is there a distortion/bandwidth correlation?

[SArepairman]

I was wondering if there are any "general" rules regarding distortions relationship with gain bandwidth product. If an amplifier has a BW of 10,000 Hz at its gain setting, will a signal of 4000 Hz be distorted less, then say a signal of 8000 Hz, because of its proximity to the end of the gain "linear region." I believe I read this somewhere but I don't remember if it was for a specific part or for op-amps in general.

Is there any good reason (assuming the bandwidth is TESTED, I.e. the reason is not to part variations) as to why you should not operate near the upper limit of your amplifier?

And yes, I know that 3dB is 70% of your signal, so your gain will not be as on paper.

[Omicron]

In principle, yes. The higher the frequency the higher the distortion. This is true for all amplifiers that use negative feedback. The reason is that the factor by which the distortion of the amplifier is suppressed by the negative feedback action depends on the open loop gain. As the GBW expresses, this open loop gain drops as the frequency increases. And hence the feedback loop becomes less and less effective in reducing distortion.

[SArepairman]

So how would I design around this fact?

I'm sure there is some "fudgeineering" technique, like "don't go more then 30% of the bandwidth" or something like this.. Do you know of any curves?

[Omicron]

Opamps that are marketed for low distortion, e.g. for audio purposes, usually have some graphs in their datasheet that show distortion versus frequency for a certain gain configuration. E.g. look at the datasheet for the LT1115 on page 4:

http://cds.linear.com/docs/en/datasheet/lt1115fa.pdf

[SArepairman]

Opamps that are marketed for low distortion, e.g. for audio purposes, usually have some graphs in their datasheet that show distortion versus frequency for a certain gain configuration. E.g. look at the datasheet for the LT1115 on page 4:

http://cds.linear.com/docs/en/datasheet/lt1115fa.pdf

Well darn the op-amp I was looking at did not hav that in the datasheet

[SArepairman]

If gain specific information is not included what can you do?

can the Av=10,100 etc curves in the datasheet you linked be computer generated from the av=1?

how can they be "inferred"?

[Bored@Work]

If gain specific information is not included what can you do?

Measuring it.

[vk6zgo]

If you are looking at THD,the harmonics of your 8kHz test frequency may be reduced in level by the roll off in level at the top end of the amplifier's frequency response,masking the effect referred to.

[T3sl4co1l]

Distortion, like frequency response, is enhanced by applying negative feedback.

Approaching the cutoff frequency, loop gain is dropping, so the voltage gain no longer exactly matches the resistor divider ratio; distortion rises in a reciprocal fashion.

Normally, distortion is spec'd at unity gain, where it is lowest (and frequency response is highest, being near GBW).  It isn't always measured at unity gain, especially for amps where the distortion is vanishingly small.  A THD figure of 0.00003% or something like that is usually measured at higher gain, 100 or 1000, where the THD is correspondingly higher.

The ratio is not exactly the gain ratio, but of the form 1 / (1 + G0/G), where G0 is the amp's DC gain and G is the gain you're measuring at.  Op-amps have very high G0, so it looks like the gain ratio.  A lower gain amp, like a single transistor stage, makes this distinction more apparent.

Tim