3
\$\begingroup\$

I am constructing a circuit designed to release a very brief pulse of current at a voltage of 1000V. A low power DC converter is used to charge a 0.1uf pulse capacitor to 1000V, and then I intend to use an n-channel MOSFET to discharge the stored energy into a very low (~10 ohm) resistance load. According to the manufacturer (KEMET) the capacitor is capable of fully discharging on the scale of multiple nanoseconds. For my application, the discharge duration is quite important, and the faster the better. In my tests, connecting the MOSFET without any pull down or gate resistors, some current leaks across the MOSFET into the load. For my application, it is very important that very little current flows across the load when the transistor is "off." The signal for "switching" (saturating?) the MOSFET is to be supplied by a function generator. As such I have two questions:

  1. Is a high voltage MOSFET appropriate for this application, keeping the requirement of low leakage current and fast switching in mind? Is there a better component? (I have purchased a Vishay IRFPG50 (1000v, 6.1A))

  2. To limit the leakage current, where and of what value resistor should I add to the circuit? Reading other posts, I have heard discussion of adding a resistor to the gate of the MOSFET. What would this achieve and are there other similar common practices I should know about?

Circuit specifications: MOSFET: Vishay IRFPG50 Capacitor: KEMET C4540H104KUGWCT100

\$\endgroup\$
9
  • \$\begingroup\$ 1000 V rating is wayyy too low for a 1000 V application. Try 1700 V. \$\endgroup\$
    – winny
    Oct 21, 2020 at 18:51
  • 1
    \$\begingroup\$ IGBT might be an option \$\endgroup\$ Oct 21, 2020 at 18:58
  • \$\begingroup\$ The discharge duration is going to be governed by the resistance of your load not the switching speed of the MOSFET. How high is "very high" exactly? \$\endgroup\$
    – Finbarr
    Oct 21, 2020 at 19:32
  • \$\begingroup\$ use an optocoupler to drive mosfet/igbt gate. \$\endgroup\$
    – user266789
    Oct 21, 2020 at 20:05
  • 1
    \$\begingroup\$ If the signal generator gate drive is low when 'off' it will (of course) drive gate low so considering only resistance a pull down gate resistor does not do much. HOWEVER: . At very high speeds gate ringing MAY be a factor. A small series resistor (start with around 10 ohms) between generator and gate may help. A reverse polarity Schottky diode connected gate-source and mounted as close as physically & electrically possible to FET will clamp any negative ringing half cycles and quickly damp gate ringing. A parallel g-s resistor then probably does no harm. | A 1,000V FET is VERY marginal. \$\endgroup\$
    – Russell McMahon
    Oct 21, 2020 at 22:51

1 Answer 1

3
\$\begingroup\$

is a high voltage MOSFET appropriate for this application, keeping the requirement of low leakage current and fast switching in mind? Is there a better component? (I have purchased a Vishay IRFPG50 (1000v, 6.1A))

The types of components that are appropriate depend on what you need your switching to look like. If you just need to turn on quickly and drain the whole capacitor into the load then you could use either an IGBT, MOSFET, or a mercury wetted relay.

If you also need to turn off quickly to control the pulse width or deliver multiple pluses then only the MOSFET is suitable. IGBTs often have long turn off times (a few us), whereas MOSFET's are capable of both fast turn on and turn off (ns to 10s of ns range).

As far as leakage goes, a mercury-wetted relay is going to be the lowest leakage in the off state. MOSFETs can also be as low as a few uA worst case. Silicon carbide MOSFETs tend to have lower leakage than regular Silicon MOSETs. For IGBTs expect leakage in the range of a few 10s of uA.

To limit the leakage current, where and of what value resistor should I add to the circuit? Reading other posts, I have heard discussion of adding a resistor to the gate of the MOSFET. What would this achieve and are there other similar common practices I should know about?

For MOSFETs the leakage is usually specified at maximum operating voltage and with VGS=0V. To decrease leakage you can get a device with a higher operating voltage relative to your working voltage, and the leakage will decrease accordingly. Also, driving VGS to a negative voltage (say -10V) will decrease leakage even further compared to just making VGS=0V.

The IRFPG50 probably isn't appropriate for your application.

