AD826 Op Amp: Signal Distortion Appears After Adding 5th Amplifier Stage at 150 kHz

I'm currently having issues with an AD826 amplifier circuit I'm breadboarding. Attached is the circuit diagram of the circuit I've built. The circuit is decoupled with 1 uF and 0.01 uF capacitors at each stage. The feedback resistance should not exceed 1 kΩ, and thus I've used 1 kΩ feedback resistors in each stage. Each stage works correctly when hooked up individually. My input is a 50 Ω function generator output set to 150 kHz and about 1 mVp-p. With this simple circuit, using 4 stages, I'm fine. The signal looks clean and it's amplified correctly. However, when I add a 5th stage, I get some distortion that I can't seem to debug. Both responses are shown in attachments. I'm wondering if I'm missing something very fundamental about this chip's operation.

I've tried switching the AD826 out with LF353 (slightly lower GBP and not quite desirable for this application). The 5 stage circuit works just fine, but if I add another I get a similar distortion. I've also tried changing power supplies, function generators, etc. I am convinced that there is something about these chips that I'm just not grasping.

Any and all advice is much appreciated.

Have you probed the supply lines to see if this distortion may be coming from there? I know that's a long shot, but worth considering.

Another long shot: Is the self-resonate frequency of the 1uF bypass capacitors higher than 80kHz?

What if you use a function generator and feed a 4.6V signal into the input of one AD826N, do you still get the distortion (I'm assuming that the output of the 4th stage is 4.6V)? Try this with an AD826N where it is the only one on the circuitboard/protoboard/?

Also, is the output of the 5th stage open? I wonder if there is some minimum loading requirement -- I didn't see that in the datasheet, but it might be assumed that the load will always be1K or less.

And, have you followed these guidelines (from the data sheet):

bq). Circuit Board Layout
Input and output runs should be laid out so as to physically
isolate them from remaining runs. In addition, the feedback
resistor of each amplifier should be placed away from the
feedback resistor of the other amplifier, since this greatly
reduces inter-amp coupling.
Choosing Feedback and Gain Resistors
In order to prevent the stray capacitance present at each
amplifier’s summing junction from limiting its performance,
the feedback resistors should be ≤1 kΩ. *Since the summing
junction capacitance may cause peaking, a small capacitor
(1 pF–5pF) maybe paralleled with R
F*
to neutralize this effect.
Finally, sockets should be avoided, because of their tendency to
increase interlead capacitance.
Power Supply Considerations
To ensure the proper operation of the AD826, connect the
positive supply before the negative supply. Also, proper power
supply decoupling is critical to preserve the integrity of high
frequency signals. In carefully laid out designs, decoupling
capacitors should be placed in close proximity to the supply
pins, while their lead lengths should be kept to a minimum.
These measures greatly reduce undesired inductive effects on
the amplifier’s response.
Though two 0.1 μF capacitors will typically be effective in
decoupling the supplies, several capacitors of different values
can be paralleled to cover a wider frequency range.

Also...

Notice that, in Figure 14, in the datasheet, the Large Signal Frequency Response graph has curves for 1K and 150Ω, lending more credence to the idea that this device requires a load on the output (and apparently at 1K or less--down to a load that will draw no more than 50ma).

Also...

If you change R2 from 330Ω tp 1K, does the distortion go away? Except, perhaps, for "Active Filter," the list of Applications for this device, are pretty much Unity-Gain.

Is the 5th stage exactly the same as the other four? What is its input impedance? With all stages connected work back from the 5th stage towards the 1st to see where the distortion originates - for example if 5th o/p has distortion, does the 5th stage's input (which is also the 4th stage's o/p) have distortion - if yes then check if the 4th stage's input. If there is no distortion, it means that the 4th stage is the one with the problem - especially its o/p. Disconnect 5th stage and re-check 4th stage - if distortion goes away then the problem is either the 4th stage opamp or something related to loading.

As an update and response to both Steve and Rohit,

*The distortion occurs only after adding the 5th stage (as shown in the figure). Once it's added I can track it back all the way to the input stage.
*The supply rails look clean
*Loading the amps made the problem worse
*Looked into the SRF of the caps, shouldn't be a problem
*I have tried placing 3 pF caps across my feedback resistors. No change
*R2 changed to 1k allows four more stages before encountering the same problem

I believe, similar to what Steve pointed out, that the layout is the problem. Of course, I'm breadboarding this circuit so I've got parasitic capacitance of the breadboard going all the way back to the original input and my decoupling caps have quite long leads. It appears that my problem is some sort of feedback/coupling though. The same sort of distortion also occurred when I loaded any stage of the amplifier significantly. Adding 1 mF caps on the breadboard as local power storage (between power and ground) in addition to the 1 uF and 0.01 uF ones I already had significantly increased the number of stages I could add before reaching the same problem, which leads me to believe layout and power line impedance are my main problems here. I'm going to whip up a PCB and see how that works out. Thanks again for your help. I don't have anyone to bounce ideas off of here so I can quickly think myself into a corner.

Layout problem, yes. Also could be problems with grounding...connect each point in the ckt that's supposed to be ground together with thick wire plus make sure that the main ground return (to the supply's 0V point) is as thick as possible.

Also, add 100nF supply bypass caps close to the ICs' supply pins - alternatively just place the cap across the supply pins with leads as short as possible.

Yes, very likely parasitic elements in the breadboard. The datasheet cautions, even, against using sockets, for this very reason.

distortion of af signal only occurs at the last three stages of the op amp. the first stage overdrive the input of the second
stage and to the third and to the fourth stage. my suggestion is to decreased the output level of the first stage not to overload the second stage and so on until the 4th stage