Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 2nd December 2011, 07:56 PM   #1
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Default Self-cascoding, a tool for creating ultra high voltage building blocks:

For jaenker, a sneak preview:

Self-cascoding allows you to build arbitrary blocks of UHV composite transistors.
Here is an example of SRPP having a 4KVpp+ output voltage capability.

The stack can be expanded arbitrarily to suit the needs.

This is the principle only, actual circuit has to include safety systems for balancing, protection, etc.

That's theory. Not something for beginners.
Attached Images
File Type: jpg UHVsrpp.jpg (170.2 KB, 274 views)
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 2nd December 2011, 08:54 PM   #2
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
I would call that cascade or series connection

cascode fixes the V across one Q - the cascade divides the V change equally

small C in the gate drive divider string are needed for high frequency division accuracy - otherwise the bottom Q can be cut off and see the whole V before the upper Q which are driven thru the divider R
  Reply With Quote
Old 7th December 2011, 01:05 AM   #3
diyAudio Member
 
Sch3mat1c's Avatar
 
Join Date: Jan 2003
Location: Milwaukee, WI
Send a message via ICQ to Sch3mat1c Send a message via AIM to Sch3mat1c
Yuck, that looks terrible. Tug on the bottom and the top transistors all avalanche in order! At the very least, you need equalizing capacitors along the resistor chain, which is an interesting challenge as Cgs and Cgd vary with voltage. Tall stacks exacerbate the problem, as the bottom transistor (in this example) serves as a common-source amp, while the subsequent stages act like source followers, setting Vds for the previous transistors, except for Cgs, which will tend to yank on the resistors unevenly as the change propagates. When the wave gets to the top, it finally starts pulling itself down, now contracting from the top rather than the bottom. After a very long time, the resistors settle out the charge and things reach equilibrium (~100us?).

Now, the cap divider action means the voltage response won't be as terrible as my hyperbolizing suggests, and the dampening of the resistors will tend to make it behave as an RC transmission line, rather than a wave traveling along an LC section (or worse yet, an LC section *with gain*). It will work fine at "DC" (which is basically audio frequencies) if that's all you're doing, but that's missing the point, DC is dividers and bias resistors, it isn't where the excitement is at.

I'd be very interested to see the square wave response, say replace V1 with a PULSE source, -4V initial, +4V pulsed, 10ns rise/fall time, 50us pulse width, 100us period, along with a plot of the voltage distribution along the stack. If it turns out more uniform than I thought, I might have an extra pair of shoes I can sautee.

Strange, you've got a stack of six 600V devices, if I'm reading it correctly? And a supply of 5000V (of which you're using about 4650V peak in this example)? Do check that LT put in the correct models, those should NOT be handling 775V. For instance, if the models are MODELs only, not SUBCKTs, they will not have representative capacitances, breakdown voltage or reverse capacity. Beware the hazards of trusting someone else's SPICE models!

The last 800V ST FET I played with broke down right on target, something like 870V.

Tim
__________________
Seven Transistor Labs
Projects and Resources
  Reply With Quote
Old 7th December 2011, 05:19 PM   #4
Elvee is offline Elvee  Belgium
diyAudio Member
 
Elvee's Avatar
 
Join Date: Sep 2006
Quote:
Originally Posted by Sch3mat1c View Post
Yuck, that looks terrible. Tug on the bottom and the top transistors all avalanche in order! At the very least, you need equalizing capacitors along the resistor chain, which is an interesting challenge as Cgs and Cgd vary with voltage. Tall stacks exacerbate the problem, as the bottom transistor (in this example) serves as a common-source amp, while the subsequent stages act like source followers, setting Vds for the previous transistors, except for Cgs, which will tend to yank on the resistors unevenly as the change propagates. When the wave gets to the top, it finally starts pulling itself down, now contracting from the top rather than the bottom. After a very long time, the resistors settle out the charge and things reach equilibrium (~100us?).

Now, the cap divider action means the voltage response won't be as terrible as my hyperbolizing suggests, and the dampening of the resistors will tend to make it behave as an RC transmission line, rather than a wave traveling along an LC section (or worse yet, an LC section *with gain*). It will work fine at "DC" (which is basically audio frequencies) if that's all you're doing, but that's missing the point, DC is dividers and bias resistors, it isn't where the excitement is at.

I'd be very interested to see the square wave response, say replace V1 with a PULSE source, -4V initial, +4V pulsed, 10ns rise/fall time, 50us pulse width, 100us period, along with a plot of the voltage distribution along the stack. If it turns out more uniform than I thought, I might have an extra pair of shoes I can sautee.

Strange, you've got a stack of six 600V devices, if I'm reading it correctly? And a supply of 5000V (of which you're using about 4650V peak in this example)? Do check that LT put in the correct models, those should NOT be handling 775V. For instance, if the models are MODELs only, not SUBCKTs, they will not have representative capacitances, breakdown voltage or reverse capacity. Beware the hazards of trusting someone else's SPICE models!

