help eliminating oscillation in circuit

[rich masel]

Hi:

I am building a preamp for a sensor, and am seeing an 2 mV oscillation at 21Khz on the output of U2. I replaced the sensor with a 1 Mohm resistor and still see the oscillation. Adding a 10 uf capacitor across R3 did not change the oscillation.

I simulated the circult in Tina-Ti and did not observe any oscillations. With the 10uf cap, the gain should be below -100 DB at 21 kHZ.

I would appreciate any suggestions. I need to use low power parts for my application, hence the choice or LMP2232.

 

[LvW]

I must admit that it is not easy to analyze the diagram as provided by you. For example, it tooks some time to realize that you are working with single supply - and I have problems to see if and how you have realized the dc biasing necessary for single supply linear operation. What is the purpose and the task of the two elements/units as shown at the top of the diagram?

 

[Arouse1973]

I must admit the circuit could do with a little tidy up. Let me see if I can help!
Adam

 

[davenn]

Hi Rich
welcome to Electronics Point

Yup not well drawn

U1 isn't even doing anything
what is VC1 in the + rail after the regulator output ?

ALWAYS put your ground (0V) rail along the bottom

those just for starters

Dave

 

[Arouse1973]

I hope you don't mind I have redrawn your circuit, please let me know if it's correct. Can you tell us what it is suppose to do, and why you have done what you have done. I could change the component idents if you want but we can still see what it does.




Thanks
Adam

[Mod Edit .... a couple of basic changes for power rail conformity ]

 

[davenn]

a lot easier to follow, Adam

as I said before ... U1 doesn't do anything

U2 isn't doing anything useful

where is your sensor and what sort of sensor is it supposed to be ?

have no idea what VG1 is ??

what is the point of R1 at 0 Ohms ?

if you read the datasheet for the REF3330, it will tell you the input bypass capacitance
must be between 0.1 and 10 uF, you have a total of 21 uF

Dave

 

[Arouse1973]

a lot easier to follow, Adam

as I said before ... U1 doesn't do anything

U2 isn't doing anything useful

where is your sensor and what sort of sensor is it supposed to be ?

have no idea what VG1 is ??

what is the point of R1 at 0 Ohms ?

if you read the datasheet for the REF3330, it will tell you the input bypass capacitance
must be between 0.1 and 10 uF, you have a total of 21 uF

Dave

Thanks Dave, Ti are a nightmare with their opamp power connections. I don't know why they do that.
Adam

 

[davenn]

in a second edit of the circuit, I also grounded the REF3330 Pin 3 which should have also gone to GND (0V)

the way the circuit stands at the moment ... it doesn't actually do anything

Dave

 

[Arouse1973]

I think that might have been intentional maybe for some form of control loop? Let see what the OP says.
Adam

 

[rich masel]

Let me explain the circuit. The objective of the circuit is to measure the current produced on a CO2 chemiresistor (R4) when 3 volts is applied across the chemiresistor. (The resistance of the chemiresistor varies non-linearly with the applied voltage and linearly with the concentration of CO2)

I found a oscillation when I ran with the real chemiresistor, so I replaced the chemiresistor with a 1 meg resistor for the tests. The output of u2 still oscillates at 21kHz.

In terms of the circuit there are two grounds in the system: an analog ground and a virtual ground.

U1 establishes a virtual ground above the analog ground. This varies with the sensor. In the example shown the virtual ground is 1.5 mV above the analog ground.
The U2/R5 produces a voltage proportional to the current through the sensor (R4).

The 3330 holds the voltage across the sensor (R4) at 3 V.
VG1 is a noise source used in the spice simulation. It is not in the real circuit.
R1 is listed as 0K. It was originally 5K to isolate the 3330 from the other capacitors, but bypassing it had no effect.

I also tried adding 10u capacitors across R5 and on the output from the 3330 to ground with no effect.

 

[(*steve*)]

please let me know if it's correct.

The redrawn circuit looks to me to be in error. The output of U1 does not appear to be grounded.

U1 establishes a virtual ground above the analog ground. This varies with the sensor. In the example shown the virtual ground is 1.5 mV above the analog ground.
The U2/R5 produces a voltage proportional to the current through the sensor (R4).

I would imagine that the voltage reference is oscillating. Try placing a low value capacitor (0.1μF to 1μF) between the output and ground.

It is hard to see what U1 is doing, but I note that your current sense resistor R5 seems very small compared to the resistance of the load. Swapping R4 and R5 would give you a signal that is about 2000 times larger. Given the 10u capacitor in the feedback of U2, it is unlikely to have sufficient gain at 21kHz to cause oscillation. Do you see the oscillation at the output of the voltage reference? What about at the output of U1?

 

[davenn]

The redrawn circuit looks to me to be in error. The output of U1 does not appear to be grounded.

Steve

I put that ground there as the REF3330 pin 3 should have been at GND(0V)
that was before the OPs following post talking about virtual GND's. In which case that GND can be taken away and let U1 provide the virtual GND to the REF3330, if that is what it is supposed to do ... yes ?

 

[(*steve*)]

Yes, I believe the function of U1 is to force a constant voltage across some of the components by referencing the ground connection of the reference to a buffered voltage equivalent to the reference voltage for U2. (how's that for three uses of the one word, each with different meanings)?

 

[Arouse1973]

So what's the circuit suppose to look like? Dave do you feel the corrections you made to my redrawn circuit are they correct? I feel the same as Steve about the purpose of the reference, but it's not the way I would do it but hey we are all different, well that's what my missus keeps telling me
Adam

 

[rich masel]

I cannot detect any oscillation at the voltage reference or at the output of u1 although it could be buried in the noise.

Adding a 1 or 10 uF ceramic cap between the voltage reference and ground had no noticeable effect. I also tried putting a ceramic cap across the voltage reference. Again, no effect.

 

[rich masel]

steve is correct on the circuit. I am trying to reference the 3330 to the buffered positive input to U2.

 

[Arouse1973]

Could you have long wires which is picking up interference? Might be noise be coupled into the ground wire of the scope probe. Connect the scope ground clip to the probe tip and move it around close to the PCB, do you see anything? What power supply are you using and can you run from a battery? Can you also take a picture of your setup?
Adam

 

[rich masel]

No effect of the scope wires. Also, if I remove the sensor (or 1 meg resistor) the oscillation disappears.

 

[(*steve*)]

Can you post a picture of your circuit, both sides of the board if it's on a PCB. I'm running out of ideas.

 

[davenn]

No effect of the scope wires. Also, if I remove the sensor (or 1 meg resistor) the oscillation disappears.

so did you read my comment way back in post #6 ?

if you read the datasheet for the REF3330, it will tell you the input bypass capacitance
must be between 0.1 and 10 uF, you have a total of 21 uF

high capacitance in the input can cause instability in the REF3330


Dave