BFC60 with as little as 40 pf but that will only handle 100mA/20W and that is just not enough for a 4kV amp. Ideally would be a mosfet in between those 2 but I have never found one. Come on guys, start producing a bfc 50
The higher the current rating the higher Ciss will be. So you need to get the smallest available mosfet that will handle the power.
Don't forget zeners, just in case.
The only decent thing to decouple such potential would be decent trasformer or sigma-delta ADC/DAC with opto in between. Tube based 8kV DD monster was posted quite a time ago.
BTW 5W power supply with few pF between in-out is much easier than most people think...
I was looking for FET's today and I found a few that look pretty good.
Some are more costly and some not so bad.
In various power ranges and all available from Digikey.
Some are even Depletion mode types.
http://ixapps.ixys.com/DataSheet/DS100187C(IXTA_TH_TV02N250_S).pdf
http://ixapps.ixys.com/DataSheet/DS100214A-(IXTH_V03N400_S).pdf
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00158273.pdf
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00159991.pdf
http://ixdev.ixys.com/DataSheet/DS100185B(IXTY-TA-TP1R6N100D2).pdf
http://ixdev.ixys.com/DataSheet/98812.pdf
http://ixdev.ixys.com/DataSheet/DS98736D(IXTA-P05N100).pdf
http://ixdev.ixys.com/DataSheet/DS100182A(IXTY-TA-TP08N100D2).pdf
http://ixdev.ixys.com/DataSheet/98545.pdf
Cheers !
jer
Some are more costly and some not so bad.
In various power ranges and all available from Digikey.
Some are even Depletion mode types.
http://ixapps.ixys.com/DataSheet/DS100187C(IXTA_TH_TV02N250_S).pdf
http://ixapps.ixys.com/DataSheet/DS100214A-(IXTH_V03N400_S).pdf
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00158273.pdf
http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00159991.pdf
http://ixdev.ixys.com/DataSheet/DS100185B(IXTY-TA-TP1R6N100D2).pdf
http://ixdev.ixys.com/DataSheet/98812.pdf
http://ixdev.ixys.com/DataSheet/DS98736D(IXTA-P05N100).pdf
http://ixdev.ixys.com/DataSheet/DS100182A(IXTY-TA-TP08N100D2).pdf
http://ixdev.ixys.com/DataSheet/98545.pdf
Cheers !
jer
Last edited:
I was looking for FET's today and I found a few that look pretty good.
Some are more costly and some not so bad.
In various power ranges and all available from Digikey.
Some are even Depletion mode types.
SKIPPED
Cheers !
jer
You've missed the point - no silicon mosfet even lateral one can outperform carbide's capacitances per se, moreover jfets are believed to be better in terms of distortions introduced: Nelson amps.
As they try to penetrate a market - you can easily get some - just wait for maturity and all you'll be getting generic smps oriented parts or worse IGBTs...
Sadly IXYS staff is very often made out of unobtanium as it was mentioned before, good luck of getting some. Being small guys they do not like small guys like diyers
better figure of merit doesn't help if the only available device is 10x too large
but what current production alt semi devices are available with reasonalbe I rating for ESL - the few I see are for 10s kW power switching and the die are just too big for the job
http://www.diyaudio.com/forums/pass-labs/222098-semisouth-goes-dodo-what-now.html
but what current production alt semi devices are available with reasonalbe I rating for ESL - the few I see are for 10s kW power switching and the die are just too big for the job
http://www.diyaudio.com/forums/pass-labs/222098-semisouth-goes-dodo-what-now.html
Last edited:
All of the ones I listed are in stock at Digikey/
Most are under $3 except for the first two.
Sic Fet's would be nice but they can't be had from a normal vender and there are problems with Semisouth now.
The current demand for an ESL depends on how much capacitance it has.
My little panels of less than 50pf (35pf) would roughly require about 100watts at 5kv peak and up to 400watts peak to fully drive them up to +10Kv peak.
10kv peak is an awfully lot of drive voltage.
