Forums

addendum to 45-year MOSFET history, comments?

Started by Winfield Hill July 28, 2019
 I've written an addendum to the 18-page 45-year
 power MOSFET history article for the x-Chapters.

https://www.dropbox.com/s/0qh8dxgielrkqvq/chap-3x.11_MOSFET-saga_2019-07-19.pdf?dl=1

 And would appreciate your comments.  Here it is:

As we leave this lengthy section, discussing various intricacies 
in the history of power MOSFETs, we have to admit not mentioning 
a few important MOSFET topics: 

(1.) RF power MOSFETs.  These are specialized beasts that work 
to high power levels (750 watts) and to high frequencies (tens 
of GHz).  This is a very active fast-growing area.

(2.) Lateral power MOSFETs.  These avoid using V-grooves and 
other high-channel-density schemes.  They feature a negative 
tempco bias property, see sections 3.6.4 and 3x.18, are 
required meat for high-power linear audio power amplifiers.  
ProFusion sells parts made by Exicon.   Hitachi abandoned the 
market, and now users pray these parts won't disappear.

(3.) Depletion-mode MOSFETs.  Unlike enhancement-mode MOSFETs, 
which are normally off and require a positive bias to turn on, 
these are normally on, and require a reverse-bias to turn off.  
See the discussion in section 3.6.2 and Table 3.6 of the main 
book, and also section 9.3.14 for applications.  In this volume 
see sections 3x.6, 4x.23, 9x.3 and 9x.12.  Very nice depletion-
mode parts are available, and seem to have a stable marketplace, 
however we aren't seeing new parts being introduced.  There are 
only n-channel types available.

(4.) Pioneered by companies like Agilent (spun off to Avago, and 
purchased by Broadcom), super-fast high-frequency small-signal 
MOSFETs flourished.  With technologies like pHEMT (pseudomorphic 
high-electron-mobility transistor) and E-pHEMT (enhancement-mode 
pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, 
a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package.  
But these handy parts have disappeared, as semiconductor 
manufacturers created IC-based solutions more attractive to the 
telecom customer base.**  Now we are left with parts like the 
SAV-551+, made by Mini-Circuits (thank you!) and not even 
appearing on Octopart  This part substitutes for Avago's 
ATF-55143, and features 2 ohms Ron and 0.4pF Cout, numbers we 
were getting used to in the good-old-days.

** footnote: Of the parts on John Larkin's 2017 small pHEMT list 
(with pinouts and SMT labels), only CEL's CE3514 is still 
available, and Mouser has a pile of SKY65050 parts left in stock.


-- 
 Thanks,
    - Win
On 28 Jul 2019 06:18:11 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

