I have used IRFBG30/IRFBG20 in the past. The amp was class A bridged; max. output voltage about 600V rms. Suitable for panels having high efficiency only or headphones.
There are many BJT transistors with higher voltage rating but their SOA is usually worse at the same supply voltage. What topology amp you are planning to build?
Last edited:
Had a thought
Just had a crazy thought - I thought I would ideally need more volts (than 600rms) to get it working well and then thought perhaps I could bridge two amps and then thought, mmmm, I could bridge the speakers!
Could work pretty well up to 2 or 3khz ....
Have you a circuit I could borrow?
Just had a crazy thought - I thought I would ideally need more volts (than 600rms) to get it working well and then thought perhaps I could bridge two amps and then thought, mmmm, I could bridge the speakers!
Could work pretty well up to 2 or 3khz ....
Have you a circuit I could borrow?
Hi,
Looks like i have lost the complete schematic. Here is the screenshot of one half of the bridge. 600Vrms is already for a bridged amp. It has some severe shortcomings however:
1) low output voltage for most ESLs(except headphones & very sensitive units)
2) high power dissipation(constantly 40w for one bridged module) - active cooling needed
3) possibly lower safety because output voltage is not isolated from the ground.
So I have abandoned the idea some time after.
BTW why not use mains toroids ? Some of them give really excellent results and are readily available. I use 6x 230/8V units for a total stepup ratio of ~1:115.
Attachments
ESL loudspeakers can require middling kV
IXYS offers MOSFET up to 4 kV - still would need to cascade for some ESL
Class A is hugely power wasting for driving pure C load to 20 kHz when most music has <5 kHz power bandwidth
and even for ES headphones several of the "audiophile" direct drive amps do draw over 50 W continuous power
IXYS offers MOSFET up to 4 kV - still would need to cascade for some ESL
Class A is hugely power wasting for driving pure C load to 20 kHz when most music has <5 kHz power bandwidth
and even for ES headphones several of the "audiophile" direct drive amps do draw over 50 W continuous power
Last edited:
You might want a look at this idea from EDN. Each amp can hit 1800 Vp-p and in push pull as needed for ESL, 3600 V p-p !! The question is can it deliver enough current to drive that big capacitor.
High-voltage amplifier uses simplified circuit | EDN
G²
I have many models that I have done in Circuitmaker using the IRFBG30's and similar FET's.
All class A as well, Bridged of course.
I have found that you need about 5Kv to 10Kv peak voltage to make an ESL really sing depending on the size of the panel.
As the bigger the panel (more surface area) the less voltage will be required to reach a certain amount of SPL's.
At the time I started on such designs the IRFBGxx series was the highest voltage that could be found cheaply.
So, in order to get a higher voltage swing I had experimented with stacking (cascoding) the FET's.
This worked very good only I never finished one for such a high range of voltages yet.
I proved my circuit for 200v using 3 or 4 IRF510's about 4 years ago.
Some of my design are similar to Bazukaz circuit only minus the gate driver transistor and they can be found here,
http://www.diyaudio.com/forums/head...discrete-class-headphone-amp.html#post2596223
This was also before I learned about how to use a current source in place of the resistor in the top half.
This is something that I now understand and hopefully will be continuing such a project soon and I would be very interesting in seeing your results as well.
All I need is some more FET's.
There are many more higher voltage FET's to choose from now, But you want one that has the lowest gate capacitance you can get and these types will generally cost much more although I have found a few that aren't to bad cost wise.
The lower the gate capacitance the less power that will be required to drive them as by the time you get up to 20Khz and higher to faithfully reproduce a square wave the gate current gets quite high in order to have an acceptable slewrate.
Also to fully drive a 1' x 4' panel at these levels it will require on the order of about 900 of power.
Being that an ESL panel has a 6db per octave rise in response after equalization this won't be much of a problem, Just something to consider wile designing such an amplifier.
I have found that considering the current requirements and Pd of some FET's, against cost, That stacking the lower voltage ones may be a viable alternative.
Last year I found some nice 500v to 650v Fet's with a fairly high Pd for like .25$ to .89$ a piece.
I was going to buy up all of the ones for .25$
They were some discontinued parts and I never jump on it, But they may still be available.
Doing it this way will also be a larger construction technique, But having it being spread out will also make it easier to work on and help keep the HV isolated from possible flash over issues between the stages causing a catastrophic failure of the whole thing as each FET stage will need to be adjusted in order to have an equal voltage across each stage.
