# How do I choose the correct op amp to amplify a high frequency sine wave (>500kHz)

I am working with ultrasonic transducers (12mm diameter, 500kHz). With my present physical setup, the transmitter is excited with a 3.3V p-p sine wave. The receiver is showing a nice clean sine wave with around 20mV p-p.

I want to amplify this output signal to around 10x the amplitude (~200mV) or greater. I have investigated op amps for this purpose, and have learned that op amps have "slew rates" and "bandwidths". I could not find any clear reference material on how to choose the correct op amp parameters for my specific application.

Can anyone provide some pointers on this?

Very roughly, the bandwidth of the closed-loop circuit will go down in proportion to the gain of the circuit. Put another way, many op-amps can be characterized by a constant "gain-bandwidth product", meaning gain x bandwidth is some constant value, with gain trading off against bandwidth as you adjust the feedback in your design.

So if an op-amp advertises a bandwidth of 1 MHz at gain of 1, then you can expect a bandwidth of 100 kHz at gain of 10.

Or, put in a way that's more useful to you, you want to look for an op-amp with bandwidth greater than 5 MHz (for gain-of-one configuration). You probably want to add some margin, because the specified bandwidth is for a 3-db drop from the low-frequency gain. So to get full gain at 500 kHz, you will need an op-amp with gain-of-one bandwidth somewhat above 5 MHz. A 10 or 20 MHz spec might be a good place to start, for a minimum.

But 100 MHz devices (and higher) are readily available, and would allow you to control the bandwidth with external components, so you might prefer that route.

The constant gain x bandwidth rule goes out the window if you start talking about current feedback op-amps. But for 500 kHz, you shouldn't need to go there.

As for slew rate, you simply need to calculate what is the slew rate of your desired output signal (take the derivative of the sinusoid, and work out the actual rate of slope in V/s), and choose an op-amp that offers a higher slew rate than that.

Finally, if the performance matters to you, simulate your circuit and check the AC performance (for bandwidth) and transient response (for slew rate limitations) before finalizing your design.

You need to pay attention to two things- 1) the GAIN BANDWIDTH PRODUCT, and 2) the SLEW RATE.

Gain-bandwith product is a first-order approximation that the product of the gain and bandwidth is near a constant. Thus, for your 500kHz application, with a gain of 10, you need a gain-bandwidth product (assuming one amplification step) of 5MHz. If you want to split this into two gain stages (which MIGHT be feasible) of around 3 each, you'd be looking at around 1.5 MHz. I tend to add about 20% to the spec as a habit.

Next, if you examine slew rate, you need 200mV out at 500kHz. Thus, the peak rate of change 0f a 200mV sinusoid at 500kHz is 0.2V*2*pi*500k, or about 630KV/sec, or 630V/ms. You need a slew rate bigger than this

Gain bandwidth product, slew rate and noise are all important. I plugged these (and a few other) values into Linear Technologies search engine and up came three offerings: -

The picture is more to demonstrate the website and how easy it is to focus in on what you might want. I preselected the "low noise" op-amps and asked for: -

GBW =>10MHz Slew rate =>5V/us Isupply <= 5mA Vsupply =>5V

I also added a few more touches like selected rail-to-rail output and single supply just to narrow the field down. I think the top choice looks good the LT6202. This should take you to the same selection page.

LT6202 is worth a consideration AND I'd also take a look at Analog Devices website and all the usual ones like TI and Maxim.

• Why is low noise important? Especially working at one frequency, where you can substantially limit the bandwidth at the low-freq end, and with such a modest gain. Even if the noise of the op amp were 50nV/rtHz, leaving the bandwidth wide open at 500kHz, and a gain of 10, the noise would come in at only 0.35mV. For a signal size on the output at 0.2V, should be fine for ultrasound purposes. Jul 29, 2013 at 18:02
• @ScottSeidman yes it seems like I'm doom mongering bringing noise into the equation but consider that the signal actually needed may be substantially less than 20mV it's best to err with caution. I don't know how much the phase integrity of the signal needs to be maintained w.r.t. the transmit signal either so a wideband amplifier without filtering may be required. It's easier to do too. Jul 29, 2013 at 19:02
• Fair enough. Whether low noise is called for or not, noise is certainly something one should habitually keep an eye on when speccing op-amps, unless one really loves surprises! Jul 29, 2013 at 19:57