4

I have an application with some constraints for which it feels difficult to find a solution.

  1. load current more than 10A, so N channel MOSFET is preferred;

  2. working voltage is a bit high, between 50-80V;

  3. logic switch control signal, no driving capability while expecting very fast response;

  4. due to some reason, high side switching is required. That is, the MOSFET will be between power source and load. The load is non-inductive like one in buck converter.

I have searched the web for high-side MOSFET driver, but all are with buck type of application, coming with external capacitor connected to source of MOSFET to get ground rail to be charged. Which I don't think it works in my application where no ground rail exists after switch-on.

I found few drivers with charge pump built-in, but below 36V.

I know a workaround that uses a battery and connects the negative of the battery to source of MOSFET, but this is not a solution as I have no way to charge the battery and so it has to be replaced regularly.

LorenzoDonati4Ukraine-OnStrike
  • 24,350
  • 5
  • 62
  • 107
Spark
  • 41
  • 1
  • 1
  • 2
  • MAXIM makes an isolated Power Converter, using a transformer. – analogsystemsrf May 13 '17 at 11:51
  • 7
    There are plenty of P-channel MOSFETS that will happily drive a 10A load. The problem with using an N-channel as a hight side switch is that you need a gate voltage higher than the voltage you're switching. See this question [Switching DC with MOSFET: p-Channel or n-Channel; Low Side Load or High Side Load?](https://electronics.stackexchange.com/questions/18884/switching-dc-with-mosfet-p-channel-or-n-channel-low-side-load-or-high-side-loa) for a full explanation. – Finbarr May 13 '17 at 11:53
  • 1
    I use IR LEDs to drive n-chans when I need a higher Vgd than handy... – dandavis May 13 '17 at 12:01
  • @dandavis How would you do that? Without further explanation, your comment is unlikely to help anyone ;) – marcelm May 14 '17 at 10:08
  • 1
    a bright white led shining directly into a red led will cause to red LED to output ~1.5v. connecting them in series adds voltage. Alternating flat LEDs are very easy to wire in series into a compact light target. Shunt a 1-10m GS bleeder to drive MOSFETS. – dandavis May 14 '17 at 16:55
  • @dandavis Ahh, interesting! Not good for high frequency switching of course, but interesting solution if you just need to switch a load on and off occasionally. – marcelm May 14 '17 at 19:01
  • Thank you all for reply...would like to add more info of my application. It is over-voltage protection switch to be switched off within 200-400 ns when a surge voltage come. Meanwhile, it work as load switch controlled by MCU. – Spark May 16 '17 at 04:25

3 Answers3

3

If the application permits, you can use an isolated power supply (12V or 15V) to create the gate driver voltage, then use an optocoupler to turn MOSFET ON or off.

You could use a photovoltaic MOSFET driver (usually used for SSR) For example the VOM1271: VOM1271

Use a automotive high side switch (it contains MOSFET and driver protection features and diagnosis), search Infineon and ST, but honestly it might be difficult to find one that works at 80V Another example: smart high side switches

winny
  • 13,064
  • 6
  • 46
  • 63
Sarah
  • 566
  • 2
  • 8
  • Thank you for reply. VOM1271 looks decent solution, only it is too slow in my application. I will expect a switch off within few hundreds nano seconds. it is a surge voltage protection switch (with a comparator). For high side switch module, it might be too expensive.. – Spark May 16 '17 at 04:37
  • The turn on is slow , but the turn off is fast relatively , (check figure 5 in the datasheet ) . It contains a turn off Jfet internally . regarding the price , it is probably expensive for such application – Sarah May 16 '17 at 16:24
3

The newish LTC7001 allows this functionality however comes in an annoying package.

ocrdu
  • 8,705
  • 21
  • 30
  • 42
John Baillie
  • 31
  • 1
0

You say that because 10A of current is to be switched, N-channel is to be preferred. I don't see why. There are quite a few P-channel MOSFEts that can fulfill your requirements, as this search on DigiKey shows.

Just as an example, take the following model, taken from that list:

enter image description here

LorenzoDonati4Ukraine-OnStrike
  • 24,350
  • 5
  • 62
  • 107
  • 1
    This, except make sure you're inside the part's SOA. – Matt Young May 13 '17 at 23:24
  • @Matt young. there are some reasons for n channel to be preferred. 1. n channel usually has lower on resistance than P channel, since the switch will be always on during load working to dissipate power (and voltage drop as well) with high working current. 2. p channel has less choices on market than n channel, usually more expensive. – Spark May 16 '17 at 04:52
  • @Spark Ok, for n.2 (but you must also weigh in the cost for your high-side N-channel mosfet gate driver). For n.1 you should consider your design target. .... – LorenzoDonati4Ukraine-OnStrike May 16 '17 at 13:36
  • @Spark ... If you want the absolutely lowest Rds(on) you'll need N-channel, but if a given P-channel MOSFET meets your goals AND it simplifies your design, you could be better off with the latter. BTW, since we are talking about switching applications, you could also parallel devices: not only this lowers equivalent Rds(on), but it increases current handling capability and helps spreading the heat dissipation among multiple devices, which in turn could make your thermal design easier. – LorenzoDonati4Ukraine-OnStrike May 16 '17 at 13:36
  • @Spark Neither of those arguments are very good. You're going to have to add complexity to just turn the thing on. Whatever pennies you might have saved are going to be added back in spades. – Matt Young May 16 '17 at 14:09
  • 1
    With 200mΩ RDSon you're going to be dissipating 20W at 10A. That's like having half a soldering iron on your board. – jcoffland Jul 23 '20 at 05:32
  • @jcoffland That MOSFET was just an example, not an optimized choice. Anyway, that Rds(on) value is a max value. If you browse the datasheet you'll see that typical value at 10A is around 140mohms for -10V Vgs drive. That makes still 14W, but the OP didn't specify which kind of switching app he is after. If the duty cycle is small the average dissipated power could be way smaller. – LorenzoDonati4Ukraine-OnStrike Jul 23 '20 at 18:14