  1. Its only rated for 1000V. That means you have no safety margin if you operate at 1000V.
  2. Its only rated for 24A peak current. For any loads less than 40 ohms you will exceed the rated maximum.
  3. The datasheet says 100uA of leakage current at VDS=1000V. You didn't really specify how much leakage was too much, but 100uA is pretty high.
  4. Some distributors are listing the IRFPG50 as obsolete. So if you plan to build more of these you may have trouble finding them in the future.

Some alternatives might be...
C2M0045170D , 1700V, 45mOhm, 160A peak, 2uA typical leakage
https://www.wolfspeed.com/downloads/dl/file/id/960/product/207/c2m0045170d.pdf

https://www.digikey.com/en/products/detail/cree-wolfspeed/C2M0045170D/6148740

STW21N150K5, 1500V, 0.7 ohm, 56A peak, 1uA max leakage
https://www.st.com/content/ccc/resource/technical/document/datasheet/0f/96/82/a0/4d/08/4b/18/DM00130400.pdf/files/DM00130400.pdf/jcr:content/translations/en.DM00130400.pdf

https://www.digikey.com/en/products/detail/stmicroelectronics/STW21N150K5/5455741

DRIVING:
For driving the gate, you can either use an isolated gate driver IC, or a pulse transformer.

Some possible gate driver ICs include...

Analog Devices ADuMxxx series.
ADuM4122, 1660V peak, 2A drive, 150kV/us immunity, 48ns tpd, magnetic
https://www.analog.com/media/en/technical-documentation/data-sheets/ADuM4122.pdf

Infineon 1EDxxx EiceDriver series.
1EDC60H12AHXUMA1, 1200V, 6A drive, 100kV/us immunity, 300ns tpd, capacitive.
https://www.infineon.com/dgdl/Infineon-1EDCxxI12AH-DS-v02_00-EN.pdf?fileId=5546d4625d5945ed015d88ad8a38235e

Silicon Labs Si826xxx series
SI8261BCD-C-ISR, 1414V working voltage, 4A drive, 35Kv/us immunity, 60ns tpd, capacitive
https://media.digikey.com/pdf/Data%20Sheets/Silicon%20Laboratories%20PDFs/Si826x_Rev2015.pdf

Texas Instruments
UCC23513BDWY, 1500V DC, 4.5A/5.3A drive, 150kV/us immunity, 105ns tpd, optical.
https://www.ti.com/lit/ds/symlink/ucc23513.pdf?HQS=TI-null-null-digikeymode-df-pf-null-wwe&ts=1606534057238

UCC5390ECQDWVRQ1, 2121V DC, 10A drive, 100kV/us immunity, 100ns tpd, capacitive.
https://www.ti.com/lit/ds/symlink/ucc5390-q1.pdf?HQS=TI-null-null-digikeymode-df-pf-null-wwe&ts=1606533698123

ISOLATED DRIVER SUPPLY:
All of the above drivers require a floating isolated supply to work. Analog devices does offer one IC with a built in isolated supply, but its not rated for 1000V so you must use an external one. Some possible options are...

MGJ6D241505WMC-R7, 3kV working isolation, +15V/-5V, 15pF isolation capacitance.
https://www.murata.com/products/productdata/8807030063134/kdc-mgj6-14mm.pdf?1583754812000

MGJ2D121505SC, 2.4KV working isolation, +15V/-5V, 3pF isolation capacitance. https://www.murata.com/products/productdata/8807029997598/kdc-mgj2.pdf?1583754812000

\$\endgroup\$
2
  • \$\begingroup\$ Thank you so much for your answer, I will use the STW21N150K5. Would you be able to give an example of an appropriate gate driver IC? I have searched on Digikey but do not know where to begin. For my application, I would like to be able to trigger a single pulse with the push of a button. \$\endgroup\$
    – Joshua S.
    Nov 27, 2020 at 19:14
  • 1
    \$\begingroup\$ @JoshuaS. I added some example gate drivers, and an isolated supply option to power them. The relevant Digi-key categories for these types of parts would be "Isolators - Gate Drivers" and "Power Supplies - Board Mount". The power supplies for driving MOSFETs are specially designed to have very low isolation capacitance across the barrier. They also typically have bipolar outputs. \$\endgroup\$
    – user4574
    Nov 28, 2020 at 4:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.