Gain bandwidth product is not what the datasheet says.

[Mercator]

Hi,
Originally, I wanted to design a single high-gain amplifier (Av = 100) for a 40kHz sinusoidal signal. So I started simulating the design in LTSpice using LT's built-in components to save some time. However, the gain at 40kHz was about 70% lower than I expected. So I stripped down the circuit until I was left with a voltage follower with unity gain (see attached image).

The datasheet says that the gain bandwidth product (GBWP) for the LTC6240 is 18Mhz. However, the bode plot of LTSpice says that it is only about 130kHz. How can that be? Or did I misunderstand the concept of GBWP?

Thanks in advance.

[Kalvin]

What happens if you give the op amp some negative power supply voltage, like -1V, instead of 0V.

Edit: You may also want to increase the AC voltage to 100mV to get started, then reduce it gradually to 50uV and observe what happens.

[Mercator]

What happens if you give the op amp some negative power supply voltage, like -1V, instead of 0V.

Giving it positive 2.5V and -2.5V (according to the datasheet it needs 5V from V- to V+) leads to a bandwidth of 8Mhz. Much better than before. Thanks  . However not 18Mhz as claimed.

Edit: You may also want to increase the AC voltage to 100mV to get started, then reduce it gradually to 50uV and observe what happens.

Same result as before. Sorry. 

[Chalcogenide]

The GBWP does not necessarily corresponds to the bandwith of the opamp when used as a buffer, it's just the product of the DC gain times the frequency of the dominant pole. If you look on the datasheet of the LTC6240 you can see that there is an additional pole at around 1MHz, followed by a zero right around 5 MHz. The gain crosses 0dB at around 5 MHz or so, and the bump you see in the frequency response around 2.5 MHz is probably a slight peaking due to the phase margin being less than 90°.

[StillTrying]

I've not looked at the data sheet, just drawn it.
GBW seems about 12MHz, at x1, and 25MHz at x100. ~18M on average.

[Mercator]

I've not looked at the data sheet, just drawn it.
GBW seems about 12MHz, at x1, and 25MHz at x100. ~18M on average.

You're right! Just went back to Av = 100 and I got a bandwidth of about 175kHz, i.e. GBWP is about 17.5 Mhz. So close enough  .

Thanks to all of you!      

[StillTrying]

Av = 100 and I got a bandwidth of about 175kHz

I seem to remember it was about 250kHz at x100, which is just about enough to output at 2Vp-p 50kHz square wave.

47R and 4K7 gave about the best BW at x100.