> I've written an addendum to the 18-page 45-year > power MOSFET history article for the x-Chapters. > >https://www.dropbox.com/s/0qh8dxgielrkqvq/chap-3x.11_MOSFET-saga_2019-07-19.pdf?dl=1 > > And would appreciate your comments. Here it is: > >As we leave this lengthy section, discussing various intricacies >in the history of power MOSFETs, we have to admit not mentioning >a few important MOSFET topics: > >(1.) RF power MOSFETs. These are specialized beasts that work >to high power levels (750 watts) and to high frequencies (tens >of GHz). This is a very active fast-growing area. > >(2.) Lateral power MOSFETs. These avoid using V-grooves and >other high-channel-density schemes. They feature a negative >tempco bias property, see sections 3.6.4 and 3x.18, are >required meat for high-power linear audio power amplifiers. >ProFusion sells parts made by Exicon. Hitachi abandoned the >market, and now users pray these parts won't disappear. > >(3.) Depletion-mode MOSFETs. Unlike enhancement-mode MOSFETs, >which are normally off and require a positive bias to turn on, >these are normally on, and require a reverse-bias to turn off. >See the discussion in section 3.6.2 and Table 3.6 of the main >book, and also section 9.3.14 for applications. In this volume >see sections 3x.6, 4x.23, 9x.3 and 9x.12. Very nice depletion- >mode parts are available, and seem to have a stable marketplace, >however we aren't seeing new parts being introduced. There are >only n-channel types available. > >(4.) Pioneered by companies like Agilent (spun off to Avago, and >purchased by Broadcom), super-fast high-frequency small-signal >MOSFETs flourished. With technologies like pHEMT (pseudomorphic >high-electron-mobility transistor) and E-pHEMT (enhancement-mode >pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, >a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package. >But these handy parts have disappeared, as semiconductor >manufacturers created IC-based solutions more attractive to the >telecom customer base.** Now we are left with parts like the >SAV-551+, made by Mini-Circuits (thank you!) and not even >appearing on Octopart This part substitutes for Avago's >ATF-55143, and features 2 ohms Ron and 0.4pF Cout, numbers we >were getting used to in the good-old-days. > >** footnote: Of the parts on John Larkin's 2017 small pHEMT list >(with pinouts and SMT labels), only CEL's CE3514 is still >available, and Mouser has a pile of SKY65050 parts left in stock.
The really sad demise is the ATF-50189, a power phemt in a SOT-89 package. It was rated for 7 volts but, in the tradition of RF parts, was good for a lot more. I have/had a beautiful pulse generator output stage, and some great laser drivers, that use them. There were nice SOT-89 mesfets, long gone. I'm having to transition some designs to GaN, which needs a lot more gate drive, but can switch a lot more volts. With a roughly factor of 4 improvement in slew and swing, we might consider opamps. SAV-551 is a nice little part. Rds-on is impressive. -- John Larkin Highland Technology, Inc lunatic fringe electronics
On 28/07/2019 2:18 pm, Winfield Hill wrote:
> I've written an addendum to the 18-page 45-year > power MOSFET history article for the x-Chapters. > > https://www.dropbox.com/s/0qh8dxgielrkqvq/chap-3x.11_MOSFET-saga_2019-07-19.pdf?dl=1 > > And would appreciate your comments. Here it is: > > As we leave this lengthy section, discussing various intricacies > in the history of power MOSFETs, we have to admit not mentioning > a few important MOSFET topics: > > (1.) RF power MOSFETs. These are specialized beasts that work > to high power levels (750 watts) and to high frequencies (tens > of GHz). This is a very active fast-growing area. > > (2.) Lateral power MOSFETs. These avoid using V-grooves and > other high-channel-density schemes. They feature a negative > tempco bias property, see sections 3.6.4 and 3x.18, are > required meat for high-power linear audio power amplifiers. > ProFusion sells parts made by Exicon. Hitachi abandoned the > market, and now users pray these parts won't disappear. > > (3.) Depletion-mode MOSFETs. Unlike enhancement-mode MOSFETs, > which are normally off and require a positive bias to turn on, > these are normally on, and require a reverse-bias to turn off. > See the discussion in section 3.6.2 and Table 3.6 of the main > book, and also section 9.3.14 for applications. In this volume > see sections 3x.6, 4x.23, 9x.3 and 9x.12. Very nice depletion- > mode parts are available, and seem to have a stable marketplace, > however we aren't seeing new parts being introduced. There are > only n-channel types available. > > (4.) Pioneered by companies like Agilent (spun off to Avago, and > purchased by Broadcom), super-fast high-frequency small-signal > MOSFETs flourished. With technologies like pHEMT (pseudomorphic > high-electron-mobility transistor) and E-pHEMT (enhancement-mode > pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, > a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package. > But these handy parts have disappeared, as semiconductor > manufacturers created IC-based solutions more attractive to the > telecom customer base.** Now we are left with parts like the > SAV-551+, made by Mini-Circuits (thank you!) and not even > appearing on Octopart This part substitutes for Avago's > ATF-55143, and features 2 ohms Ron and 0.4pF Cout, numbers we > were getting used to in the good-old-days. > > ** footnote: Of the parts on John Larkin's 2017 small pHEMT list > (with pinouts and SMT labels), only CEL's CE3514 is still > available, and Mouser has a pile of SKY65050 parts left in stock. > >
Thanks Win, I have just dipped in so far but on page99, second line (about SIC gate drive): ...s (early SiC parts required +20 V and 5 V), b... Shouldn't that be -5V? Over-enhancing (a depletion mode part) and over-depleting (an enhancement mode part) are worthwhile tricks to keep in the designers toolbox. After-all the zero bias distinction between enh and depl is not always so simplistic. piglet
piglet wrote...
> >...s (early SiC parts required +20 V and 5 V), b... > > Shouldn't that be -5V?
Yes, thanks, I've fixed it. -- Thanks, - Win
On Sun, 28 Jul 2019 22:13:07 +0100, piglet <erichpwagner@hotmail.com>
wrote:

