ADS1232 Problem

In debugging what you have, I would make a few measurements.

First, I would measure the refernce pins to make sure the reference is getting to the REFP and REFN pins. There should be 5V on pins 16 and 15. Then I would measure the input pins AINP1 and AINN1 (if those are the input you are using. Those would be pins 11 and 12. You should note that you are currently in a gain of 128V/V, so the signals you are looking at are very small. Even 200000(dec) which is the largest deviation is almost 0.9mV in signal. Just make sure that when everything is hooked up, that there is a signal and a reference going to the ADC. Measurements should be made at the pins just to make sure the signals are reaching the ADC. You could also reduce your gain to 1 or 2 just to see that you have a realistic reading from the ADC. You can also check to see that the load cell is changing the input when it is pressed.

After that, I would disconnect the load cell (to make sure the problem isn't there) and then short the inputs. This should give you the least amount of noise in any given measurement. I would then take an ADC measurement to see if this does indeed give output codes close to 000000h.

I'm not sure if its possible the problem is coming from the PCB material. I would think this is unlikely.

It is certainly possible that the noise from power lines and ground paths can disrupt a reading. However, the amount of noise seems to be too large. I would need to know more about your circuit, including the schematic and layout. I would also need to know about how the load cell is driven. I would also check to see if there is a problem in the schematic or layout to see if there is a problem in the hookup of the circuit.

Joseph Wu

Saeed,

I've looked through your layout and I don't see too much that would be a problem.

In comparison between your old and newer layout, I see that they are very similar to each other. However, I noticed that you use a completely different reference scheme, with a resistor divider supplying a lower voltage. Is there a reason for this? It certainly changes the input range and may decrease the performance. Generally, I wouldn't have expected this to degrade performance to the level you describe, but it might be an issue.

Also, how do you connect the analog and digital grounds on this design? It's not shown on the schematic (that I could find). This may also cause ground loop problems and create extra noise for the ADC.

As for the layout, the suggestions I have are generic suggestions that I would normally make.
1. I wouldn't run any signals or vias under the device. In general these can create or pick up EMI disturbances and create noise in the measurement.
2. When running the signal path, make sure that they run away from the the digital lines particularly SCLK and CLK. When I refer to the signal path, this would be any line going into the inputs and the reference.
3. The ground plane under the device should be continuous, without lines running through it (or vias as mentioned before).
4. Ground and supply connections should be direct to the local connection on the board. Both ground and supplies can be susceptible to digital currents which can cause spikes in voltage with longer ground or supply traces.
5. Supply decoupling caps can also help reduce noise. These should be placed on the supply near the ADC. In the old design, the reference was the supply and had a capacitor. In the new design, this is no longer the case.

In the comparison between the old and new designs, I would try to revert as much back to the old design as possible. See if you can reset the reference back to the supply instead of having a resistor divider. Also, I would try to clean up the power supply (by using an external supply) and see if the noise drops as well. As I'd mentioned before, I would check to see that a shorted input gives generally good results (disconnect the load cell and short inputs).

Joseph Wu