However, I was able to do this for a short while with my little panel.
Other than that I don't have any exact figures at the moment.
Even at 5Kv peak my panels gets in excess of 105db at 1 meter and a larger panel would do much better than that.
But the bigger the panel the more power is required of course.
I think I have found it to be around 1000watts to fully drive a 1' X 4' panel at 5Kv to 6kv rms I believe.
I forget what the D/S was that I used,I think it was .062" for that estimation.
Since ESL's have a rising response with increasing of frequency the largest voltage swing at the lowest frequency of operation is what is most important and is what well set your usable maximum SPL.
When eqaulized to be flat the drive voltage at 200hz will be -20db lower at 20Khz.
This pretty much keeps the power demand the same across the whole bandwidth for 200 to 20Khz as an example as the capacitive reactance falls at the same rate as the frequency rises.
As the voltage is halved as per octave due to the -6db per octave equalization, at the same rate, the power demand stays the same.
Even with a gate capacitance of say 650pf it would still only require less than a watt of drive power even with a 500khz sine wave considering a full 15v swing on the gate in which in most cases will not be required.
Yes, many of the FET's are optimised for switching but this doesn't mean that the can't be used for linear circuits.
I called IRF and talked to an engineer a few years ago on this and that is exactly what he had told me.
He said that all of the FET's can be used in linear circuits just some a more suited for switching circuits than others.
I tried every type of FET I had in my circuit from IRF510's to IRF840's and IRFZxxx switchers, They all worked just fine in the linear circuit.
I will be exploring this again more sometime very soon.
Right now, I almost have a new little panel finished.
The details of its construction can be found here,
A Segmented Stator Desktop ESL
It is of a segmented stator design and should have even less of capacitance and current demand on an amplifier than what a regular panel would be.
jer
Most are under $3 except for the first two.
Sic Fet's would be nice but they can't be had from a normal vender and there are problems with Semisouth now.
The current demand for an ESL depends on how much capacitance it has.
My little panels of less than 50pf (35pf) would roughly require about 100watts at 5kv peak and up to 400watts peak to fully drive them up to +10Kv peak.
10kv peak is an awfully lot of drive voltage.
However, I was able to do this for a short while with my little panel.
Other than that I don't have any exact figures at the moment.
Even at 5Kv peak my panels gets in excess of 105db at 1 meter and a larger panel would do much better than that.
But the bigger the panel the more power is required of course.
I think I have found it to be around 1000watts to fully drive a 1' X 4' panel at 5Kv to 6kv rms I believe.
I forget what the D/S was that I used,I think it was .062" for that estimation.
Since ESL's have a rising response with increasing of frequency the largest voltage swing at the lowest frequency of operation is what is most important and is what well set your usable maximum SPL.
When eqaulized to be flat the drive voltage at 200hz will be -20db lower at 20Khz.
This pretty much keeps the power demand the same across the whole bandwidth for 200 to 20Khz as an example as the capacitive reactance falls at the same rate as the frequency rises.
As the voltage is halved as per octave due to the -6db per octave equalization, at the same rate, the power demand stays the same.
Even with a gate capacitance of say 650pf it would still only require less than a watt of drive power even with a 500khz sine wave considering a full 15v swing on the gate in which in most cases will not be required.
Yes, many of the FET's are optimised for switching but this doesn't mean that the can't be used for linear circuits.
I called IRF and talked to an engineer a few years ago on this and that is exactly what he had told me.
He said that all of the FET's can be used in linear circuits just some a more suited for switching circuits than others.
I tried every type of FET I had in my circuit from IRF510's to IRF840's and IRFZxxx switchers, They all worked just fine in the linear circuit.
I will be exploring this again more sometime very soon.
Right now, I almost have a new little panel finished.
The details of its construction can be found here,
A Segmented Stator Desktop ESL
It is of a segmented stator design and should have even less of capacitance and current demand on an amplifier than what a regular panel would be.
jer
Last edited:
Here is a recent thread of similar if not the same topology that I have been messing with,
http://www.diyaudio.com/forums/solid-state/170945-8-watt-hifi-few-parts.html#post2255998
And the THD figures have been shown to be very clean.
