noise problem with ADA4898-1

[glias]

hello all,

I have some problem with the ADA4898-1 which I use for my application (IR sensor polarization/preamplifier). I tested the first and second stage which seems to be OK, but when I tried to look at the output of the third stage I have a noise (about 2mV max). I'm wondering if I use correctly the op amp, in the datasheet they use it with small RF values to reduce the thermal noise, do you know if this will be the cause of my noise ? can I use a 1.5k feedback resistor when using it in non inverting stage ? I also made a mistake in my PCB : in the datasheet they said that all ground and power planes under the pins of the ADA should be cleared of copper, and it is not my case my PCB is 4 layers with integral ground plane on the layer 2... do you think it the reason of my noise ?
I also use 100nF and 10uF ceramic instead of tantalum capacitors, it may be better isn't it ?


hope that you could help me
regards

**broken link removed**

 

[keith1200rs]

Have you a bigger diagram? - it is tricky to read.

Keith.

 

[glias]

**broken link removed** clic on it to have a lager view.

 

[keith1200rs]

I cannot see an obvious problem, but are you sure the noise is caused by the third stage? 2mV on the output is only 100uV on the previous stage so would you be able to see it?

I don't think the decoupling or ground plane would be the problem.

Keith.

 

[glias]

Hello Keith, I have some noise on the output of U1 (test made by deconnecting R59), I can't really measure it (vertical sensibility is limited to 2mV) but I see ripple. As you can see I have made an error by placing the bypass capacitor to GND, it must be connected to the output of the regulator, but the measurement is still "little" noisy, so maybe the regulator is altered (I see this error after 1-2 hours) I hope that is not my PCB placement/routing.
My decoupling capacitors (ceramic 10µF) are placed on the bottom of the PCB, and what do you think of choosing ceramic instead of tantalum caps for decoupling ? as I know ceramics caps have a low ESR/ESL ... ? better than tantalum caps ?

 

[keith1200rs]

As you can see I have made an error by placing the bypass capacitor to GND, it must be connected to the output of the regulator

I assume you mean the LT1964 bypass pin. I suggest you correct that - it is easy enough to do with a short peice of wire and either a cut track or raising one leg of the capacitor. It may not be the cause of your problem, although you can check by seeing how clean the regulator output is.

It may be worthwhile trying a decoupling capacitor on R1/R2 junction to ground.

Ceramics are very low ESR and so very good for decoupling. They can cause instability with some LDO regulators which require a minimum ESR for stability.

What frequency range are you working in?

Tracking down noise problems like this are tricky. It is one reason I designed & built a low noise oscilloscope amplifier many years ago, with the intention of selling it. However, in the end I didn't market it for sale but I still use one of my 20 year old prototypes. It gives me 20uV/division on my oscilloscope.

Keith.

---------- Post added at 10:01 ---------- Previous post was at 09:59 ----------

 

[houly]

Thanks Keith,
Yes as I said I have already correct it (for the by pass pin), but there still have noise on output. I wait for battery supply this afternoon, I suspect my power supply (I observed noise on the supply rails), however op amp have a very good PSRR @ low frequencies.

my frequency domain is 50Hz-350kHz

 

[keith1200rs]

Have you removed the sensor to see if that is the cause of the noise?

I am a little puzzled by the circuit. The reference for the first stage is slightly above -5V. I assume you are running the opamps from +/-12V or more?

You don't show any decoupling - I assume that is adequate?

Try connecting the first stage non-inverting input to ground and see what the noise is like.

How is the photodiode connected?

Keith.

 

[glias]

the sensor is equivalent to a variable resistor. I simulated with a 50 ohm resistor (see the input notified on the schematic : S5 and sensor[1] on the op amp)
the first op amp reacts from a variation of the sensor (resistor variation) and give a voltage from the regulator... we want to have the variation and amplify it, so the high pass filter is there.

I made a measurement with battery, but it's the same results.
Then I tried to look on the ouput of U6 and shunt the capacitor C9. It's better (about 1mV of noise) but the op amp saturates (normal...) so the result couln't be relevant.
I don't have idea where does it come from, I tried to reduc the value of feedback resistor because the datasheet mention that the ADA must be feedbacked with low resistor value and it's not my case : 1.5k and 2k...
For the first stage I tried to measure the noise with a low value of feedback resistor and I can't see on the scope a real difference. So i suppose that the influence of the value is not the cause of my noise.
I also placed in parallel caps to create a pole and therefore reduce risks of the instability. I have no ideas...

 

[keith1200rs]

Have you tried it without the sensor? Have you tried decoupling the junction of R1/R2?

To track down noise problems you need to make changes. Even if the change makes the noise worse it can help to point out where the problem lies.

Keith.

 

[glias]

Hello Keith,
Thanks a lot for your help. Well, as I said before I don't have the sensor yet ! I only place an equivalent resistor between S5 and SENSOR[1] (see the schematic), the value of the resistor is 50 ohms.
No I don't try to place a cap on R1/R2 => I'm going to test it and tell you it's pertinent.

I will send you the PCB.