# Required GBW of an op-amp

So if my Vin signal is of 50kHz and I want a gain of 50, would this mean that the GBW of the op-amp has to be at least 2.5Mhz? What happens if its below 2.5Mhz

Study the blue line (a half-decent op-amp's open-loop gain) in the picture below: -

• At 10 MHz the gain is unity (0 dB) hence the GBP is 10,000,000
• At 10 kHz the gain is 60 dB and, a gain of 60 dB is 1,000 hence, the GBP is 10,000,000
• At 10 Hz the gain is 120 dB and, a gain of 120 dB is 1,000,000 hence, the GBP is 10,000,000

So, this op-amp has a gain-band-width-product of 10,000,000 and if we wanted a closed-loop gain of 34 dB (a gain of 50) we can draw the purple line and see where it intersects the blue line. In this example, we can see that it is 200 kHz and, not surprisingly that means a GBP of 200,000 x 50 = 10,000,000.

If we expected a band-width of 1 MHz we wouldn't get it with a gain of 50.

With a 2.5MHz opamp, gain of 50 is already starting to drop off at 50 kHz, and even more drop above 50 kHz. It acts very similar to a low-pass filter.

We often use opamps to simplify cascaded stages in a design. For example, an opamp circuit might exhibit high input impedance and low output impedance so that a particular gain stage does not load a previous stage, and is not loaded by a following stage. We expect that the gain of our amplifier stage is set by the ratio of a few resistors.
At frequencies approaching the GBW limit, a circuit using an opamp has less-ideal input and output impedance.... for example, instead of near-zero output impedance, output Z may rise to hundreds-of-ohms. Stage-to-stage loading can suppress gain even more than the simple GBW calculation of the opamp alone. Having excess opamp GBW may be required to have flat gain up to 50 kHz.

GBW requirement of at least 2.5 MHz is correct. However it does not leave any margin.

If the op-amp has smaller GBW, it can't provide the required gain at 50 kHz.

• It's got nothing to do with slew rate. Consider this: If the signal amplitude were reduced to 1%, the GBP would still restrict the amplification at high frequencies whereas slew rate problems would largely disappear. Please fix your answer because slew rate limitations and GBP limitations are not the same. Jul 13 '20 at 15:44

If you want to preserve the INFORMATION in the waveform, then you need FLAT GAIN and FLAT PHASE responses.

You need to establish budgets for those parameters.

Then design a signal chain that respect the budgets.

However, if you just want wiggles out of your amplifier, any opamp will do.

I recall reading on analog phonograph vinyl playback PreAmplifiers. The various home experimentors (this was diyAudio.com "SIMPLISTIC NJFET RIAA" thread) would write of:

• precisely matched Left_to_Right audio channel gain, within 0.1dB across the audio band

• matched Resistors and Capacitors, matched at the 10pF out of 50,000 pF value, or 10/50,000 = 1/5,000 or 0.02% matching, for PHASE matching

• nanoVolt trash and spikes and random noise (thermal) from the special SHUNT REGULATOR power supply

• placing the Power Supply at least a meter away

and suddenly, as the human ear/brain had a few seconds to listen, the SOUND STAGE would appear and the music instruments would be detectably physically placed in the concert floor.

Now how do you write a budget for that? From decades of experience, with high focus on precision GAIN and PHASE, etc.

=============================

No matter how much you spend on speakers, the electronics have to be matched in Gain and Phase. The "diyAudio.com" people had 0.02% budgets for matching.

They also had minimal electronic ECHO, where the RIAA delays with 50Hz pole and thus 3 millisecond delays, leak from the Output (to the Power Amplifier) across the PCB 2_D structure and back into Ground paths around the First Input JFET, at the [ 0.1 milliamp * 1 milliohm == 0.1 microVolt] 100 nanoVolt level.

The input from a Moving Coil cartridge may have been 200 microVolts at strong signals.

The ratio of 200uV/100nanoVOlts is only 2,000:1 (-66dB), and ECHOES occur at the 3milliSecond level. The PCB must be designed, to get -100 or -120dB echoes. You can do that. But requires a budget. And 2_D thinking.

YOu asked this before. Which part was not understood? The loss of gain or the tolerance?

If you use a 1MHz GBW Op Amp on the above 2.5MHz GBW demand your gain is reduced from 50 to 20 with some wide tolerance. (Nominal only)

If you use the JFET TL062 with A typical GBW of 1MHz it will reduce further to 90% of the manufacturing tolerance @ 70’C so is very unreliable for gain tolerance and SHOULD NEVER BE RELIED ON FOR ACCURACY.

Depending on your gain or phase shift requirements you should choose a lower gain by a factor of 2 to 10 or increase your choice of IC to 5 to 10MHz GBW for accuracy. Phase shift starts around 1 decade below the break point.

An example of a CMOS 10MHz GBW Op AMP is the LF356.

https://www.digikey.ca/en/products/detail/texas-instruments/LF356N-NOPB/6135