Datasheet Jargon Translation Required (2023 Update)

For context, this comes from page 13 of the datasheet found here:

Pole is the opposite of zero. Namely, resonance pole.

Here is more technical explanation.

There's always a bit of capacitance between the op amp input and ground. If you use a big resistor from the op amp output to the input where you are applying the feedback, this can give you enough lag at a frequency where the op amp still has gain to let it oscillate at that frequency.

To prevent this put a small capacitor in parallel with the feed-back resistor. It will knock off some gain at high frequency, but it will stop it oscillating.

It's op amps 101.

A pole is basically a frequency rolloff corner, a resistor and a capacitor as a lowpass filter. The corner frequency is 1/(2*pi*R*C)

The term actually refers to LaPlace transforms and complex plane plots and stuff.

Opamps usually have an internal pole in the 10 Hz sort of range.

A pole has phase shift, 45 degrees lag at the corner frequency, creeping up towards 90 at high frequencies.

Two poles in a loop approaches 180 degrees shift, so negative feedback becomes positive and things can get unstable. Your feedback network might create a significant pole.

Got a circuit sketch? You can Spice this.

The part about supply decoupling and pcb capacitance is usually no big deal. Data sheets like to be alarmist about that stuff.

I'm just following page 16 of this document (diagram 2 at the bottom of the page). Couldn't be simpler - or so I thought til they started on about feedback poles.

Is that what the C1 'virtual capacitor' between the inverting and non-inverting inputs is all about?

It means, "Watch out for stray capacitance on the summing junction or across the feedback resistor." ;)

Cheers

Phil Hobbs

That 3 pF is internal to the opamp. PCB pads are typically a fraction of a pF.

If R1||R2 were, say, 1K, and C1 were 3 pF, the feedback pole would be tau=3 ns, corner frequency around 50 MHz, assuming pi=6. That's way above the freq where this opamp is out of gain, so it's OK. But if the resistance were 100K, there could be trouble.

Seems I needn't worry, then. I'm only going to be using this as the first stage of an amp for a dynamic microphone and the 20hz-20khz audio band is way too low for tiny stray capacitances to matter. There may be other pitfalls, but stray caps isn't something I need to worry about here.

Cursitor Doom wrote: ===================

(2) refers to the formula relating "slew rate" to full output bandwidth.

..... Phil

The issue is not so much the bandwidth of the signal you wish to amplify, but the bandwidth of the amplifier. I recall early solid state amps that would oscillate at ultrasonic frequencies if you weren't careful about how they were connected. That's a bad thing.

The op-amp inverting input has some capacitance (you can think of it as to ground) Some op-amps are worse than others (esp. those that use large mosaic of input transistors internally). Also there is stray capacitance from the 'input' end of the feedback resistor and input resistor to ground (especially if you have ground or power planes under the connections or parts).

That acts like a low-pass filter to the inverting input so the negative feedback signal lags and is attenuated at higher frequencies. That can cause overshoot or even oscillation.

In micropower circuits you typically want to have very high value feedback resistors so the effect of a given capacitance is higher. Similarly, high frequency op-amps can have plenty of gain at hundreds of MHz so even a tiny capacitance can have a significant effect. It's not typically an issue with jellybean op-amps and normal resistance values.

Poles (and zeros) refer to the numerator and denominator of the transfer function (in the complex plane), but for the purposes of a simple low-pass filter, the feedback pole is just the cutoff frequency of the filter.

H(s) = Vout(s)/Vin(s) = w0/(w0+s) so there's a pole at s = -w0 =

-1/(RC), where the denominator becomes zero.

A 1M ohm resistance and a 5pF capacitance has a pole at f =

If you have a low-power inverting amplifier with an input+stray capacitance of 5pF, with a 10M feedback resistor and a 1.11M input resistor (gain of -9, parallel resistance of 1.0M), you can parallel the 10M resistor with 0.5pF to compensate. That's a fairly small capacitance so you might have to cut and try. If the capacitance is too high you get more sluggish response than necessary, if it's too low you may get overshoot.

Much like adjusting the probe compensation on an oscilloscope to get a nice square wave from the reference square wave test point.

Post a sketch of the circuit. Words don't tell the story.

The signal you CARE about might be limited to 20 kHz, but if there's a bit of stray pickup (switchmode power supply, perhaps?) at higher frequency, and the op amp has gain and positive feedback... you'll suffer.

The gain-bandwidth product of this amp is only 5 MHz. While it is possible for the amp to oscillate at a lower frequency, this is not a particularly tricky part to use. If you don't have a bunch of stray capacitance between the inverting input and ground you should not see any problems. The words of caution in the data sheet appear to be generic and are much more important in amps with higher gain-product bandwidths.

Rick C snipped-for-privacy@gmail.com wrote in news: snipped-for-privacy@googlegroups.com:

snip

Totally off topic, BTW (my post, not yours). You're just the e-trans guy and you sparked a thought. Imagine that.

Chortle... 'gain product'

The new Harley E-Bike is nearly $10k cheaper this year and gets a 149 mile per charge range instead of only 100.

I call that a gainful product. ;-)

Esoteric speaker connection cables can make certain modern hifi amplifiers go wildly unstable. You know the sort woven by virgin mermaids from 99.999% pure copper in an oxygen free atmosphere...

High gain phono amplifiers for some expensive pickups in the days of old had a nasty habit of picking up nearby taxi radios too. You can have too much gain bandwidth in the pursuit of high fidelity.

It's on page 16 of the datasheet I linked to (diagram 2). Really couldn't be any simpler; just a case of observing the layout directions and chosing the right support components for the op-amp itself.