Forums

fast, small low-capacitance zener to protect MOSFET gates

Started by Winfield Hill December 11, 2018
On Tuesday, December 11, 2018 at 4:40:29 PM UTC-5, John Larkin wrote:
> On Tue, 11 Dec 2018 13:15:44 -0800 (PST), George Herold > <gherold@teachspin.com> wrote: > > >On Tuesday, December 11, 2018 at 3:15:57 PM UTC-5, John Larkin wrote: > >> On 11 Dec 2018 06:25:40 -0800, Winfield Hill > >> <hill@rowland.harvard.edu> wrote: > >> > >> >I became discouraged of finding a suitable low-capacitance zener to protect the > >> >MOSFET gates in my high-voltage amplifier designs. (The protection may be > >> >optional, but one sleeps better with it in place.) Even the 50uA-rated > >> >low-current zener diodes are in fact larger-die types, with high capacitance. > >> >For example, the 7.5-volt MMSZ4693 has about 130pF at zero volts. > >> > > >> >My solution was to use a Diodes, Inc. D1213A-01WS TVS device, which has a 6 to > >> >10V breakdown and an amazing low 1pF of capacitance. But this can carry limited > >> >current in the forward direction, so I added a BAT54WS Schottky diode in > >> >parallel. Both devices are in small SOD-323 packages. Total capacitance, 11pF > >> >at their zero volt maximum, which doesn't add too awfully much to the 38pF Ciss > >> >of the small high-voltage power MOSFETs I'm using. > >> > > >> >Or maybe one of you has discovered a better solution? > >> > >> You could put a schottky in series with a zener. The first couple of > >> pulses will charge up the zener capacitance. > >Huh, I was going to suggest that, but I figured the series schottky > >would be forward biased, and so have a lot of capacitance. > >Is that 'wrong' thinking on my part? > > It wouldn't be forward biased much unless the zener were conducting, > which would mean it's protecting the fet.
Hmm, I guess I was thinking that it would be charging the zener C. How fast does it leak off through the schottky? (~1uA?)
> > One could add a resistor to keep the zener conducting, so that the > schottky was usually reverse biased. Or, equivalently, anchor the > schottky to some power supply.
Ahh... I'm not seeing something, if the zener is always conducting then isn't the series shocttky also? I'm not sure how to think about diodes on short time scales. A switch, with appropriate RCL, and then charge conservation.? George H.
> > > -- > > John Larkin Highland Technology, Inc > picosecond timing precision measurement > > jlarkin att highlandtechnology dott com > http://www.highlandtechnology.com
Winfield Hill <hill@rowland.harvard.edu> wrote in
news:puon7t01th6@drn.newsguy.com: 

> Phil Hobbs wrote... >> >>On 12/11/18 9:25 AM, Winfield Hill wrote: >>> I became discouraged of finding a suitable low-capacitance zener to >>> protect MOSFET gates in my high-voltage amplifier designs. (The >>> protection may be optional, but one sleeps better with it in place.) >>> Even the 50uA-rated low-current zener diodes are in fact larger-die >>> types, with high capacitance. For example, the 7.5-volt MMSZ4693 has >>> about 130pF at zero volts. >>> >>> My solution was to use a Diodes, Inc. D1213A-01WS TVS device, which >>> has a 6 to 10V breakdown and an amazing low 1pF of capacitance. But >>> it can only carry limited current in the forward direction, so I >>> added a BAT54WS Schottky diode in parallel. Both devices are in >>> small SOD-323 packages. Total capacitance, 11pF at their zero volt >>> maximum, which doesn't add too awfully much to the 38pF Ciss of my >>> small high-voltage power MOSFETs. >>> >>> Or maybe one of you has discovered a better solution? >> >> The Central Semi CSL05 looks decent at 1.2 pF, but it has a series >> diode to prevent forward current. How about the ESD0P2RF-02LS at >> 0.25 pF? > > Missed the CSL05, SL05, a typo? Anyway, yes, dozens of amazing low- > capacitance TVS protection devices available, handling high peak > currents for a few microseconds. But my designs (similar to AoE III > Fig 3.111, page 209) must handle 150mA continuous forward current, > ordinary zeners can do it just fine. That's what the added Schottky > diode is for. It contributes most of the capacitance, but increasing > the node capacitance from 38pF to 50pF is an acceptable penalty. > Maybe some small unipolar TVS can handle that, 10pF capacitance max? > >
I thought this item was commonly known as a tranzorb.
John Larkin wrote...