I didn't have any way to measure THD when I built my circuits but I was able to listen to them using a Piezo driver and as well as monitor it with my scope and it was very clean with music as well as with Sine and Square waves.
I have no doubt in mind that some thing like this would work nicely for ESL's especially a smaller one at least.
This topology has been done before and is also used for ESphones as well.
Here is a design by Neil S.Mckean that is the same as well,
Do It Yourself - Electrostatic Speakers - Project: ESL H.V. Amp by Neil S. Mckean
Cheers !!
jer
http://www.diyaudio.com/forums/solid-state/170945-8-watt-hifi-few-parts.html#post2255998
And the THD figures have been shown to be very clean.
I didn't have any way to measure THD when I built my circuits but I was able to listen to them using a Piezo driver and as well as monitor it with my scope and it was very clean with music as well as with Sine and Square waves.
I have no doubt in mind that some thing like this would work nicely for ESL's especially a smaller one at least.
This topology has been done before and is also used for ESphones as well.
Here is a design by Neil S.Mckean that is the same as well,
Do It Yourself - Electrostatic Speakers - Project: ESL H.V. Amp by Neil S. Mckean
Cheers !!
jer
I just received a few HV FET's today from STMicroelectronics to start working on my direct drive amplifier design.
Here is what I have so far,
STW4N150 1500v 4A N-CH TO-247
STF6N95K5 950V 9A N-CH TO-220FP
STP21N90K5 900V 18.5A N-CH TO-220
STTH812DI ULT Fast 1.2Kv Diode
STTH512DI ULT Fast 1.2KV 5A diode
And I should have a few IRFBG30's on their way as well as various other samples.
ST also has some 1700V FET's but I didn't see them until after I put in my order so I will get some next time.
Plus I also got a few high current drive opamp's and assorted precision opamp's as well.
I already have an old Heathkit SB-101 power supply that provides a good source of filtered 900Vdc to start with and for use with my DIY ES headphones.
Then from there I will build a higher voltage switcher for bigger design's and FET stacking types of configurations.
My goal is to get between 6 to 7KV peak voltages although even 1500V will make my little desktop panel ESL comfortably loud as well.
My Desktop panel design is less than 50pf so little current will be needed, Although my end goal is to drive at least a 400pf panel to 6 to 7Kv peak voltages to start with and from there who knows.
jer
Here is what I have so far,
STW4N150 1500v 4A N-CH TO-247
STF6N95K5 950V 9A N-CH TO-220FP
STP21N90K5 900V 18.5A N-CH TO-220
STTH812DI ULT Fast 1.2Kv Diode
STTH512DI ULT Fast 1.2KV 5A diode
And I should have a few IRFBG30's on their way as well as various other samples.
ST also has some 1700V FET's but I didn't see them until after I put in my order so I will get some next time.
Plus I also got a few high current drive opamp's and assorted precision opamp's as well.
I already have an old Heathkit SB-101 power supply that provides a good source of filtered 900Vdc to start with and for use with my DIY ES headphones.
Then from there I will build a higher voltage switcher for bigger design's and FET stacking types of configurations.
My goal is to get between 6 to 7KV peak voltages although even 1500V will make my little desktop panel ESL comfortably loud as well.
My Desktop panel design is less than 50pf so little current will be needed, Although my end goal is to drive at least a 400pf panel to 6 to 7Kv peak voltages to start with and from there who knows.
jer
parasitic C of the multi-Amp power fets will be much larger than your desktop panel
you should probably look at IXYS HV Mosfets for a better match of die size to your application
for the 400pF panel 1 Apk should be fine for listening to music - even if thats not adequate for 20 KHz full amplitude sines
and before you start - could I interest you in some Life Insurance?
you should probably look at IXYS HV Mosfets for a better match of die size to your application
for the 400pF panel 1 Apk should be fine for listening to music - even if thats not adequate for 20 KHz full amplitude sines
and before you start - could I interest you in some Life Insurance?
