• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 3 answers
  • Subscribers 48 subscribers
  • Views 445 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Noise in current measurement

Sebastian Siegfried
Intellectual 680 points
Part Number: INA286-Q1
Other Parts Discussed in Thread: DRV8300, INA286, INA281, INA240, INA293

Hi,

I'm trying to build a custom motor control board, using a C2000 uC.

As a driver IC, I use the DRV8300 and for current measurement, I use the INA286 (because it is one I could find easly, evertyhing with less than 100V/V gain is really hard to get).

Now, all the signals (Vsupply, Va, Vb, Vc, Ia, Ib, Ic) can be measured, but I noticed, that when I start the controller PWM (without motor at the moment), I immediately get really high noise peaks on the current measurement and I don't really know, how they occur. I was pretty careful with the design, but it seems, that I have done something wrong or that the INA286 with such a low shunt is not handable. I will try to desolder the cap at the differential input, maybe that could be the problem? 

There is no noise on the 3.3V supply for the IC, so also this cannot be the source of the noise.

Here is the schematics of phase A with the current measurement path:

Layout TOP:

Layout GND:

Layout VCC_SIG:

Layout BOT:

Osci INA286 OUT (huge peak):

So, can you see something that is really wrong here or is the IC not really dedicated for motor control? There are no peaks on the shunt itself, when I measure the HIGH side of the shunt.

So it really is somehow coupled into the measurement IC.

Edit:

So I also measured the high side shunt and the INA286 output. I could not use a sping probe, thats why you see the PWM signals, but I think that makes the capture more valuable.

It seems, that the start of such a overshoot is somehow triggered by a PWM signal, I really cannot explain that behaviour

Edit2:

The INA281 has this switching capacitor structure:

Might this be a problem?

Edit3:

The pulses seem to occur always around 3-4us after a PWM Signal:

And the pulses seem to be pretty random:

  • 0
    TI__Mastermind 18790 points

    Sebastian, 

    What your most likely seeing here is the common mode response of the INA286. Have a look below:

    If you are measuring in line with the motor current, then the shunt is also being subjected to the PWM of the motor, which causes the common mode voltage of the INA286 to rapidly transcend between the V_HI of the pulse, and GND. Even with the common mode rejection present, there will still be small pulses present as the amplifier works through these transients, and this is most likely what you're seeing on the scope. If you put the PWM on the other channel of your scope, it will most likely line up that that the up and down ramp of the PWM are in line as per the above images. 

    For such an application, I would recommend looking at the INA240, or the new INA241, as these devices are designed for this type of application, and exhibit a much smaller response here. Note the difference not only spike magnitude, but also the time divisions of the graph:

  • 0 in reply to Carolus Andrews
    TI__Mastermind 18790 points

    Sebastian, 

    I just saw your schematic. You are on the low side actually. Interesting. Let me have a deeper look and provide additional comments. There may be more to this. 

  • 0 in reply to Carolus Andrews
    TI__Mastermind 18790 points

    Sebastian, 

    The only other thing I see here is that there is quite a lot of capacitance on the output of the device, have you tried increasing the output resistor R12? The maximum output capacitance sustainable by the INA286 is 1nF, and you have 2.2nF present on the line. 56Ω's is not a large amount of isolation resistance. Could it be what we're seeing here is simply marginal instabilities on the output? 

    Is it possible to get the high side of the shunt measurement you mentioned earlier on the scope with the output?

  • 0 in reply to Carolus Andrews
    Intellectual 680 points

    Hi Carolus,

    Thanks for your answer. Yes, I 'm on the low side only, so I don't understand it either. It is exactly the same noise on all 3 amplifiers.

    Hmm interesting thing about the maximum load capacitance, I have not seen this when I designed the circuit, but since I cannot really detect any input noise to the amplifier it must have something to do with that. I will try to increase that next. But isn't that load capacitance only for 0Ohm resistance at the output?

    I can only see the noise spikes, when I turn on the PWM, so when I just turn the motor without PWM (all phases are shorted when DRV8300 PWM pins are low) I can measure the current pretty accurat (+-200mA @ +-55A range).

    The high side of the shunt does not show any spike (measured with the spring probe). I can try to capture them at the same time, but it seems anyway it does not occur at every PWM periode (you can see this already in the capture I uploaded).

