MOSFET Squarewave amp

Discussion in 'The Projects Forum' started by welkin87, Oct 4, 2011.

  1. welkin87

    Thread Starter Member

    Nov 18, 2010
    Due to my lack of experience in choosing MOSFET's, I am in need of some suggestions. I am working a small part of my senior design project.

    Incoming signal:
    - 10 MHz squarewave
    - 25mA (at most)

    - voltage swing roughly -3 to +3 volts (two power supplies)
    - up to 500mA constantly

    The graduate student I'm working with suggested using a MOSFET and 2 power supplies: +3V and -3V. He already has a non-inverting opamp circuit doing what he needs, but he has me looking for an alternative. My idea was to use 2 cascaded common-source amps then a final common-drain stage.

    Any suggestions on MOSFET's to use? There are so many out there and I need to find one so I can model it in LTSpice.
  2. mcasale


    Jul 18, 2011
    I assume the output is also a square-wave?

    If you use an op amp for this, make sure it's spec'ed for the high frequency - 10MHz is sorta-high. Also, selection of the dual power supplies will depend on how close to the rail(s) the output can go. It's all in the spec sheets.

    Why not use a dual supply comparator, like an LM311? Same rules apply about frequency and rail-to-rail outputs.

    A discrete MOSFET design should be pretty straight-forward.

    Go to a distributor like DigiKey or Mouser and search for whatever parameters are important.
  3. welkin87

    Thread Starter Member

    Nov 18, 2010
    Yeah, we need to output a square wave as well. 10MHz is much higher than I've ever worked with.

    I've never worked with a voltage comparator before. I've only completed my junior year in electrical engineering; just started my senior year.

    The graduate student had already designed the opamp and wants something else. That's what I'm left with doing, so another opamp isn't an option.

    Will my DC drain current need to be 500mA to get an AC current of 500mA? What kind of specs at Digikey do I need to look for on a MOSFET?
  4. mcasale


    Jul 18, 2011
    Well, a comparator is like an op amp, only different :) It's output is only HIGH or LOW, and those values vary with the supply voltages.

    For the MOSFET, you can run it like an "open-collector", except it's an "open-drain". This depends on how you connect your load. Can you post a schematic showing what you are trying to do and what you are driving?

    For switching applications, which are not linear, the vendor may specify switching times, or maybe something else. You will want to look at FAST or Ultra-fast FETs because different vendors call them different things.

    Same with the comparator - I don't know if an LM311 will work at 10MHz, but this is an old part. There's probably something newer that will work.

    Where are you attending school?
  5. mcasale


    Jul 18, 2011
    If you decide to go discrete, here's a useful web page for calculating the bias points on a "cascode" BJT amplifier. You don't need to use FETs if you find BJTs easier to understand.

    Years ago, I used a circuit like this as a buffer amplifier for a 40MHz oscillator in a real product. So if you get decent high speed transistors, it should work fine. You can then clamp the output signal to whatever you want with zeners, or whatever.

    The output stage can still be a FET depending on what you are driving.

    Good luck.
  6. oldtech33709

    New Member

    Sep 24, 2011
    A full bridge converter could do this with one rail, I've never tried it at that high a frequency though.
  7. welkin87

    Thread Starter Member

    Nov 18, 2010
    Ok, met with graduate student today and he clarified some things. Pretty much, I need to take a 0 to +3V square wave signal and make it a -3V to +3v square wave signal. I need to maintain 1 volt per nanosecond (1 V/ns) on the rising edge. The reason he suggested the MOSFET is that they have an infinite current gain. I need 500 mA.
  8. mcasale


    Jul 18, 2011
    Just getting back to this. Sorry. I still don't know how you are driving the load, 500mA. Obviously, you need to do some level shifting, which a comparator will do. Then you can drive the gate of an n-channel MOSFET. 1v/nS is pretty fast. You will need to drive the gate pretty hard and fast. Note that MOSFETs do have input & output capacitances. Is there any spec on delay time from input to output, or is it just the slope that's important? A schematic would be helpful.
  9. Ron H

    AAC Fanatic!

