convert 1Mhz square wave pwm (5v and Gnd) to 1 Mhz square wave (+20v and -20v)

hello all.
im planning to convert 1mhz pwm from a microcontroller (5v and gnd) to a positive and negative (+20v and -20v) pwm (1mhz). I already did this by using a bjt transistor (bd139 and bd140), please see the pict below.



the circuit above works to convert the (+5v/gnd) pwm to (+20v/-20v) pwm with 100 kHz frequency. but when i raised the frequency to 1mhz the output was gone missing (drop to 0v). could anyone help me to get the solution of my problem? sorry if my explanation is pretty confusing, but it understanable, thank you.

Where are the Base Voltages of the output transistor ?? At least, there should be 2 diodes in series for a proper biasing.

What is the maximum output load current?

- - - Updated - - -

BigBoss said:

Where are the Base Voltages of the output transistor ?? At least, there should be 2 diodes in series for a proper biasing.

Click to expand...

That's only really needed for an analog output to minimize crossover distortion.
It's not needed for a digital out, which the TS wants.

After simulating the circuit, the problem is the saturation delay of the BJT transistors.
Below is the simulation with modifications to reduce this delay.
I added speedup capacitors at the bases along with a Schottky clamp on Q1 to prevent its saturating.
Also changed Q4 to a faster transistor.

Why do you need such a fast PWM signal?

crutschow said:

What is the maximum output load current?

- - - Updated - - -

Click to expand...

sorry i forgot to mention the current, actually the current i need is around 1-1,5 amps

thank you so much for the reply sir. i need the fast pwm signal to trigger an ultrasonic transducer for my thesis. the trigger voltage must be a pwm (positive and negative) with around +20 and -20v voltage, not to mention that the current needs is around 1 - 1,5amps.

edobp said:

sorry i forgot to mention the current, actually the current i need is around 1-1,5 amps

Click to expand...

Then you'll need larger transistors or an added buffer stage since the devices you show are only good to 0.2A maximum.

crutschow said:

Then you'll need larger transistors since the devices you show are only good to 0.2A maximum.

Click to expand...

like TIP3055 series? is it also work with the circuit you posted above sir?

Hi,

thank you so much for the reply sir. i need the fast pwm signal to trigger an ultrasonic transducer for my thesis. the trigger voltage must be a pwm (positive and negative) with around +20 and -20v voltage, not to mention that the current needs is around 1 - 1,5amps.

Click to expand...

Afaik ultrasonic transducers don´t need DC-free (symmetric) drive voltage.

Maybe it´s more simple to use a half bridge (there are ready to buy single IC solutions), or a full bridge.

Klaus

KlausST said:

Hi,



Afaik ultrasonic transducers don´t need DC-free (symmetric) drive voltage.

Maybe it´s more simple to use a half bridge (there are ready to buy single IC solutions), or a full bridge.

Klaus

Click to expand...

thank you so much for the reference sir. i will surely think about it.

edobp said:

like TIP3055 series? is it also work with the circuit you posted above sir?

Click to expand...

Not readily.
What is the actual impedance of the transducer you want to drive?
That will affect what drive circuit you need.

Here's the LTspice simulation of a circuit with a MOSFET output that will drive a 1.5A peak load at ±20V into the load @ 1MHz.
It uses a MOSFET half-bridge driver with a bootstrap high-side output to control two N-MOSFETs supplied by 40V to generate a 0-40V output.
A capacitor blocks the output offset to give ±20V to the load. If the load is a capacitive transducer then the output capacitor may not be needed.

The cross-coupled NAND gates at the input generate two non-overlapping signals for the high and low MOSFETs to avoid high feedthrough current during the switching time.

N-MOSFETs other than shown on the simulation can be used but to get proper switching times and minimize power dissipation in the MOSFETs, they must have a low gate charge, as shown in their data sheet.
It should be below 20nC, (below 10nC would be better) with an ON resistance ≦0.1Ω, and a voltage rating of 60V or higher.

Very clever, Cruts.