>Winfield Hill wrote: >> John Larkin wrote... >>> >>> You could put a schottky in series with a zener. The first >>> couple of pulses will charge up the zener capacitance. >> >> Interesting idea! > > I looked at the AoE page and see that this is a linear amp, not > the Pockels cell driver. So charging the zener isn't as obvious. > > I like to use PV optocouplers as gate drivers, because they float > on the fet source and don't have the huge gate over-voltage hazards. > > https://www.dropbox.com/s/pwi91ext9cclhzz/Gate_Driver_3.JPG?dl=0
The PV drivers are slow, I'm dealing with ns time-scales here. Both schemes involve two parts, so I'll stick with the parallel Schottky approach. In many linear amplifier applications, one has no control over when the device operation is invoked. Thanks for the suggestions. Your PV coupler scheme is frequently useful. We suggest it as well, AoE III, Fig 107, page 205, and Table 3.5, shows various floating low-resistance bipolar switches. -- Thanks, - Win
On Tue, 11 Dec 2018 16:26:27 -0800 (PST), George Herold
<gherold@teachspin.com> wrote:

>On Tuesday, December 11, 2018 at 4:40:29 PM UTC-5, John Larkin wrote: >> On Tue, 11 Dec 2018 13:15:44 -0800 (PST), George Herold >> <gherold@teachspin.com> wrote: >> >> >On Tuesday, December 11, 2018 at 3:15:57 PM UTC-5, John Larkin wrote: >> >> On 11 Dec 2018 06:25:40 -0800, Winfield Hill >> >> <hill@rowland.harvard.edu> wrote: >> >> >> >> >I became discouraged of finding a suitable low-capacitance zener to protect the >> >> >MOSFET gates in my high-voltage amplifier designs. (The protection may be >> >> >optional, but one sleeps better with it in place.) Even the 50uA-rated >> >> >low-current zener diodes are in fact larger-die types, with high capacitance. >> >> >For example, the 7.5-volt MMSZ4693 has about 130pF at zero volts. >> >> > >> >> >My solution was to use a Diodes, Inc. D1213A-01WS TVS device, which has a 6 to >> >> >10V breakdown and an amazing low 1pF of capacitance. But this can carry limited >> >> >current in the forward direction, so I added a BAT54WS Schottky diode in >> >> >parallel. Both devices are in small SOD-323 packages. Total capacitance, 11pF >> >> >at their zero volt maximum, which doesn't add too awfully much to the 38pF Ciss >> >> >of the small high-voltage power MOSFETs I'm using. >> >> > >> >> >Or maybe one of you has discovered a better solution? >> >> >> >> You could put a schottky in series with a zener. The first couple of >> >> pulses will charge up the zener capacitance. >> >Huh, I was going to suggest that, but I figured the series schottky >> >would be forward biased, and so have a lot of capacitance. >> >Is that 'wrong' thinking on my part? >> >> It wouldn't be forward biased much unless the zener were conducting, >> which would mean it's protecting the fet. >Hmm, I guess I was thinking that it would be charging the zener C. >How fast does it leak off through the schottky? (~1uA?) >> >> One could add a resistor to keep the zener conducting, so that the >> schottky was usually reverse biased. Or, equivalently, anchor the >> schottky to some power supply. >Ahh... I'm not seeing something, if the zener is always conducting >then isn't the series shocttky also?
Not if the resistor connects to the schottky-zener junction. Keep the zener voltage up, but don't forward bias the schittky. -- John Larkin Highland Technology, Inc lunatic fringe electronics
John Larkin <jjlarkin@highlandtechnology.com> wrote in 
news:tj111etlcst5pujo281qs9bpp0u8h5aa6o@4ax.com:

> schittky.
Oh Look! Larkin is into scatology. Oh... that's right... that is the stupid shit he accuses others of.
