For several month we are using the AD7190 in an instrument of us. We use 4 AD7190’s on a single PCB.
Two are used in differential input mode with a amplification of 128 for direct measurement of thermocouples. Two others are used with a gain of one.
We already produced dozens of these instruments without having any problem.
Here is a schematic of how one of the AD7190s with the 128x amplification is used in this instrument:
However, now we have a new instrument and again I use the AD7190 with a 128x amplification to measure directly a differential thermocouple signal and I have problems.
Note that the same firmware is used to set-up and read-out the AD7190.
This is my problem (note that the signal is also filtered in the PC software that has drawn this picture, running average with dynamic array width):
The input signal of the ADC is about 2µVpp, so the output code of the AD reaches from -630 to +200 in the measurement given above.
The input signal is a real triangle. I measured it also with the AD7190 of our first instrument, but that gives a correct signal.
This problem arises very often with our new PCB, but not always! I have soldered on the PCB and changed a resistor. After that it functioned correctly for a certain time and other PCBs always give this result and again others always give a correct result.
When I have a PCB that is correct, this is the result:
Does someone have an idea of what is wrong?
The schematic of the AD7190 PCB for this new instrument is teh same as the AD7190 given in the designa above, only a new PCB with only that one AD7190.
However, as I already wrote: also sometimes the signal is correct after soldering on the PCB (and switching off and on the instrument).
I also connected pin 9 to AGND (which is not given in the schematic above).
After start-up communication with the AD7190 is as given in the overview below:
So after this, the AD7190 is running in continuous conversion mode with 10Hz update rate.
Can someone please help me with this problem. I am testing already for more than a week on the issue, without any results and production of the instrument has to start!
Thanks in advance.
The measurements given below were both performed with the PCB with the one AD7190, causing the problems.
Now the measurements were performed without filtering, it’s directly the ADC output signal. Note that these are measurement of two separate PCB’s. The first measurement results in the filtered signal with the steps, while the second measurement results in a correct triangle.
However, although the filtered signal of the second measurement shows a correct triangle, you also see the effect of the problem a little in the highlighted box. It’s only much less.
What you also see in the above measurement is that the noise becomes greater when the signal is near the midscale and that the noise is not symmetrical. Noise peaks are much larger in the direction of the midscale!
Also noise is much higher in the above measurement.