CS30 - susceptibility to noise?

[R K]

Posted on June 22, 2018 at 11:56

Placing a CS30A next to a 0.05ohm sense ohm successfully measured output currents from 24v boost converter.

Since the microcontroller was not nearby, to reduce noise thought it might be better to place the CS30A next to the uC (with a low pass filter) and run two lines to the sense resistor. On the grounds that a) the longer traces should be low impedance one; b) any noise picked up would cancel in the differential pair;

Doesn't work well. According to my Fluke multimeter, the correct mV differential arrives at the CS30A, but it outputs voltages that are considerably less than the 20 times multiplication expected. So the ADC reports numbers for unrealistically low currents.

It's difficult to measure noise but the scope (with ground probe across the CS30 input pins) suggests the differential pair has picked some up.

Is it possible that it's the noise which is preventing the CS30 (tried multiple devices) from giving good results?

Any comments or suggestions?

[AUPETIT nicolas]

Posted on June 28, 2018 at 15:53

Hi R K

I'm afraid to have not correctly understood your point. Let me know if my understanding is correct. In your first sentence you said that the CS30A works properly.

Then the location of CS30 changed and it is placed far away from the Rsense (shunt) and close to the µC. And so you need to use long wire to connect the shunt resistor to the inputs of CS30. And in this configuration it does not work anymore. Am I right?

Is it the only change between the first configuration where it works properly and the second configuration (with long wire)?

Which kind of noise might happened in your application? I mean is there any possible EMI signal that can happened?

Because this kind of HF signal might disturb the CS30 behavior.

In a first simple trial,  May I suggest you to add some input capacitors (close to the input of CS30) to see if this have an impact on the performance of CS30. You can have a look to the application note AN4304 which explain how to filter input of High side current sensing by clicking on the link below:

https://www.st.com/content/ccc/resource/technical/document/application_note/a3/84/7c/19/21/a8/40/f3/DM00086777.pdf/files/DM00086777.pdf/jcr:content/translations/en.DM00086777.pdf

Let me know if something change

BR

nicolas

[R K]

Posted on July 02, 2018 at 16:21

This is most helpful. Even messily tacking on a couple of decoupling capacitors improved the performance of the CS30A. (The board has a 250kHz 24v boost convertor and a 1MHz 12v battery charger and could be picking up noise from both.)

Having studied the excellent app note wonder if the most intelligent thing to do would not be to: a) move the CS30 as close as possible to the sense resistor (since already know that gives better results); b) run the output the 10-15cm to the uC (it will have a much better S/N ratio); c) place a simple RC LPF at the input pin for the ADC; d)  implement a moving average filter in code? 

btw. is it not necessary to place a decoupling capacitor on the Vcc pin of CS30A? Have not seen that in any circuit schematics.

[AUPETIT nicolas]

Posted on July 03, 2018 at 16:00

Hi R k

Effectively the CS30 seems in a noisy environment, in this case the use of decoupling capacitances is highly recommended. And you are totally right power supply decoupling is extremely important as well. let's says 10µF and 100pF on Vcc should be good....

But one of the most important point is to take care to PCB.

EMI can couple directly with PCB traces, flat cables and IC package lead frame the higher the interference frequency the shorter the length of the wiring at which parasitic coupling is effective.

Effectively the PCB trace can be seen as for sure a resistive path, but also as an inductive component.

In low frequency the traces acts as a resistance but till the frequency increase the trace between more and more inductive and will act as an antenna which catch and then conduct any RF or EMI noise. And the inductance of the PCB trace is linked to its length. This is why it is important to keep the traces as short as possible in order to limit the antenna effect in high frequency. Using a ground plan is also recommended.

Filter can be added, but to reinforce the robustness of CS30. I will add a RC filter close to the output of the CS30, avoiding any parasitic to enter by the output of the CS30 and one more RC filter close the µC to protect the ADC against noise. And it would be preferable to keep the CS30 close to the shunt.

Kind regards

Nicolas

[R K]

Posted on July 05, 2018 at 18:00

I will  certainly take your advice on this.

Whilst having used 10uF at Vcc-in of larger components and modules, the CS30 consumes so little current would never have guessed such a high capacitance might be useful.

Thanks again.