Lowpass filter oscillations 2

[djs]

Hello;
I have designed and built the attached amplifier/ lowpass filter. The circuit works pr say, however the low pass filter starts to have a small oscillation when its output approaches 2 volts. the period of the oscillation is about 14 us. U4 power supply pins are bypassed with 0.1 uf caps. The +5 power supply has no measureable ripple, and the -5 power supply has about 30 mv of ripple. The oscillations are about 0.1 volt amplitude.

Can some one give me some suggestions?

 

[Skogsgurra]

Some amplifiers with low current consumption cannot drive capacitive loads, the LM324 is an example and I guess the LMC6064 could also have that problem. Read the foot-notes in the data sheet. The capacitive load here is the feedback capacitor. Also, there is positive feedback through the protective diodes. Do they have to be connected to R8 etc? Wouldn't GND be better?

Have you tried other amplifiers? Have you tried to remove the protection?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[OperaHouse]

That input sure jumps out at you. Strange things can happen with a floating input. Since you are probably enamored with the diodes and this is low pass, try a .01uF or larger from pin 3 to ground. That would negate any capacitor effects of the diodes. I don't think the diodes are needed at all. Regardless, it is always a good idea to filter any input noise before amplifying it.

Another tale from the crypt. More than 40 years ago at my first job they had this transistor/FET transducer amp that they built. It had this $45 FET that they were so proud of (I bought a good used car for about that price). I looked at the schematic and it had this big cap on the output to ground, like about .1uF. They told me it would oscillate without that. Thinking of tube circuits I disconnected the gate lead of the FET and put a 100 ohm resistor in series with it. That was enough dampening to stop the oscillations.

 

[djs]

The input stage is not giving me a problem. The output of U4, pin 1 has zero or close to zero ripple. Also I am feeding the input with a function generator so is not open circuit. The ripple or noise shows up at U4 pin 14. It almost looks like the transients from a chopper stabilized amplifier.

The diodes are there for surge protection. I live in west central Florida, which is the lighting capital of the US!

I am attaching the data sheet for the LMC6064.

On further checking it looks like the ripple maybe caused by oscillations in other amps with int the package. I am continuing to work on it. Thanks for all of the replies and keep them comming.

 

[Skogsgurra]

Page 8 says exactly what I said about capacitive load. "Stability vs Capacitive Load, RL = 20 kohms and 1 Mohm". Please note that the stability is also output voltage dependent - just what you say you experience.

There are good application notes in that data sheet. I think you should study them and also read the foot notes.

lightning does not hit instrumentation dirctly - not even in Florida. There are far better methods to protect the input, if you think it is necessary.

Do not take offence if I point out that the phrase is 'per se' and the meaning is 'in itself' or 'as such'.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[djs]

An update on this issue. The output of U5 pin 1 is feed to an analog input on a MCU. The ADC on the MCU is a switched capacitor unit. The oscillations I am observing are transients caused by the ADC feeding back through resistors R13 and R12. The LMC6064 is a low bandwidth IC so it is not capable of quickly responding to switched transients.

The good news is that these switched transients do not affect the overall performance of the circuit. I have checked linearity throughout the range of the circuit and it is to witnin two or three ADC counts for the 12 bit ADC.

 

[benta]

I wonder why anyone today would even think about using a Sallen-Key filter, which is probably the worst topology ever made.
Please change to a working topology, I can suggest MFB as a possibiliy.
TI has excellent application otes on filter design, listing the limitations of different types. Example:

http://focus.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sloa049b&fileType=pdf

Cheers,

Benta.

 

[zappedagain]

To continue benta's point - the parasitic elements on the components in a Sallen Key low pass filter make a really nice high pass filter.

Z

 

[djs]

I have not observed any artifacts from the Sallen-Key filter. The signals in this design are so low freq that any design would work. The Sallen-Key was the easiest to implement.