    Btw: The PWM frequency is 20kHz

    Edit: Might it be a problem to use a senstive amplifier (100V/V) for a low side PWM measurment? Normally I always use something like 10V/V or 20V/V and I never had any problem.

    Kind Regards,

    Sebastian

  • 0 in reply to Sebastian Siegfried
    TI__Mastermind 18790 points

    Sebastian, 

    Please let me know if this has any effect on the behavior. Also, based on the schematic, it seems they are the same, but just to verify: the ground of the INA286 and the reference to IN- are the same GND, correct? Is there any potential for GND variation?

  • 0 in reply to Carolus Andrews
    Intellectual 680 points

    I removed R12 (56Ohm resistor) and now it is even worse. The INA286 GND is connected to IN- (Kelvin Connection) and I'm just giving a PWM signal on the outputs, there is no current flowing that could cause a GND variation. I'm really struggling to understand this. I also don't see, how anything should couple into the INA286 output line, since it is on the third layer, where the GND layer is in between and there are no parallel lines on top or bottom.

    Edit: I cut now the pin from the INA286 and I measure exactly the same without any output connection. The ouput trace itself is noiseless, so coupling from the PWM gate signals to it is definitely not the problem. Also I desoldered the 1nF cap between IN+ and IN-, but without any difference. So it is just the amplifier and the shunt. I really don't see, where this should come from.

    Edit 2: So i have a second board with a INA293 (I accidentely have choosen a unidirectional amplifier in the first iteration) with exactly the same layout and I don't have any peaks on the output when I activate the PWM signal with exactly the same shunt and output filter. I think somehow the INA286 is really not meant for PWM applications, only for slow slew rates like in a battery application. Can you recommend a bi-directional current measurement IC with the same housing (SOIC8) and the same pinout?

  • 0 in reply to Sebastian Siegfried
    Intellectual 680 points

    I removed R12 (56Ohm resistor) and now it is even worse. The INA286 GND is connected to IN- (Kelvin Connection) and I'm just giving a PWM signal on the outputs, there is no current flowing that could cause a GND variation. I'm really struggling to understand this. I also don't see, how anything should couple into the INA286 output line, since it is on the third layer, where the GND layer is in between and there are no parallel lines on top or bottom.

    Edit: I cut now the pin from the INA286 and I measure exactly the same without any output connection. The ouput trace itself is noiseless, so coupling from the PWM gate signals to it is definitely not the problem. Also I desoldered the 1nF cap between IN+ and IN-, but without any difference. So it is just the amplifier and the shunt. I really don't see, where this should come from.

    Edit 2: So i have a second board with a INA293 (I accidentely have choosen a unidirectional amplifier in the first iteration) with exactly the same layout and I don't have any peaks on the output when I activate the PWM signal with exactly the same shunt and output filter. I think somehow the INA286 is really not meant for PWM applications, only for slow slew rates like in a battery application. Can you recommend a bi-directional current measurement IC with the same housing (SOIC8) and the same pinout?

    Edit 3: I also desoldered R11 and D3, since they are coupling into the high side shunt, but still no success. I really cant help me where this issue come from

  • 0 in reply to Sebastian Siegfried
    TI__Mastermind 18790 points

    Sebastian,

    On the output tests, when you removed R12, did you replace it with another value? The recommendation I was making was to increase the resistance between the amp and the capacitor to further isolate the capacitance from the output node, not decrease it. Or was you testing on the floating node once you removed the resistor? 

    Given that this is a switching capacitor front end, I agree with you that there may be a better fit for this application. The INA240 I mentioned above comes in an SOIC-8 and is pin for pin with INA286 here. 

  • 0 in reply to Carolus Andrews
    Intellectual 680 points

    I was just testing on the floating node, I know that it should be increased. I will try the INA240, I'm now 95% sure, that the switched capacitor input stage makes the problems. Would be great, if you can verify this internally and maybe write it into the datasheet.

  • 0 in reply to Sebastian Siegfried
    TI__Mastermind 18790 points

    Sebastian, 

    Understood. I will see if I can replicate this on the bench when I get a moment. Please let me know how testing with the INA240 goes.