Then there is the cooling issue and I have thought about putting the whole thing in an oil bath as well.
I have put a lot of thought into these and I still like the idea of a large solid state direct drive amplifier.
I have a design somewhere in my archives that I was going to try using some IRFBG30's stacked up powered from a MOT that I have been saving for some time.
It should produce about 6KV p-p on a 3100v supply.
Even some thing like this may be ample for a larger panel.
But at least twice this will be required for the little panels that I have been experimenting with to get over 105db.
Then there is the issue of the filtering the supply!
A switching supply may be more suitable and cheaper as they are available and used in today's modern microwave ovens.
I found a very nice one on the web that a HAM had done up as it was quite impressive!
But is was only at 1200 volts output.
I had finished my Variable HV supply about a year ago and it is precision regulated and has enough current to pull a constant arc at 5KV.
I have been bitten by this supply as it does a peak voltage of 13.8kv and it is no joke!!
Just a few things to consider.
Cheers !!!
jer
All class A as well, Bridged of course.
I have found that you need about 5Kv to 10Kv peak voltage to make an ESL really sing depending on the size of the panel.
As the bigger the panel (more surface area) the less voltage will be required to reach a certain amount of SPL's.
At the time I started on such designs the IRFBGxx series was the highest voltage that could be found cheaply.
So, in order to get a higher voltage swing I had experimented with stacking (cascoding) the FET's.
This worked very good only I never finished one for such a high range of voltages yet.
I proved my circuit for 200v using 3 or 4 IRF510's about 4 years ago.
Some of my design are similar to Bazukaz circuit only minus the gate driver transistor and they can be found here,
http://www.diyaudio.com/forums/head...discrete-class-headphone-amp.html#post2596223
This was also before I learned about how to use a current source in place of the resistor in the top half.
This is something that I now understand and hopefully will be continuing such a project soon and I would be very interesting in seeing your results as well.
All I need is some more FET's.
There are many more higher voltage FET's to choose from now, But you want one that has the lowest gate capacitance you can get and these types will generally cost much more although I have found a few that aren't to bad cost wise.
The lower the gate capacitance the less power that will be required to drive them as by the time you get up to 20Khz and higher to faithfully reproduce a square wave the gate current gets quite high in order to have an acceptable slewrate.
Also to fully drive a 1' x 4' panel at these levels it will require on the order of about 900 of power.
Being that an ESL panel has a 6db per octave rise in response after equalization this won't be much of a problem, Just something to consider wile designing such an amplifier.
I have found that considering the current requirements and Pd of some FET's, against cost, That stacking the lower voltage ones may be a viable alternative.
Last year I found some nice 500v to 650v Fet's with a fairly high Pd for like .25$ to .89$ a piece.
I was going to buy up all of the ones for .25$
They were some discontinued parts and I never jump on it, But they may still be available.
Doing it this way will also be a larger construction technique, But having it being spread out will also make it easier to work on and help keep the HV isolated from possible flash over issues between the stages causing a catastrophic failure of the whole thing as each FET stage will need to be adjusted in order to have an equal voltage across each stage.
Then there is the cooling issue and I have thought about putting the whole thing in an oil bath as well.
I have put a lot of thought into these and I still like the idea of a large solid state direct drive amplifier.
I have a design somewhere in my archives that I was going to try using some IRFBG30's stacked up powered from a MOT that I have been saving for some time.
It should produce about 6KV p-p on a 3100v supply.
Even some thing like this may be ample for a larger panel.
But at least twice this will be required for the little panels that I have been experimenting with to get over 105db.
Then there is the issue of the filtering the supply!
A switching supply may be more suitable and cheaper as they are available and used in today's modern microwave ovens.
I found a very nice one on the web that a HAM had done up as it was quite impressive!
But is was only at 1200 volts output.
I had finished my Variable HV supply about a year ago and it is precision regulated and has enough current to pull a constant arc at 5KV.
I have been bitten by this supply as it does a peak voltage of 13.8kv and it is no joke!!
Just a few things to consider.
Cheers !!!
jer
I built a Sanders-like 8023 (??) tube amp to drive my Dayton-Wright 6-panel DIY speakers and later, to drive my off-the-shelf D-W XG8 speakers. B+ was 2400 and being direct connected, that reduced the negative bias I needed. I think the clean swing voltage was like 600vrms. BTW, DC-coupled all the way, if I recall. I drove the speakers roughly 140 Hz to 3500 Hz, with sharp crossover curves. It played plenty loud in a medium sized live room but not as loud as I wanted, now and then.