On Tuesday, December 11, 2018 at 10:56:18 PM UTC-5, John Larkin wrote:
> On Tue, 11 Dec 2018 16:26:27 -0800 (PST), George Herold > <gherold@teachspin.com> wrote: > > >On Tuesday, December 11, 2018 at 4:40:29 PM UTC-5, John Larkin wrote: > >> On Tue, 11 Dec 2018 13:15:44 -0800 (PST), George Herold > >> <gherold@teachspin.com> wrote: > >> > >> >On Tuesday, December 11, 2018 at 3:15:57 PM UTC-5, John Larkin wrote: > >> >> On 11 Dec 2018 06:25:40 -0800, Winfield Hill > >> >> <hill@rowland.harvard.edu> wrote: > >> >> > >> >> >I became discouraged of finding a suitable low-capacitance zener to protect the > >> >> >MOSFET gates in my high-voltage amplifier designs. (The protection may be > >> >> >optional, but one sleeps better with it in place.) Even the 50uA-rated > >> >> >low-current zener diodes are in fact larger-die types, with high capacitance. > >> >> >For example, the 7.5-volt MMSZ4693 has about 130pF at zero volts. > >> >> > > >> >> >My solution was to use a Diodes, Inc. D1213A-01WS TVS device, which has a 6 to > >> >> >10V breakdown and an amazing low 1pF of capacitance. But this can carry limited > >> >> >current in the forward direction, so I added a BAT54WS Schottky diode in > >> >> >parallel. Both devices are in small SOD-323 packages. Total capacitance, 11pF > >> >> >at their zero volt maximum, which doesn't add too awfully much to the 38pF Ciss > >> >> >of the small high-voltage power MOSFETs I'm using. > >> >> > > >> >> >Or maybe one of you has discovered a better solution? > >> >> > >> >> You could put a schottky in series with a zener. The first couple of > >> >> pulses will charge up the zener capacitance. > >> >Huh, I was going to suggest that, but I figured the series schottky > >> >would be forward biased, and so have a lot of capacitance. > >> >Is that 'wrong' thinking on my part? > >> > >> It wouldn't be forward biased much unless the zener were conducting, > >> which would mean it's protecting the fet. > >Hmm, I guess I was thinking that it would be charging the zener C. > >How fast does it leak off through the schottky? (~1uA?) > >> > >> One could add a resistor to keep the zener conducting, so that the > >> schottky was usually reverse biased. Or, equivalently, anchor the > >> schottky to some power supply. > >Ahh... I'm not seeing something, if the zener is always conducting > >then isn't the series shocttky also? > > Not if the resistor connects to the schottky-zener junction. Keep the > zener voltage up, but don't forward bias the schittky.
Right, I saw this in bed last night... but didn't get up to correct my post :^) George H.
> > > > -- > > John Larkin Highland Technology, Inc > > lunatic fringe electronics
On 12/12/18 10:02 am, Clifford Heath wrote:
> On 12/12/18 3:05 am, Winfield Hill wrote: >> Phil Hobbs wrote... >>> >>> On 12/11/18 9:25 AM, Winfield Hill wrote: >>>> I became discouraged of finding a suitable low-capacitance zener to >>>> protect >>>> MOSFET gates in my high-voltage amplifier designs.&Acirc;&nbsp; (The protection >>>> may be >>>> optional, but one sleeps better with it in place.)&Acirc;&nbsp; Even the 50uA-rated >>>> low-current zener diodes are in fact larger-die types, with high >>>> capacitance. >>>> For example, the 7.5-volt MMSZ4693 has about 130pF at zero volts. >>>> >>>> My solution was to use a Diodes, Inc. D1213A-01WS TVS device, which >>>> has a >>>> 6 to 10V breakdown and an amazing low 1pF of capacitance.&Acirc;&nbsp; But it can >>>> only carry limited current in the forward direction, so I added a >>>> BAT54WS >>>> Schottky diode in parallel.&Acirc;&nbsp; Both devices are in small SOD-323 >>>> packages. >>>> Total capacitance, 11pF at their zero volt maximum, which doesn't >>>> add too >>>> awfully much to the 38pF Ciss of my small high-voltage power MOSFETs. >>>> >>>> Or maybe one of you has discovered a better solution? >>> >>> The Central Semi CSL05 looks decent at 1.2 pF, but it has a series diode >>> to prevent forward current.&Acirc;&nbsp; How about the ESD0P2RF-02LS at 0.25 pF? >> >> &Acirc;&nbsp; Missed the CSL05, SL05, a typo?&Acirc;&nbsp; Anyway, yes, dozens of amazing low- >> &Acirc;&nbsp; capacitance TVS protection devices available, handling high peak >> &Acirc;&nbsp; currents for a few microseconds.&Acirc;&nbsp; But my designs (similar to AoE III >> &Acirc;&nbsp; Fig 3.111, page 209) must handle 150mA continuous forward current, >> &Acirc;&nbsp; ordinary zeners can do it just fine.&Acirc;&nbsp; That's what the added Schottky >> &Acirc;&nbsp; diode is for.&Acirc;&nbsp; It contributes most of the capacitance, but increasing >> &Acirc;&nbsp; the node capacitance from 38pF to 50pF is an acceptable penalty. > > Can you tune out the Schottky diode's capacitance with a small series L? > > Is there anything like a string of PIN diodes in a single package? > > I've just used Skyworks SMP1330 to protect an RF LNA. The combination of > low capacitance (0.7pF) and ability to sink power (almost 1 watt) is > remarkable. If I understand it, as the device approaches conduction the > I layer thins, increasing capacitance rapidly, so it becomes > "conductive" even before the diode turns on.
I'm still hoping someone will try to answer these questions. Winfield perhaps? Clifford Heath.
"Clifford Heath" <no.spam@please.net> wrote in message 
news:1gkQD.5$rw6.0@fx27.iad...
