Distortion Output LM7171 Opamp Not According to Datasheet

[Mechatrommer]

LM7171 is 2X gain stable VFB opamp up to 100MHz+ bandwidth (https://www.ti.com/lit/ds/symlink/lm7171.pdf)... but when i'm feeding 4Vpp signal (or larger) from FG, the output signal got distorted starting as low as 30MHz (attached sequence, yellow trace is opamp's output 2X gain and terminated with 50 ohm, blue trace is -ve (-12V) rail AC coupled to see PSU oscillation), input power is ±12V. input (FG) signal is checked to be ok up to 60MHz (UTG962) Any experience as to why and its likely reason? my calculation indicates slew rate is not exceeded at that signal amplitude and frequency. i cover the top opamp IC with flux to help spread the heat because opamp can get really hot when output got distorted. schematics follows, 1uF elec, and 100nF + 10nF ceramics installed as in attached schematics, PCB layout also provided, 2 layers 1.6mm thickness from jlcpcb. this is really basic circuit but difficult to get right. any idea? thanks.

[MasterT]

Have you seem Figure 51. Harmonic Distortion vs. Frequency

[Mechatrommer]

yes, extrapolating it, at 60MHz its about -50dB 2nd harmonics, thats very negligible (should be invisible) compared to what i got. 3rd harmonic is wobbly but maybe at -40dB? thats still very little. fig 48 shows undistorted 1% vs Vpp, that may explains it but still not really obvious, maybe later i need calculate THD of signal that i got, but from visual, its very unacceptable and THD should be several magnitude higher than -40dB. are you saying this is normal? the 200MHz BW only for small signal? attached is another output signal how bad it is... cheers.

[MasterT]

Right, fig.48 explains this phenomenon better. I think it's normal, GBW doesn't apply to output BJT driver, especially for 20+ years old IC.
 I remember 3 MHz power BJT miserably failed to deliver required power in the audio amp even at 20 kHz

[Mechatrommer]

Thanks.. so i guess this is normal, its just i overestimated or misunderstood what a gain bandwidth is.. ie mainly for small signal only, i thought lm7171 is a new opamp.. some newer opamps will specify its 'full power' or 'large signal' bandwidth.. i guess this is what i'm looking after, which is not available in lm7171's ds...

[MasterT]

Alternative:

[Mechatrommer]

Alternative:

coincidentally, i retested THS3095 (sibling of 3091) in my stock yesterday to work as expected on the same pcb and setup. another candidates are THS3001 and THS3491. but those CFB opamp are multiple times pricey than VFB like LM7171, i'm just testing for another possible cheaper and cooler temperature options. but clearly $0.50 LM7171 cant do the job, i was just misleaded. currently testing another LM7171 as controller for a class AB power darlingtons, previously tested with TL071 to work up to 1MHz BW... the higher BW LM7171 introduces some oscillation (phase shift) on power stage that needed to work on and investigate at 3-6MHz, but thats another story... thanks mate.

[MasterT]

Out of curiosity bought buf634 and lmh6321, to see if I can improve drive strength capability of LM7171.
None of them can work at 20 MHz 20 Vp-p w/o distortion, LMH6321 hardly could get 5 MHz 10 Vp-p. Just FYI.
 

[Kleinstein]

The circuit in the photo does not match the plan. Especially it seems to miss the termination at the output. The cable capacitance can really upset the amplifier.
P.s. just noted, just the resistor lables are mixed up - so the terminaion is there.


Another point the the PCB layout: the ground layout look really strange with the cuts on the bottom to destroy the ground plane.  A ground plane with sich a cut is no longer a good ground and a possible source of all sordes of trouble with a fast amplifier.

One can not really expect a 100-200 MHz GBW OP-amp to get low THD at 30 MHz with a gain of 2. There would be only a loop gain of seom 2-4 left to compensate for linearity errors of the output stage.

[Mechatrommer]

The circuit in the photo does not match the plan. Especially it seems to miss the termination at the output. The cable capacitance can really upset the amplifier.
P.s. just noted, just the resistor lables are mixed up - so the terminaion is there.

if you referred to OP, both terminated and not terminated are tried, none can give good result, and coax is short about 5cm length, so i guess not quite an issue in term of reflection and capacitance at few tens MHz in nonterminated setup, ymmv.

Another point the the PCB layout: the ground layout look really strange with the cuts on the bottom to destroy the ground plane.  A ground plane with sich a cut is no longer a good ground and a possible source of all sordes of trouble with a fast amplifier.

cut ground pour is the recommended way by almost all high speed CFB opamp (look at major TI CFB's datasheets) this is to reduce stray capacitance introduced into feedback line and hence screwing stability.