Cascading Op-Amps suffer from oscillation

[bleedingwolf]

Hi All, I am now using AD8099 for a low noise preamplifier, which is followed by a low pass filter for filtering out low frequency noise and DC created by the unbalance impedences between V+ and V-. For single stage very thing is fine. However when I cascade two stage, oscillation happens. Please see circuit in attachment. Anyone tell me what cause this problem? Thank you.

 

[crutschow]

You are likely getting feedback from the output of the second amp to the input of the first. Are they well separated in your layout?

What is the frequency of the oscillation?

 

[audioguru]

I betcha the circuit was made on a breadboard with long wires all over the place.
It must be made on a double-sided pcb with one side as a ground plane.
The output must be kept away from the input.

 

[MRCecil]

Unless you have gotten evaluation board(s) from AD for this amp or made your own, per their AN-720 ap note, it's likely your results via sim or alternate real-world layout is the oscillation issue coupled with a few other items. My advice is to read over the datasheet very carefully for this special use, very touchy device before going further.

Also of note, if you look at the datasheet, you'll notice discrepancies in your decoupling of +Vs and -Vs, the input resistance on the non-inverting input, the labels on pins 5 & 8 inferring they are of the same net (tied) and the capacitive loading on the output. Take a look at page 18, upper right, for the recommended configuration for either the CSP or SOIC package for the desired programmed gain of 20 as shown in your schematic. With gains like this and slew rates up to 1350 V/usec, singing amps are easy to generate, and require due caution.

Good Luck with your project.

 

[bleedingwolf]

The Oscillation frequency is 100Mhz with 3V p-p voltage. Do you mean there are coupling between output of the second amp and the input of the first? They are on the same board and far from each other. Coaxis cables are used at the input and output of each amplifer. Is it possible the coupling is coming from the GND layer? Or other facter?

Thanks for you answer.

You are likely getting feedback from the output of the second amp to the input of the first. Are they well separated in your layout?

What is the frequency of the oscillation?

 

[bleedingwolf]

I betcha the circuit was made on a breadboard with long wires all over the place.
It must be made on a double-sided pcb with one side as a ground plane.
The output must be kept away from the input.

The circuit was made on a double sided pcb and I made the ground fill all over the two sides of the board beyond the circuit itself. Is this a problem

Thanks

 

[bleedingwolf]

Hi MRCecil,
Thanks for the feedback. I made the evaluation board by my own. And like you said I made two changes on the provided circuit diagram for application reasons. One is the input resistance on the non-inverting input and the another causing the capacitive loading on the output. Is these the reason causing the oscillation?

Bo Hu

Unless you have gotten evaluation board(s) from AD for this amp or made your own, per their AN-720 ap note, it's likely your results via sim or alternate real-world layout is the oscillation issue coupled with a few other items. My advice is to read over the datasheet very carefully for this special use, very touchy device before going further.

Also of note, if you look at the datasheet, you'll notice discrepancies in your decoupling of +Vs and -Vs, the input resistance on the non-inverting input, the labels on pins 5 & 8 inferring they are of the same net (tied) and the capacitive loading on the output. Take a look at page 18, upper right, for the recommended configuration for either the CSP or SOIC package for the desired programmed gain of 20 as shown in your schematic. With gains like this and slew rates up to 1350 V/usec, singing amps are easy to generate, and require due caution.

Good Luck with your project.

 

[audioguru]

A capacitive load on an opamp can cause oscillation. The small amount of capacitance could be from a 'scope probe or from a shielded cable.

Try adding a series resistor to the output of the circuit to isolate a capacitive load. Try 10 ohms to 100 ohms.

 

[MRCecil]

From this side, it's not possible to determine the actual source(s) of the oscillations save noting the departures from the manufacturer's recommendations displayed in your as-built schematic. I would recommend the following:

1. Uncouple the second stage input and remove +Vs & -Vs;
2. Remove C6 from the first stage and configure the decoupling ckt. as shown in the datasheet;
3. Ensure C2/C3 & C4/C5 are placed as close as possible to the supply pins using the recommended 0508 case size;
4. Replace R1_1 with a 50Ω resistor;
5. Remove R6_1, R7_1 and C7_1 and place a 100Ω load on the output;
6. Pulse the input at an appropriate level, to avoid overdriving the amp, and check output for any ringing, excessive overshoot and for settling time within specified limits;
7. If ringing or prolonged settling is observed, look over your board and layout for issues such as long traces, poor etching (leakage), irregular separation between traces and/or ground plane, etc;
8. If the above appears OK, try experimenting with minimum values in a compensation ckt (pin 5) (see datasheet recommendations);
9. Also, one could try a feed forward cap across Rf (see datasheet);
10. Repeat process for second stage independent of the first;
11. Once both stages are working are within specs, couple them together, close & tight, and repeat pulse tests for results remembering the gain is now at 400 so input should be set accordingly;
12. If all is fine at this point, put your test signal to the input and check output for anomalies;
13. If that proves out, then start making your modifications to the circuit and test as you go.

I may have overlooked something along the line here, and others may see any gaps or plain old mistakes.

Something one should always keep in mind is that the semi manufacturers have put a bunch of R&D time into their designs, and know the advantages and shortcomings of their devices. They put both in their datasheets for specific reasons, though being couched at times.

Good Luck

 

[mneary]

Pin 5 of U1 appears to be connected to pin 5 of U2. Likewise pin 8 of U1 appears to be connected to pin8 of U2.

This is a fatal design flaw, and until corrected, would make all other work pointless.

 

[bleedingwolf]

Hi,
I finally fixed this oscillation by replacing R1 by a 50ohm resistor. I can not tell why. Do you have only idea about this?

Thank you.

 

[bleedingwolf]

Hi MRCeil,
I finally fixed this oscillation by replacing R1 by a 50ohm resistor. I can not tell why. Do you have only idea about this?

Thank you a lot.

 

[audioguru]

Hi,
I finally fixed this oscillation by replacing R1 by a 50ohm resistor. I can not tell why. Do you have only idea about this?

Maybe the input is like a receiver antenna. The low resistance of R1 damps radiated pickup from the output.

 

[bleedingwolf]

Hi MRCecil
I mean the single stage is always stable. But when I cascaded them together, it started to oscillate. And right now I replaced R1 by a 50ohm resistor and then everything is fine now. But why does it happen? I don't understand.

Bo Hu

 

[crutschow]

Reducing the value of R1 may reduce the high frequency gain of the path.

Incidentally, the filters between stages and at the output are high-pass, not low-pass.

What is the highest frequency you want to amplify?

 

[MRCecil]

Hi MRCeil,
I finally fixed this oscillation by replacing R1 by a 50ohm resistor. I can not tell why. Do you have only idea about this?

Thank you a lot.

It is likely there was a mismatch of impedance between your signal source, nominally low at ~52Ω-76Ω for many HF generators, and the input 10kΩ resistance initially presented. The output reactance of your source may have aided in starting the oscillations possibly via some feedback loop on your board or may have been isolated at the input stage.

Just guessing at this as I have no emperical data to evaluate, but stuff happens. This same situation should not be much of a problem on your second stage given the Zout of the first stage.

Good Luck

Edit: I have no idea what that angry face is doing on this post. Can't delete the thing either!