I used a Sanders-type HV direct-drive amp for maybe 2-3 decades. With barely a hiccup. I believe that I and a "test panel" of friends are convinced it is audibly superior even to the 39 lb matching transformers in driving Dayton-Wrights and open-air D-W panels.
My amp really drive a large bunch of resistors with the panels barely counting in the load. Makes it easier to create a stable wide-band amp, rather than an amp to drive a largely capacitive load.
But with a 2400vdc B+ supply (and covering 140-3000 Hz range), I still wished for a bit more ooomph in a middle-sized living room (note, the D-W panels have quite wide spacing).
Ben
My amp really drive a large bunch of resistors with the panels barely counting in the load. Makes it easier to create a stable wide-band amp, rather than an amp to drive a largely capacitive load.
But with a 2400vdc B+ supply (and covering 140-3000 Hz range), I still wished for a bit more ooomph in a middle-sized living room (note, the D-W panels have quite wide spacing).
Ben
Last edited:
Jcx, Yes ,I understand, I have much experience when working with HV cricuits.
I have repaired tube electronics (TV's) for many years as well as many tube RF linear amps, Neon Transformers and such.
But, Yes the risk is always there, Thanks !!!
The FET's do have a bit large of input capacitance, But their output capacitance is much smaller than the inputs and yes it is much larger than my current panels.
I have some High current opamps to drive the gate loads I figured about 40ma. but the ones I have are good for 150ma and I have some that will do 3 amps as well.
According to the data sheets they make nice square waves into a .01uf capacitance load......And stable!
I do like the IXYS stuff and would be my choice but I got these as samples just to get started.
I think IXYS requires me to have a Business domain type Email address so those will have to wait until I get one or just buy them outright as the ones I wanted were only $.87 each.
It was almost more for the shipping!!!
Ben, I have all of the Sanders and David P. Hermeyer schematics!!
I have been wanting to build them for many many many years, But I never got the tubes at a cheap enough price when I should have.
And, The power resistors were quite costly as well!!!
FET's will handle this quite nicely at a fraction of the cost!
I do know that at 5KV peak the panels get to about 106db and at 2.5KV peak the are in the +99db range.
I have measured this with an SPL meter while watching the input level going to a 1:256 step-up transformer on a scope and typically anything more than 5Vpeaks (1280Vpeak) was more than enough for average everyday listening.
A ClassA P-P amp running at 2.5kv to 3kv will easily produce my main goal and my 900V supply will be enough to make some decent sound.
I have a few MOT's lying around that I could use but I do plan on using a much higher frequency supply in order to cut the cost and space of the filter section of the main supply.
This will also allow me to better regulate it as well as I have had great success with my Variable HV Bias Supply.
As far as DC output protection I have thought about making the capacitors out of either a stack of glass plates or plastic material, I am not sure which one will be the least noisy, I probably won't use them at all.
I am not sure if I will be allowed to post the circuits here or not.
But,there are many of such circuits in these threads.
I will let you know as I go along of my progress, But, I will definitely have them up on the ESLDIY site.
If I can just get one to work well on only a 1500v supply I think that I may decide that it is ample for my needs.
Even my little panel needs about 300 to 400 watts to fully drive them, But again that is about 106db to 110db at one meter so add another 6db for my nearfield listening distance!!!!
Cheers!!!!
jer
I have repaired tube electronics (TV's) for many years as well as many tube RF linear amps, Neon Transformers and such.
But, Yes the risk is always there, Thanks !!!
The FET's do have a bit large of input capacitance, But their output capacitance is much smaller than the inputs and yes it is much larger than my current panels.
I have some High current opamps to drive the gate loads I figured about 40ma. but the ones I have are good for 150ma and I have some that will do 3 amps as well.
According to the data sheets they make nice square waves into a .01uf capacitance load......And stable!
