Other Parts Discussed in Thread: LP5907
Tool/software:
Hi all,
I'm having a hard time nailing down a noise problem with INA240, though I am pretty convinced it is a layout issue.
Before I commit to redesigning part of my board, I wonder if anyone can weigh in on this issue.
My board is a 3ph motor controller and I'm seeing huge noise on the ina240 output when switching is enabled.
I'll include some oscilloscope shots. In these tests I'm looking at the INA140A2 after a passive low pass filter (fc ~ 1.6MHz, R=10R, C=10n) with a DC bus voltage of 51V. I'm switching at 30kHz, and the phase B rise time is about 10ns. Im not applying any current. This is just switching at 50% duty cycle. In fact I have no load connected at the output of the phase. (Having a load connected had no effect on the noise I'm seeing)
On the INA240, I expect to see the short 2us peak due to the common mode rise of the inputs. However, it definitely does not settle to a measurable value. It seems to oscillate out until the next PWM transition. I see similar noise regardless of whether I have a load attached and am applying some current, or if the load is not attached.
Edit with some extra information:
- The noise goes away (with switching turned on) if I disconnect INA240 inputs and solder them to ground.
- I found this post and its not looking good for my layout. Sadness...
Can I email someone with images of my layout?
Thanks
An update on this.
I've tried the following, with no real change in INA240 output:
- I disconnected the differential inputs and connected them to the sense resistor via twisted leads to see if the PCB routing was having a large effect. The signal output remained pretty much the same.
- Cut a slot in my ground plane between the switching electronics and the current sensors.
- Increase my FET gate resistor from 30R to 100R. This took the VDS turn off time down to about 150ns. There is no overshoot or ringing at the switching node of the FETs. (There was none before too)
- Disabled phase A and C, so only phase B is switching.
- The INA240s are powered by a LP5907. I had 22u ceramic as the input/output caps. I changed these to 1u ceramic. They were too high according to the datasheet.
- INA240 ref is connected to 3v3a and gnd for bidirectional operation. Here is a scope shot of 3v3a (green), ina240 (yellow) phase b switching node (blue)
An extra note: I originally had this problem when I was using INA240A3 100V/V gain. I dropped down to A2 with 50V/V gain, wondering if my layout was not good enough for 100V/V gain. The results presented are with INA240A2. I'll probably drop down to A1 20V/V and see how that looks too..
Thanks again
Update the 3rd: I think we've cracked the code here people.
I found a post talking about output filter stability. The output of my INA240 has ringing at multiples of my switching frequency, so.....
RE: INA240: how to design anti-alias filter for INA240+ADS7253
I have an RC low pass straight into my ADC. R=10R, C=10n. For funsies, I removed the filter R and attached the oscilloscope:
This is much better! Now I need to figure out some sane filter values. I thought R=10R, C=10n would be reasonable, but perhaps not.
Thanks again
