Walkie talkie signal interferes with load cell signal?

[seyyah]

We have developed a device that measures weight using loadcell. It works in a normal condition. But if a walkie talkie runs near it, it affects the device and the device shows wrong values. Loadcell cable is shielded and when we properly
connect it to the ground, the cable is not affected. But if we move the walkie talkie near the device it is affected. Pcb routing is one thing but i believe we should solve the problem mainly in the program. we use cs5532 as ADC. What do
you suggest to solve this problem? I believe the walkie talkie frequency is in the hundreds of khz range. So what should we do to eliminate this kind of noise? By the way we use 50-100sps conversion rates. there is a differential and common mode hardware filter at the inputs of the adc at a cut off frequency of ~2KHz.

 

[joseMiguel]

Hi,
1) Do you have differential input signal?
2) have you tried a faraday cage on you device?
3) Do you use a spectrum analyser to investigate?
Hope it help you.
JoseMiguel

 

[seyyah]

1- yes the input is differential
2- faraday cage is not practical at the moment because it has a designed and manufactured box and also a different brand but same function device which has no faraday cage has no problem like this, so we should do something without cage first.
3- No, only oscilloscope.

 

[mister_rf]

Maybe need to use a combined RF filter( minimum design using ferrite beads +capacitors) + low pass filter (also good decoupling on power supply lines), see an example in the attached schematics.

 

[FvM]

The wrong display is most likely caused by RF signals rectified at input stage of the ADC, an additional input amplifier, or any external protection diodes. Also the bridge supply may be affected. The "walkie talkie" will typically use 27 or 433 MHz band. It's very unlikely, that the problem can be solved by software means. Well considered connection cable shield and circuit grounds, and RF filtering in bridge supply output and differential measurement input should help.

 

[seyyah]

Maybe need to use a combined RF filter( minimum design using ferrite beads +capacitors) + low pass filter (also good decoupling on power supply lines), see an example in the attached schematics.

It has a similar configuration except that we used analog front end instead of instrumentation amplifier and we used a series voltage resistor instead of buffered resistor voltage divider. Also ferrites and ground couplings are in place.

---------- Post added at 21:42 ---------- Previous post was at 21:33 ----------

As they said me, they used a modified walkie talkie which has lower frequency then normal in order to lengthen the communication distance but i don't know the exact frequency. Anyway, the cable has shield and when the walkie talkie is over it, there's no affection if we properly ground it. But if we directly move it over the open circuit or adc there's substantial slip in the measurement. Input line of the adc circuit is similar to the one of mister_rf's schematic. There is no additional buffer or amplifier. Also there is no diode on differential signal line. But i don't know the afe's internal circuit exactly. It has an amplifier at first stage but i don't if it has a diode or other similar thing.

 

[FvM]

But if we directly move it over the open circuit or adc there's substantial slip in the measurement.

This may be normal operation. Generally, sensitive elecronics may require shielding for sufficient EMI immunity. As long as you don't know the frequency and output power of the device, the observation is almost meaningless.

 

[seyyah]

My friend has made a measurement with the device. I still don't know the frequencies and amplitudes he showed me the loadcell return signal and i tried to sketch it. It is similar to the one in the attachment. The signal(dc component) is jumping %50 or more of the initial (dc component)value when the walkie talkie is on. It seems there is an ofset. How can this be eliminated?

 

[mister_rf]

In my opinion there’s a ground noise as this may impose to redesign the whole PCB. Ground planes provide a common reference point for electrical signals, and they can also be used for shielding, so need to pay attention to this aspect. Some materials related to:

Analog Devices : Analog Dialogue : PCB Layout

**broken link removed**

https://focus.ti.com/lit/an/szza009/szza009.pdf

**broken link removed**

 

[seyyah]

Yes you may be right, he had redrawn the board to add something and i don't know how it is drawn. I also want to mention that it is difficult to design this type of circuit pcb. It is not suggested to seperate the analog ground and digital ground(actually there are different opinions) and they should be shorted in the shortest mean of the way to prevent leakage currents (related to 24 bit delta sigma converter inner circuit). The board is small and it has 24v, two 12 volts which of one is isolated, 10V, 5V, 3.3V, 2.5V rails. And a 4 layer pcb. I find it difficult to design power distribution of the pcb. Any suggestions?

 

[FvM]

I keep my opinion, that a load cell amplifier circuit may need external shielding to achieve immunity against strong RF fields. Because the test results don't give any quantitative data (frequency, power/field strength, distance), it's difficult to determine, if it's a particular problem of unsuitable PCB layout or, as already guessed, just normal operation.

On the other hand, you'll find some potential for improvements in most designs. A RF hardened would be comprised of
- continuous ground plane
- multiple low inductance bypass capacitors for all supply nodes
- LC filters for all power and signal lines at the board interface
- additional bypass capacitors for low level signals near the analog chips to reduce residual RF
- local shields over sensitive circuit parts