DANGEROUS.........
Best sound I ever made and after no small amount of fiddling with feedback loops, etc. the amp ran with no troubles for almost 20 years. I am convinced direct drive is audibly superior to anything using a step-up transformer. There are compelling reasons to manufacture ESLs with transformers but those reasons don't include striving for best quality sound.
It is conceivable that a half-way step-up would be a good compromise if you can't achieve the full swing you need to drive ESLs. More than 40 years ago, Mike Wright made such a tube amp coupled to his first speakers.
Remember, an amp with distortion anybody would consider weakish by today's standards is a perfect match for even great ESL speakers.
DANGEROUS.........
Best sound I ever made and after no small amount of fiddling with feedback loops, etc. the amp ran with no troubles for almost 20 years. I am convinced direct drive is audibly superior to anything using a step-up transformer. There are compelling reasons to manufacture ESLs with transformers but those reasons don't include striving for best quality sound.
It is conceivable that a half-way step-up would be a good compromise if you can't achieve the full swing you need to drive ESLs. More than 40 years ago, Mike Wright made such a tube amp coupled to his first speakers.
Remember, an amp with distortion anybody would consider weakish by today's standards is a perfect match for even great ESL speakers.
Last edited:
been there, done that... All the way up to that 16 mosfet cascaded output stage running from a 4000V power supply, putting out more that 7.5 kV.
The irfbg20 and BFC40 were the best options available at the time I worked on this (couple of years ago). Could be there are better alternatives available now. You need high voltage and low gate capacity, approx 2A is ideal.
Take into account that for fullrange you need approx. 25 mA per kV, for a decent output you need at least 8kV voltage swing (4 kV power supply and bridging). That requires a current capacity of 200mA ruling out class A (2*.2*4000 = 1600W dissipation at least, per channel!). You have only N-channel fets available so class B is almost impossible to construct in such a way that it is reliable and low-distortion.
In short: Don't do it. IMHO best option is an impedance compensated transformer in a feedback loop.
I summed it all up in this topic:
http://www.diyaudio.com/forums/planars-exotics/80714-another-direct-drive-thread.html
The irfbg20 and BFC40 were the best options available at the time I worked on this (couple of years ago). Could be there are better alternatives available now. You need high voltage and low gate capacity, approx 2A is ideal.
Take into account that for fullrange you need approx. 25 mA per kV, for a decent output you need at least 8kV voltage swing (4 kV power supply and bridging). That requires a current capacity of 200mA ruling out class A (2*.2*4000 = 1600W dissipation at least, per channel!). You have only N-channel fets available so class B is almost impossible to construct in such a way that it is reliable and low-distortion.
In short: Don't do it. IMHO best option is an impedance compensated transformer in a feedback loop.
I summed it all up in this topic:
http://www.diyaudio.com/forums/planars-exotics/80714-another-direct-drive-thread.html
Last edited:
The Acoustat is a very compromised design. Nowhere near enough output current. Not very good sounding either.
Forget spice for high voltage mosfet applications. Reliability is zero. The internal capacities of MOSFETs are voltage (and even current) dependent. At these high voltages this effect is very pronounced. The models used for MOSFETS in spice do not account for that. My experience is that the real world and spice are truly two very different things
Forget spice for high voltage mosfet applications. Reliability is zero. The internal capacities of MOSFETs are voltage (and even current) dependent. At these high voltages this effect is very pronounced. The models used for MOSFETS in spice do not account for that. My experience is that the real world and spice are truly two very different things
The problem is that you have to drive a capacitor. At low frequencies you need many kV's swing to get decent output. At high and midrange frequencies you need hundreds of mA current. Combining the whole spectrum in one amplifier means it has to provide both, making it impossible to use any class A (or single ended) design.
So the challenge of building a good direct drive amplifier really comes down to how to build a push pull output stage using only one polarity of devices that see the full voltage. So far I have not found a satisfactory solution for this without using a transformer. Very interested in how you get along... So good luck and keep us posted!
So the challenge of building a good direct drive amplifier really comes down to how to build a push pull output stage using only one polarity of devices that see the full voltage. So far I have not found a satisfactory solution for this without using a transformer. Very interested in how you get along... So good luck and keep us posted!
Ummm, drive a capacitor? All you need is to run a big clean voltage. I drove a large bunch of power resistors push-pull which had an ESL speaker (not too big as capacitors go) in parallel.
Ben
I'm not sophisticated in these things, but I thought that was what you always wanted to do - drive a well-behaved resistor.
Ben
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.