>On 28/07/2019 2:18 pm, Winfield Hill wrote: >> I've written an addendum to the 18-page 45-year >> power MOSFET history article for the x-Chapters. >> >> https://www.dropbox.com/s/0qh8dxgielrkqvq/chap-3x.11_MOSFET-saga_2019-07-19.pdf?dl=1 >> >> And would appreciate your comments. Here it is: >> >> As we leave this lengthy section, discussing various intricacies >> in the history of power MOSFETs, we have to admit not mentioning >> a few important MOSFET topics: >> >> (1.) RF power MOSFETs. These are specialized beasts that work >> to high power levels (750 watts) and to high frequencies (tens >> of GHz). This is a very active fast-growing area. >> >> (2.) Lateral power MOSFETs. These avoid using V-grooves and >> other high-channel-density schemes. They feature a negative >> tempco bias property, see sections 3.6.4 and 3x.18, are >> required meat for high-power linear audio power amplifiers. >> ProFusion sells parts made by Exicon. Hitachi abandoned the >> market, and now users pray these parts won't disappear. >> >> (3.) Depletion-mode MOSFETs. Unlike enhancement-mode MOSFETs, >> which are normally off and require a positive bias to turn on, >> these are normally on, and require a reverse-bias to turn off. >> See the discussion in section 3.6.2 and Table 3.6 of the main >> book, and also section 9.3.14 for applications. In this volume >> see sections 3x.6, 4x.23, 9x.3 and 9x.12. Very nice depletion- >> mode parts are available, and seem to have a stable marketplace, >> however we aren't seeing new parts being introduced. There are >> only n-channel types available. >> >> (4.) Pioneered by companies like Agilent (spun off to Avago, and >> purchased by Broadcom), super-fast high-frequency small-signal >> MOSFETs flourished. With technologies like pHEMT (pseudomorphic >> high-electron-mobility transistor) and E-pHEMT (enhancement-mode >> pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, >> a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package. >> But these handy parts have disappeared, as semiconductor >> manufacturers created IC-based solutions more attractive to the >> telecom customer base.** Now we are left with parts like the >> SAV-551+, made by Mini-Circuits (thank you!) and not even >> appearing on Octopart This part substitutes for Avago's >> ATF-55143, and features 2 ohms Ron and 0.4pF Cout, numbers we >> were getting used to in the good-old-days. >> >> ** footnote: Of the parts on John Larkin's 2017 small pHEMT list >> (with pinouts and SMT labels), only CEL's CE3514 is still >> available, and Mouser has a pile of SKY65050 parts left in stock. >> >> > >Thanks Win, I have just dipped in so far but on page99, second line >(about SIC gate drive): > >...s (early SiC parts required +20 V and 5 V), b... > >Shouldn't that be -5V? > >Over-enhancing (a depletion mode part) and over-depleting (an >enhancement mode part) are worthwhile tricks to keep in the designers >toolbox. After-all the zero bias distinction between enh and depl is not >always so simplistic. > >piglet >
The gates of silicon mosfets can be driven plenty hard without fear of blowing out the gate oxide. The margin is roughly 8:1. SiC fets seem to have very little margin between good gate drive and destruction. Given the high equivalent gate resistances of many SiC parts, they need to be driven hard to switch fast. GaN gates are similarly restricted. Not much more than a healthy drive voltage can damage them. -- John Larkin Highland Technology, Inc lunatic fringe electronics
I have not had the time to read it through thoroughly, but I do not see the Spirito effect mentioned:

https://www.google.com/search?client=safari&channel=iphone_bm&source=hp&ei=1i8-XdiZKoOKrwSf0q4I&q=spirito+effect&oq=spirito+eff&gs_l=mobile-gws-wiz-hp.1.0.0l2j0i22i30l6.9576.15921..16887...0.0..0.164.1147.10j3......0....1.......0..46j46i275j0i10.vrrvDa3Mf-g

Cheers 

Klaus
Klaus Kragelund wrote...
> > I have not had the time to read it through thoroughly, > but I do not see the Spirito effect mentioned:
It's going to get a proper mention. -- Thanks, - Win
On 28.07.19 2:18 PM, Winfield Hill wrote:
> With technologies like pHEMT (pseudomorphic > high-electron-mobility transistor) and E-pHEMT (enhancement-mode > pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, > a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package.
Small suggestion: Mentioning e-pHEMT right next to the ATF-38143, which is a depletion-mode part seems a bit suboptimal. Maybe just leave out the e/d distinction altogether? &mdash; David
On 7/29/19 9:13 AM, David Nadlinger wrote:
> On 28.07.19 2:18 PM, Winfield Hill wrote: >> With technologies like pHEMT (pseudomorphic >> high-electron-mobility transistor) and E-pHEMT (enhancement-mode >> pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, >> a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package. > > Small suggestion: Mentioning e-pHEMT right next to the ATF-38143, which > is a depletion-mode part seems a bit suboptimal. Maybe just leave out > the e/d distinction altogether? > > &nbsp;&mdash; David
RIP. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Phil Hobbs wrote...
> > On 7/29/19 9:13 AM, David Nadlinger wrote: >> On 28.07.19 2:18 PM, Winfield Hill wrote: >>> With technologies like pHEMT (pseudomorphic >>> high-electron-mobility transistor) and E-pHEMT (enhancement-mode >>> pHEMT), we got inexpensive discrete parts like Avago's ATF-38143, >>> a 10GHz 4-volt 200mA FET in a convenient 4-lead SC-70 package. >> >> Small suggestion: Mentioning e-pHEMT right next to the ATF-38143, >> which is a depletion-mode part seems a bit suboptimal. Maybe just >> leave out the e/d distinction altogether? >> >> &nbsp;&mdash; David > > RIP.
Yes, I noticed that issue, will re-write a few sentences. -- Thanks, - Win