Dear All,

I would like to get some advise on analog circuit for signal measurement.

My circuit is quite simple, I have two op-amp for amplification in a non-inverting circuit and then go into a 12bit A/D. The first op-amp is a low-offset and second one adjustable gain.

The fact is this circuit needs to be as stable as possible as absolute measurement will be done with it. The accuracy shall be below 0.2%. As the system has a "software" calibration, the important part is that it stays stable over the time.

The noise is also a concern, based on some publications, it is recommended for the op-amp resistor circuit two be as low as possible, in my case 10ohm and 200ohm (20x amplification).

But now I'm doing some tests and there is an other concern, as the resistors are low, the influence of the soldering resistance will have higher impact.
As for example, if the soldering resistance is 0.05ohm, on a 10ohm resistor it represents 0.5% but will only represents 0.005% on a 1kohm.

The problem is with thermal stress and ageing, the soldering resistance is likely to vary and comprise the stability.

All the circuit is SMD with 0603 resistors.

Anybody has experience with this, and is the thinking correct ?

Thanks

 

Very few if any opamps can drive a feedback resistor with a value as low as 200 ohms.
Some opamps can drive a total load as low as 600 ohms but then their feedback resistor value is much higher.

Resistors with values in the high hundreds of thousands of ohms and in the Megohms are noisy. Resistors with values of thousands of ohms are not.

 

The noise of an op amp with gain is primarily determined by the input resistor value and the op amp noise, since that is the only resistor whose noise is amplified. The feedback resistor noise is not amplified.

A low noise op amp such as an OP-27 has an input intrinsic noise about equal to a 700 ohm input resistor.

Solder resistance should be negligible for SMD resistors.

 

Mhh quite good feed back thanks.

My choice was probably not good but still it seems to work.

I will change that to 1k and 20k and perform some tests.

 

Feed Back:

1k 20k gave me too much noise, I should precise also i'm in a range of 0 to 2.5V.

100ohm - 2k gave me a good deal

 

Your estimate of solder junction resistance is a bit high.

PCB traces have more of an effect than solder junctions do. How wide are the traces on the board, and what weight copper? Do the signals go through vias/layers?

 

The signal does not go through via. The length of the signal track are max 5mm and 0.3mm wide, total area of the analog zone is about 1cm2 which has separated ground and VCC.

PCB is a four layer FR4, I think 4 OZ.

Here is the layout for reference, it's only the top layer. the two amp-op are msop

 

If I understand this link correctly, the surface mount solder resistance is in the neighborhood of 0.2mΩ.

 

I also forgot to mention that the manufacture did really crap job for soldering and there is a lot of cold soldering. (I've changed the manufacturer since).

 

Some people (animals?) solder electronic circuit boards by DRIPPING the solder from up high!

The company I worked for began importing Korean telephone systems in 1980 before LG and SAMSUNG began to do things correctly. The soldering was horrible and was unbelievable.
We didn't need a technician because when something failed we simply threw it away because it was very cheap.

Customers complained about the awful crossover distortion in the speakers on the phones so I added a resistor to change the class-B output amplifier to class-AB.