The last 800V ST FET I played with broke down right on target, something like 870V.

Tim
You probably missed the warning message:

Quote:
This is the principle only, actual circuit has to include safety systems for balancing, protection, etc.

That's theory. Not something for beginners.
This illustrates the principle, with the models available in LTspice's native libraries.

The actual circuit has to include balancing, protection, slew-rate control, etc (and obviously devices having very good safety margins for voltage, current, power and SOA)

In fact, the actual circuits have more protection components than "functional" ones, but the way they are implemented is highly dependent on the application: the situation for a HV ballast is not that of an amplifier.

Anyway, here is the sim you asked for, first completely raw, then with a single 1.5pF capacitor across the 2Meg feedback resistor:
Attached Images
File Type: jpg SelfCasc2.jpg (172.5 KB, 142 views)
File Type: jpg SelfCasc3.jpg (176.0 KB, 126 views)
File Type: jpg 100_5112.JPG (310.8 KB, 128 views)
__________________
. .Circlophone your life !!!! . .
♫♪ My little cheap Circlophone© ♫♪
  Reply With Quote
Old 7th December 2011, 06:19 PM   #5
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
I have used this type of configuration successfully although I have not ventured into the kilovolt range yet.

Here is a link were I show my designs,

All Discrete Class-A Headphone Amp

The first two I have built ,the third one shows how the added capacitor should be configured for each level and this shows a much better performance in the sims.

The rest of them are some hotrodded versions of some Aleph amp configurations or Class A Push-Pull For driving some ESL panels.

The 50v version was designed for a 1 ohm load and is a monster for a Class A amp,Although I wouldn't suggest using the IRF530's for the output devices,IRF240's or something would be a better choice.

I have used the second circuit in a push-pull design driving a little piezo waffer and it worked really good and sounded very clean for what it was.
I thought the little disk was going to shatter but it didn't.

That was as far as I got and I plan to do some more work like this soon.

I also have several versions using a 1000v IRF BG30's drawn up, as Jameco had a stockpile of these for less than $2 a peice.
I am not sure if they have anymore but they might as last time I checked they did.

I was going to use a couple stacks (4 X 4,16 devices) powered off of a MOT for a DD Esl amp.
This was before I had figured out how to design a CCS as the power resistors required for the thing are quite costly for a 900 watt power dissapation for each leg (X4).

The 1000v version sims with no errors as well as the others.

Great Subject !!!

jer

Last edited by geraldfryjr; 7th December 2011 at 06:24 PM.
  Reply With Quote
Old 7th December 2011, 06:29 PM   #6
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
a Leach amp used 2 series devices sucessfully

the Koss E-90 electrostatic headphone amp uses 3x series 450 V TO-92 devcies (and 2x parallel) for each side of a complementary push-pull output running from +/-600 V

I anticipate using 2x IXYS series cascaded 1 kV Mosfets for a project but may just buy the higher V device which came out later - there are now several multi-kV MOSFET and SiC FET devcies that would cut the number required in series
  Reply With Quote
Old 7th December 2011, 06:35 PM   #7
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Yes, Those Higher voltage devices just surfaced in the last two or three years.
When I started messing with this concept 800v to 1000v Fets were all that I could find. Also it was hard to find them with a very low gate capacitance as well.

jer
  Reply With Quote
Old 8th December 2011, 12:00 AM   #8
diyAudio Member
 
Sch3mat1c's Avatar
 
Join Date: Jan 2003
Location: Milwaukee, WI
Send a message via ICQ to Sch3mat1c Send a message via AIM to Sch3mat1c
Quote:
Originally Posted by Elvee View Post
You probably missed the warning message:


This illustrates the principle, with the models available in LTspice's native libraries.
Oops, fair enough..

Quote:
The actual circuit has to include balancing, protection, slew-rate control, etc (and obviously devices having very good safety margins for voltage, current, power and SOA)

In fact, the actual circuits have more protection components than "functional" ones, but the way they are implemented is highly dependent on the application: the situation for a HV ballast is not that of an amplifier.

Anyway, here is the sim you asked for, first completely raw, then with a single 1.5pF capacitor across the 2Meg feedback resistor:
Dang, seems a little high on the voltage, like it's close to or in clipping. Waveforms don't look too bad but there might be some interesting transient behavior going on. That bounce on the first waveform was interesting!

Tim
__________________
Seven Transistor Labs
Projects and Resources
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Safety Practices, General and Ultra-High Voltage DrDeville Tubes / Valves 296 11th October 2013 12:42 AM
Building the ultra light boombox voyage Class D 7 14th June 2011 01:42 PM
New high end audio translation tool at Bablefish jackinnj Everything Else 21 8th December 2004 10:12 PM
Safety Practices, General and Ultra-High Voltage DrDeville Everything Else 49 23rd September 2004 09:31 AM


New To Site? Need Help?

All times are GMT. The time now is 08:01 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2