Warning: Creating default object from empty value in /homepages/11/d132647312/htdocs/Amfone/mkportal/include/SMF/smf_out.php on line 47
Stacking MOSFETs for high voltage




 
The AM Forum
May 22, 2019, 11:23:35 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: [1]   Go Down
  Print  
Author Topic: Stacking MOSFETs for high voltage  (Read 436 times)
0 Members and 1 Guest are viewing this topic.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 7157



WWW
« on: March 17, 2019, 02:32:19 AM »

I was inspired to do some learning by the audio driver project recently discussed.
"Board for WA1GFZ MOSFET Audio Driver"
http://amfone.net/Amforum/index.php?topic=44619.0


It's interesting to learn about analog/linear use of MOSFETs at high voltages and moderate currents. What I have been thinking about is not for grid drive but more to appreciate what can be accomplished with higher voltages than usual. The bottom line is that with the FET, only a few parts are needed compared to bipolar transistors. Just like good ole tubes. The project inspired me to try a few things in LTspice.

Messed around with slightly different FETs, resistor values, etc. cranked the virtual supplies to +/-500V and changed biasing to suit the operating points, and put together in there a series arrangement of MOSFETs for the higher voltage. At the higher supply voltages, LTspice showed 400V peak to peak voltage @1A peak 'grid' current on the positive half cycle. I make no claims about it in real life and high current would mean large heatsinks. I can only offer caveats at 1KV with transistors.

Just for fun here is the series MOSFET follower circuit that looks good for higher voltage than one device alone could handle with a safety margin.

When substituted for one output device in the +/-500V simulation based on the driver article, it performed like the single part, with margin to spare for the higher voltage. The drive from one side of the phase splitter was just over 400V p-p, so that's all this was simulated with today, but it should be able to get closer to the rails if driven harder, carefully biased, and everything is perfect, which it never is.

This has little to do with the original product's purpose, but might be useful where higher dynamic and variable voltages are wanted for ham radio purposes. It's based on ideas from many published circuits. There's no 'protection' on it so beware the real-life experiments. FETs are always cheap in LTspice.

Can several MOSFETs be stacked next to a resistive voltage divider and serve as one very high voltage part? why not. Last time I saw that kind of thing in real life was as a regulator in a 2200V current limited power supply. It had a row of very expensive Darlington transistors in there. FETs didn't exist then.


* ltspice seris HV fet circuit.png (49 KB, 700x815 - viewed 94 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
VE3ELQ
Contributing
Member
*
Offline Offline

Posts: 253


« Reply #1 on: March 17, 2019, 07:49:56 AM »

Your design looks workable to me but I have no experience with tubes.  A newer wide bandgap SIC FET might be a better choice.   The 1.7 KV rating is more than double that of the silicon FETs you propose so you could do it with only one in a simpler circuit. Need to ensure that the power dissipation is not exceeded when providing grid current.
 
https://www.wolfspeed.com/downloads/dl/file/id/173/product/13/c2m1000170d.pdf

73s  Nigel
Logged

A "Tube Free" zone.
KK4YY
Member

Offline Offline

Posts: 238


Golí na vydumku khitra


« Reply #2 on: March 17, 2019, 09:28:20 AM »

Only 200 pf Ciss on the C2M1000170D. That could come in handy.
Logged

Fate does not protect its worshipers any more than its deniers.
PA0NVD
Member

Offline Offline

Posts: 529


Nico and Chappie (Chappie is the dog...)


« Reply #3 on: March 17, 2019, 10:36:38 AM »

I worked with PEF technology, Pulsed Electric Field for sterelisation of liquids. There pulses in the order of 1/2 uSec were generated with FET's upto several megawatts. There are applications were a string of FET's is used to switch say 10 kV, so absolutely doable.
I did design with FET's but for higher power levels in the order of 100 MWatt, hollow anode hydrogen thyratrons are used. The first one I ever made for the university of Wageningen in the Netherlands used tubes and generated 100 - 1000 uSec pulsed at 50+ kWatts and 10 - 20 kV. Still more reliable in case of (frequent) short circuits or arcing. The short circuits and arcing did give problems for the FET strings.
Logged
KD6VXI
Contributing
Member
*
Online Online

Posts: 1995


Making AM GREAT Again!


« Reply #4 on: March 17, 2019, 02:57:12 PM »

I have fets that go to 1.5kv.  I use them in my pwm to reliably scream modulate. I've used them to 1.2kv.

BJFET I've seen even higher ratings on.

--Shane
KD6VXI
Logged
WBear2GCR
Member

Offline Offline

Posts: 3680


Brrrr- it's cold in the shack! Fire up the BIG RIG


WWW
« Reply #5 on: March 18, 2019, 10:18:27 PM »

A thought...

The follower shown looks like it's a stacked SE (single ended) type. So there is a lot of dissipation in the
series resistor from the -500vdc supply side. It may be a better idea to substitute a choke for the series
resistor... alternately a current mirror is commonly employed.

I guess the reason that a PP circuit isn't suggested is the lack of suitable Pch devices.

There are some Nch only follower circuits that use more than one device and manage to put some
feedback in, and this results in vanishingly low distortion specs. SPICE will likely give some ballpark
indication of the harmonic content of the proposed circuit. It may not matter much what the THD
is in ham radio application.
Logged

_-_- bear WB2GCR                   http://www.bearlabs.com
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 7157



WWW
« Reply #6 on: March 19, 2019, 01:01:24 PM »

On the thyratrons and liquid sterlization, that's very interesting! Guess it is not practical for killing bugs in home water though.
The largest thyratron here is the old glass 12 megawatt 5948A (attached) but there is a new ceramic one avail to those with deep pockets.

I am interested in a high voltage push pull audio circuit, so called 'totem pole' type. I just used what there was in LTspice to see if the stack looked like it would work. It seems simpler than bipolar. The interest in the single ended cirtcuit would be to do something like screen voltage modulation to follow the plate votage modulation. The high negative voltage isn't even needed, I think.. Who needs so much bias? Much of that can be lowered when only using the positive half cycle.

The primary push-pull application is a deflection current amplifier with a frequency response DC to >60KHz, and the challenge is to overcome the inductance of the inevitable coil windings so that the current is linear to the input voltage. I worked on this before, but for over a year I have had no time for it. The thing that makes the push pull project worthwhile is to get Asteroids to play on an old surplused air traffic control radar CRT.

It's like the old vector-display video games -but those used +/-35V at most and as I understand them, their bandwidth was wide but even the custom wound deflection yoke limited it severely.

The high voltage amp is there to overcome the inductance. The device dissipation would be a big deal. It rears its ugly head no matter what. Class G or H might work, but H is just moving the dissipation to another device. Class D for a 60KHz signal means a too-high switching frequency, I think.


* 5948A_huge_thyratron_tube.pdf (363.18 KB - downloaded 24 times.)
* DTS_HY-5948A_Thyratron.pdf (1367.42 KB - downloaded 20 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Pages: [1]   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.051 seconds with 18 queries.