• TI Thinks Resolved

INA3221: Kickback noise from the ADC like on PAC1934

Prodigy 30 points

Replies: 5

Views: 65

Part Number: INA3221

I use the PAC1934 to monitor currents and voltages on rails with relatively high impedance (1k or so). I found that it creates very fast transients on the signal when its ADC samples the inputs. This is not acceptable in my circuit and it was not possible to reduce this to an acceptable value by the suggested filtering measures (series resistors, parallel cap).

Is the INA3221 better in this respect? Or do I have to use a part like the INA219, which has a PGA in front of the ADC, for which reason I wouldn't expect any such kickback noise.

  • Hi Gerhard,

    Have you found out why the PAC1934 injects fast transients to your signal? What are the characteristics of these transients? Since this is not a TI device, we don’t have enough information to make a prediction.

    However I haven’t heard of similar issues for any of our digital power monitors. There are precautions in our design to minimize charge injection. I suggest try with INA3221 EVM for a quick experiment. Fly wire the EVM to your PCB likely will work just fine.

    Regards, Guang

  • In reply to Guang Zhou:

    Hi Guang,

    thanks for your reply. As I wrote in my post, the PAC1934 injects the transients each time it samples the signal.

    Microchip support says, that it discharges it's sampling caps before each new conversion. Thus, each time it samples it's inputs, it has to charge the internal sampling caps. This creates extremely fast transients that are hard to filter. Obviously it just switches the input signal on the sampling caps with no buffering in between.

    As buying and trying out on the INA3221 EVM takes money and time, I would really prefer to get a more well informed response than "I haven't heard of such issues". It would help me a lot, if you could check with someone who knows definitely, whether it doesn't inject such transients when sampling the input signal. If e.g. the INA3221 uses a buffer before its sampling cap, we can be sure, that there are no transients injected.

    Thanks a lot!

    Gerhard

  • In reply to Gerhard Petrowitsch:

    Hi Gerhard,

    INA3221 is based on sampled capacitor topology; there is going to be certain degrees of charge injection. But it is impossible to say how it compares with the competition device without actual measurements. Because the circuits are different, your design also plays a role in this.

    INA219 is based on similar topology as INA3221, the internal PGA is part of the ADC function and will not act as a driver that eliminates charge injection.

    Even though it takes effort, the best option is still to actually try with INA3221. Please let me know if there is anything we can do to help.

    Regards, Guang

  • In reply to Guang Zhou:

    Hi Guang,

    thanks for the info. Is there a similar device in your portfolio, which uses an amplifier before the ADC so that there is not charge injection to be expected? Or should I just use separate devices, like the INA138 plus some ADC at  its output.

    Regards,

    Gerhard

  • In reply to Gerhard Petrowitsch:

    Hi Gerhard,

    Currently we don’t have a current sense amplifier with front end buffer or PGA. INA138 will not have this issue as you pointed out. It might be worthwhile to look into and evaluate this device further.

    Regards, Guang