I do like the IXYS stuff and would be my choice but I got these as samples just to get started.
I think IXYS requires me to have a Business domain type Email address so those will have to wait until I get one or just buy them outright as the ones I wanted were only $.87 each.
It was almost more for the shipping!!!
Ben, I have all of the Sanders and David P. Hermeyer schematics!!
I have been wanting to build them for many many many years, But I never got the tubes at a cheap enough price when I should have.
And, The power resistors were quite costly as well!!!
FET's will handle this quite nicely at a fraction of the cost!
I do know that at 5KV peak the panels get to about 106db and at 2.5KV peak the are in the +99db range.
I have measured this with an SPL meter while watching the input level going to a 1:256 step-up transformer on a scope and typically anything more than 5Vpeaks (1280Vpeak) was more than enough for average everyday listening.
A ClassA P-P amp running at 2.5kv to 3kv will easily produce my main goal and my 900V supply will be enough to make some decent sound.
I have a few MOT's lying around that I could use but I do plan on using a much higher frequency supply in order to cut the cost and space of the filter section of the main supply.
This will also allow me to better regulate it as well as I have had great success with my Variable HV Bias Supply.
As far as DC output protection I have thought about making the capacitors out of either a stack of glass plates or plastic material, I am not sure which one will be the least noisy, I probably won't use them at all.
I am not sure if I will be allowed to post the circuits here or not.
But,there are many of such circuits in these threads.
I will let you know as I go along of my progress, But, I will definitely have them up on the ESLDIY site.
If I can just get one to work well on only a 1500v supply I think that I may decide that it is ample for my needs.
Even my little panel needs about 300 to 400 watts to fully drive them, But again that is about 106db to 110db at one meter so add another 6db for my nearfield listening distance!!!!
Cheers!!!!
jer
Last edited:
The STF6N95K5 950V 9A N-CH TO-220FP ones seem to be the best of the bunch I got right now.
Even though they are the TO-220FP fully encased version even stacking them will hardly exceed their 25 watt disapation factor in my application.
A standard TO-220 version is rated at 90watts.
It has a gate capacitance 450pf.
And the output capacitance is only 30pf !!!
The STW4N150 1500V 4A is not to bad either as I have seen worse.
Rated at 160watts with an input capacitance of 1200pf.
And an output capacitance 120pf.
It looks like stacking two 950V Fet's will yield a better overall performance for higher powers if the non Isolated version (TO-220) is used.
Using the Isolated version will be nice with low power small panels and headphones. They could be all mounted on the same heatsink as they are isolated!!
Stacking 3 for 2850V ( 6 required) is still only 50watts peak for my 50pf (160Kohm) load alone so it may be marginal and run warm but i think it will work.
If I do have any heat issues then I will either lap them down and insulate them or just get some of the plain TO-220 versions.
This will definitely be better than the wasted power that is used up in a step-up transformer at more than 600pf to 800pf per core to start with!!!
jer
Even though they are the TO-220FP fully encased version even stacking them will hardly exceed their 25 watt disapation factor in my application.
A standard TO-220 version is rated at 90watts.
It has a gate capacitance 450pf.
And the output capacitance is only 30pf !!!
The STW4N150 1500V 4A is not to bad either as I have seen worse.
Rated at 160watts with an input capacitance of 1200pf.
And an output capacitance 120pf.
It looks like stacking two 950V Fet's will yield a better overall performance for higher powers if the non Isolated version (TO-220) is used.
Using the Isolated version will be nice with low power small panels and headphones. They could be all mounted on the same heatsink as they are isolated!!
Stacking 3 for 2850V ( 6 required) is still only 50watts peak for my 50pf (160Kohm) load alone so it may be marginal and run warm but i think it will work.
If I do have any heat issues then I will either lap them down and insulate them or just get some of the plain TO-220 versions.
This will definitely be better than the wasted power that is used up in a step-up transformer at more than 600pf to 800pf per core to start with!!!
jer
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.