# How do I amplify a square wave pulse of 4 to 8 periods at 3.5 MHz from +-2V to +-60V or more?

I am working with ultrasound sensors optimized for 3.5 MHz signals, and I want to test these using my PicoScope 5000 with integrated signal generator. The PicoScope is able to generate arbitrary square waves with an amplitude of max +-2V, and I would like some circuit to be placed between the PicoScope output and sensor which amplifies the square wave to +-60V or more.

The ultrasound sensor has an impedance of approximately 41 ohm, and is connected in parallel with a TX810 evaluation board used as a T/R switch.

The square wave pulse needs to be from 4 to 8 periods long, and it will be repeated approximately every 70ms.

I've been experimenting with Microchips MD1213DB1 evaluation board, but find that it is not a perfect match as it requires a phase shifted square wave and an Output enable signal in addition to the original square wave. Since the PicoScope only has one output (being the square wave) this became difficult to implement. Also, I just fried my MD1213DB1 board and I'm hoping that I don't have to order a new one.

Is there a cheaper and easier way to achieve the pulse amplification?

Edit:

I want to test how the ultrasound sensor performs using square waves compared to gaussian sinusoidal waves. I have already tested using sinusoidal waves at +-60V, by amplifying them through ADA4870 and PA107 evaluation boards. My hope is that transmitting square waves will result in an equally good result, enabling us to design a simpler pulser circuit for the final product instead of constructing the expensive sinusoidal variant using DAC and amplifiers.

I was thinking if there were some digital / pulser solution in which the original square wave might be used to open / close some MOSFET transistors, consequentially creating a higher voltage square wave on their output.

• No. Driving pulses that are 1-3 usec long at +/-60V "or more" into 41 ohms is neither cheap nor easy. Also, you need to specify how "square" the pulses have to be after amplification. What is the rise/fall time requirement? What is the max allowed amplifier delay from input to output? What does "60V or more" mean? How much capacitance is the amp driving into? What power supplies are available? If you're using the MD1213DB1 then you're getting off light. Doing this custom is complex and difficult. May 25, 2020 at 15:29

You want to drive a capacitive transducer at 3.5MHz with 120Vp-p, which is a fairly tall order. The peak power is rather high (3A peak at 60V).

You can purchase high-voltage power op-amps, some of which have example circuits that show driving heavily capacitive loads. I've done a discrete booster amplifier for driving piezos, but it's tricky to get it to work in a stable manner with that kind of heavy capacitive load.

• Interesting. Could you point me in the direction of such an op.amp and example circuit? May 26, 2020 at 7:51
• I doubt that there is any opamp at these voltage levels which could simultaneously deliver that power, dissipate the heat and have a flat gain up to many, many harmonics of 3.5MHz. In every case a dedicated power transistor is needed in the end stage with some massive thermal interface. May 26, 2020 at 21:45
• @HorrorVacui Apex has ~100W amplifiers that can handle up to a couple hundred V. The problem is more the GBW product, which either a resonant configuration or a bridging amplifier could help with. May 26, 2020 at 21:52
• @SpehroPefhany. Exactly. These frequencies, especially if square wave is required are more of RF applications, where the "simple" switching power transistors are not fast enough. Plus the required 30dB gain. It has to be possible, but it is not a jelly-bean component. May 26, 2020 at 22:01
• @HorrorVacui Oh, definitely not, they're \$. But cheap compared to engineering time (for a one-off or two). May 26, 2020 at 22:01

Instead of an analog amplifier, consider a digital approach, in which a 60V source is switched on and off by the lower voltage square wave.

• This is what I was thinking. In order to generate a generate a high voltage square wave pulse-train I could use the original low-voltage square waves to toggle MOSFETS. Could this be an option? Is there a schematic I could use as reference for such a circuit? May 26, 2020 at 7:00
• This is more of a comment than an answer May 27, 2020 at 15:22

You can increase the voltage with a transformer, or a chain of transformers. At 3.5MHz there has to be some available off-the-shelf component. Though you could have a problem with the power. 60 V on 41 Ω is equivalent to 88 W.

Another option is to use transistor amplifiers. There are many devices which withstand 120V.

• I want to test how the ultrasound sensor works using square waves in comparison to sinusodial waves. I'm afraid the inductance introduced by the transformers will disturb the pulse shape. May 26, 2020 at 7:49
• There are signal transformers close up to 1GHz. But if its not an option, go for a transistor amplifier. The mobil base stations and radio transmitters also use solid state transmitters. 3.5MHz and 120V is well within the range of possibility. May 26, 2020 at 21:49