>> Can you tune out the Schottky diode's capacitance with a small series L? >> >> Is there anything like a string of PIN diodes in a single package? >> >> I've just used Skyworks SMP1330 to protect an RF LNA. The combination of >> low capacitance (0.7pF) and ability to sink power (almost 1 watt) is >> remarkable. If I understand it, as the device approaches conduction the >> I layer thins, increasing capacitance rapidly, so it becomes "conductive" >> even before the diode turns on. > > I'm still hoping someone will try to answer these questions. > > Winfield perhaps? >
You can only tune out by about 50%, corresponding to the famous doubling of bandwidth for peaked amplifiers. The ideal is closer to 3x for an infinite order network, but those are hard to build and tune, ;-) so we settle for 2nd order (CL), sometimes 3rd order (CLC). In this case, you can't argue with the capacitance being a parallel impedance, and all you can do is flatten the impedance with respect to frequency, i.e., swamp it with resistance. Class A gate drive? The solution is worse than the disease. :^) String of diodes, no idea. High voltage rectifiers (over 2kV or so). Think I've seen "thermistors" like that, for use in temperature compensation of audio amplifiers (a stack of diodes bonded to the heatsink, instead of a Vbe multiplier). Dunno if that's a standard product. Low voltage zeners work about as well as a stack of diodes, so *shrug*. There's also LEDs, which probably work better, though they aren't rated for all that much peak forward current (like a few amps for less than a microsecond?). Maybe a big one, but the capacitance will be high yet again. Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: https://www.seventransistorlabs.com/
On Thursday, 13 December 2018 08:35:01 UTC, Tim Williams  wrote:
> "Clifford Heath" <no.spam@please.net> wrote in message > news:1gkQD.5$rw6.0@fx27.iad... > >> Can you tune out the Schottky diode's capacitance with a small series L? > >> > >> Is there anything like a string of PIN diodes in a single package? > >> > >> I've just used Skyworks SMP1330 to protect an RF LNA. The combination of > >> low capacitance (0.7pF) and ability to sink power (almost 1 watt) is > >> remarkable. If I understand it, as the device approaches conduction the > >> I layer thins, increasing capacitance rapidly, so it becomes "conductive" > >> even before the diode turns on. > > > > I'm still hoping someone will try to answer these questions. > > > > Winfield perhaps? > > > > You can only tune out by about 50%, corresponding to the famous doubling of > bandwidth for peaked amplifiers. The ideal is closer to 3x for an infinite > order network, but those are hard to build and tune, ;-) so we settle for > 2nd order (CL), sometimes 3rd order (CLC). > > In this case, you can't argue with the capacitance being a parallel > impedance, and all you can do is flatten the impedance with respect to > frequency, i.e., swamp it with resistance. > > Class A gate drive? The solution is worse than the disease. :^) > > String of diodes, no idea. High voltage rectifiers (over 2kV or so). Think > I've seen "thermistors" like that, for use in temperature compensation of > audio amplifiers (a stack of diodes bonded to the heatsink, instead of a Vbe > multiplier). Dunno if that's a standard product. > > Low voltage zeners work about as well as a stack of diodes, so *shrug*. > There's also LEDs, which probably work better, though they aren't rated for > all that much peak forward current (like a few amps for less than a > microsecond?). Maybe a big one, but the capacitance will be high yet again. > > Tim
A stack of low v zeners gives a very soft knee. The softness only adds. And a lot more C than vanilla diodes. NT
On 13/12/18 7:34 pm, Tim Williams wrote:
> "Clifford Heath" <no.spam@please.net> wrote in message > news:1gkQD.5$rw6.0@fx27.iad... >>> Can you tune out the Schottky diode's capacitance with a small series L? >>> >>> Is there anything like a string of PIN diodes in a single package? >>> >>> I've just used Skyworks SMP1330 to protect an RF LNA. The combination >>> of low capacitance (0.7pF) and ability to sink power (almost 1 watt) >>> is remarkable. If I understand it, as the device approaches >>> conduction the >>> I layer thins, increasing capacitance rapidly, so it becomes >>> "conductive" even before the diode turns on. >> >> I'm still hoping someone will try to answer these questions. >> >> Winfield perhaps? > > You can only tune out by about 50%, corresponding to the famous doubling > of bandwidth for peaked amplifiers.&nbsp; The ideal is closer to 3x for an > infinite order network, but those are hard to build and tune, ;-) so we > settle for 2nd order (CL), sometimes 3rd order (CLC).
Hmmm, fair enough.
> String of diodes, no idea.
A stack of low-C PIN diodes, to be specific, with the sudden C change around threshold.
> High voltage rectifiers (over 2kV or so).
Those actually are PIN, but just too big (too much C, even in series).
> Low voltage zeners work about as well as a stack of diodes, so *shrug*.
As a stack of PIN diodes? These don't turn on like ordinary diodes, is my point. I bet there's a package that has four pairs that could be strung together. Clifford Heath.