    Apr 14, 2005
    MOSFETs have ON resistance. Big MOSFETs have very low ON resistance, but they are very difficult to drive at 10MHz, due to the large gate charge/capacitance. Small MOSFETs are easier to drive. IMHO, If you need ±500mA at ±3V, you will need supply voltages that are higher than ±3V.
  10. welkin87

    Thread Starter Member

    Nov 18, 2010
    Thanks for help guys, I really appreciate it.

    I don't have a schematic for the MOSFET amp I'm considering but I do have one for the opamp circuit the graduate student is currently using. I will upload it later today. The opamp is a TI OPA 2674. It has the exact specs that are needed for this project. That's why I have a feeling I'm not gonna be able to find any other better alternatives. Anything in the BJT department that might work? I'm searching as we speak.
  11. gootee

    Senior Member

    Apr 24, 2007
    It seems like you should be able to do it with an opamp amplifier (maybe the one he already has) that has a discrete high-speed power-booster amplifier inside its feedback loop, or after it.

    You might be able to get some ideas from application notes AN-21 and AN-18, at One of those has a 3000V/us amplifier (Fig 12 in AN-21) that should do what you need and one has one that does in excess of 1000v/us (Fig 3 in AN-18) that is almost there.

    The LT1363/64/65 are 1000V/us single/dual/quad opamps, from Linear Technology Corp (LTC), It seems like one or more of those and a discrete buffer/follower would do the trick (or help), too.

    [Edit:] I suggest also looking at AN-94 at Among other good tips, it mentions that the LT1818 (single) and LT1819 (dual) amplifiers can do 2500 V/us, with 70 mA output currents (Maybe you could just run eight or more of them in parallel?). Also make sure that you take a look at Appendixes B and C.

    You will want to try to also model the parasitics of your conductors and components, in spice.

    Here are links to over 20,000 spice models of components: List/ Part List/



    You'll probably need the following, while trying to get the thing to work with 1 V/ns slew rates.

    (This is pretty slick.)

    GOT RINGING? I re-discovered a great way to easily determine the optimal value of R
    (termination), or R and series C (snubber), to damp it out.

    This is a very simple method for determining the parasitic capacitance and the
    parasitic inductance of a resonance in a circuit or a transmission line or PCB trace,
    which also gives you the characteristic impedance for the resonance, which is
    everything you need to know to damp it, optimally.

    This assumes that you have a ringing condition, already, such as might occur on a
    digital buss or a transmission line or PCB trace, or in a switch-mode power supply or
    even an AC-to-DC transformer/rectifier circuit. (If you don't have ringing and just
    want to determine some of these parameters, I guess you could try hitting your
    circuit with a pulse train until it rings.)

    1. Measure the frequency of the spike resonance (ringing) voltage, using an
    oscilloscope. (You might need a very-low-capacitance probe or probing technique.)

    2. Add a shunt capacitor across whatever is ringing (or in parallel with the
    parasitic capacitance) and adjust the value of this capacitor until the frequency of
    the spike resonance/ringing is reduced by a factor of two. The value of this resulting
    capacitor will be three times the value of the parasitic capacitance that is creating
    the voltage spikes or ringing.

    3. Because the parasitic capacitance is now known, the parasitic inductance can be
    determined using the formula: L = 1 / [(2 x Pi x F)² x C] where F = (original) resonant
    frequency and C = parasitic capacitance

    4. Now that both the parasitic capacitance and inductance are known, the
    characteristic impedance of the resonance can be determined using the following
    Z = SQRT(L/C) where L = parasitic inductance and C = parasitic capacitance

    5. The resistor for the terminator or for the RC snubber circuit should be sized to
    the value of the characteristic impedance, and the capacitor should be sized between
    four and ten times the parasitic capacitance. The use of larger (than 4X) capacitors
    slightly reduces the voltage overshoot at the expense of greater power dissipation in
    the resistor.

    NOTE: An R alone would work to prevent or damp-out the ringing (or reflections, as the case
    may be). But if power dissipation in the R would then be too high, a C is added in
    series with the R.


    Last edited: Oct 17, 2011