Author Topic: Getting the MOSFET to act as much as possible as an ideal switch  (Read 1382 times)

0 Members and 1 Guest are viewing this topic.

Offline sagar448Topic starter

  • Newbie
  • Posts: 2
  • Country: ca
Getting the MOSFET to act as much as possible as an ideal switch
« on: February 08, 2022, 08:37:30 pm »
Hello All,

I've been playing around with Mosfets for a couple months now and long story short i've been trying to get the Mosfet to act as much as possible as an ideal switch.

Caution, not suitable for beginners :P

A mosfet has 3 types of parasitic capacitances due to its intrinsic and extrinsic properties. There are two types of capacitances. Overlap and junction. Anyways The 3 types of capacitances are gate to drain, gate to source, and drain to source. The issue is the drain to source capacitance. This is the reason why the MOSFET cannot act like an ideal switch. The drain and source are connected even though the MOSFET is off. There's obviously no way to change that property but maybe there's something can be placed after or before the mosfet to make it closer to an ideal switch?

Look at attached image, I put two mosfets back to back with their sources connected. G1 is the gate signal to turn both of them on and off. The voltage being carried through the mosfet is approx 300VDC or higher. I'd like to remain in the DC domain e.g. no transformers etc.

I cannot talk about the application as this is something i've been working on independently for a year now but feel free to ask any other questions!


 

Offline magic

  • Super Contributor
  • ***
  • Posts: 7184
  • Country: pl
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #1 on: February 08, 2022, 08:57:38 pm »
And why does it bother you? Is the capacitance loading some circuit? Is it causing signals or noise to leak from one side to another?

If the latter, put two of your MOSFET SSRs in series and ground the link between them with yet another SSR. That will take care of capacitive coupling, methinks. Make sure to route cables apart from each other or all will be in vain :P
 

Offline sagar448Topic starter

  • Newbie
  • Posts: 2
  • Country: ca
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #2 on: February 08, 2022, 09:13:03 pm »
Thanks for the reply!

It is the latter, its the leak.

Unfortunately cannot use an SSR. They cannot switch fast enough, these mosfets are being switched in the kilohertz range.
 

Offline magic

  • Super Contributor
  • ***
  • Posts: 7184
  • Country: pl
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #3 on: February 08, 2022, 09:19:40 pm »
By SSR, I mean your combination of two MOSFETs. I presume they are meant to switch some voltage and block in both direction. There exist off the shelf SSRs which contain exactly this inside, plus some gate drive circuitry. (I don't know if they are fast enough for kHz, never used that stuff).

The point is, if you double your switches and ground the internal connection when the switches are off, signal leaks will disappear.
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 8550
  • Country: us
    • SiliconValleyGarage
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #4 on: February 08, 2022, 09:20:26 pm »
what do you mean with 'leak' ? are you switching an AC signal ?
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3849
  • Country: us
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #5 on: February 08, 2022, 09:20:55 pm »
What type Ron do you need?
Maybe consider a depletion mode device like GaN?

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline Manul

  • Super Contributor
  • ***
  • Posts: 1158
  • Country: lt
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #6 on: February 08, 2022, 11:09:11 pm »
It is not clear, are you worried that there is capacitive coupling from one side to another OR that the switch presents itself as a capacitance for the signal source?
 

Offline Terry Bites

  • Super Contributor
  • ***
  • Posts: 2504
  • Country: gb
  • Recovering Electrical Engineer
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #7 on: February 08, 2022, 11:48:50 pm »
Often overlooked, is the fact that no Mosfet can be completely turned off.  IDSS gets worse as you go up in current rating and temperature.  its incurable.
All semiconductor switches leak.
There is AC  and DC leakage. AC via internal capacitances which exist between all terminals and internal junctions. DC resistive paths existist between all terninals. Internal junction potentials exist beween all terminals.
Its doesnt matter what kind of device or material you use.

Tough days down there at DARPA labs huh!
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22382
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #8 on: February 09, 2022, 03:52:49 am »
Also, if you can afford some charge injection, or some other means of managing it -- the capacitance will be lower under bias, i.e. high Vds.  This is especially pronounced for power MOSFETs, which are not lateral (so not subject to overlap, etc. capacitances familiar to lateral IC MOSFETs), but of vertical and trench designs where the bulk of the substrate itself (or the lightly-doped epitaxy layer on top of it, as the case may be) serves as the off-state capacitor: in particular, SuperJunction trench MOS can have a capacitance ratio of >100:1, with the biggest change in the 10-100V range (depending on rating).  The SuperJunction design is... hard to describe, but it has the effect of the capacitance itself practically turning off like a switch, as almost the entire bulk becomes depleted at once (hence the abrupt change in C).  Put another way, the thickness of the capacitor suddenly increases from 100s nm, to 10s of um.

Integrated analog switches have a good compromise between on/off ratio; they are specified in terms of insertion loss when on, and attenuation when off, with respect to a given (e.g. 50 ohm) source/load impedance.  Capacitances are somewhat irrelevant in that case.  Whether this helps you consider your application, I don't know.

Analog switches may internally use tee structures (alternating series/shunt/series switches) to achieve such high ratios.  The main downside is probably the limited voltage range, which must be within the supplies; a switch capable of some 100s MHz bandwidth with quite respectable specs, might only be rated for 3.3 or +/-5V.  There are fortunately fairly few applications you'd need more (and, when you do, there are proper RF switches to handle that).

If the source/load impedances cannot be well defined (e.g., electrometer input something or other?), a custom solution may be required, or at least ye old fashioned mechanical switches (which offer truly incredible on/off ratios -- do not underestimate them!).


As for the application -- regarding high voltages, if this might be something like mains switching, keep in mind not just the expected (nominal peak) voltage, but also the surge rating, and any characteristics the load might have (inrush current, inductive turn-off).  For these reasons, semiconductor applications have largely been confined to only the most robust devices (diodes, SCRs, TRIACs).

Modern devices are just beginning to offer adequate ratings for real power line switching use: SiC MOSFETs with ~kV ratings are adequate to handle surge at low voltages (i.e. 240VAC + 2.5kV surge), but tall stacks of them are needed to also pass the available fault current on such a circuit (~1kA)!

Tim
« Last Edit: February 09, 2022, 03:54:53 am by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3570
  • Country: fr
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #9 on: February 09, 2022, 09:21:52 am »
Bonjour A waste of our time if you can't reveal the environment and applications requirements.

I can assure you that your application has been done long ago. Search a few patents.

A complete system diagram or schematic will yield better answers.

Finally no "added component" can change the fet self capacité DS, and any switch eg GaN or SiC has capacity


Jon


Jon

Jean-Paul  the Internet Dinosaur
 

Offline Terry Bites

  • Super Contributor
  • ***
  • Posts: 2504
  • Country: gb
  • Recovering Electrical Engineer
Re: Getting the MOSFET to act as much as possible as an ideal switch
« Reply #10 on: February 09, 2022, 08:48:41 pm »
A waste of time indeed.  Ars longa, vita brevis and such.

Here the real subject is the MOSFET's environment and how well adapted it is to it.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf