Precise measurement using AD8228

I am developing a precise current sensor based on shunt and instrumentation amplifier AD8228 sensing the load current of H-bridge.

Figure 1 is the H-bridge. The switches are IGBT switching at 30k~40k on fixed duty cycle. Vsense is the voltage of shunt ranging from -10mV to 75mV.

Figure 1

Because the IA cannot handle the common-mode voltage of 160V, I use an isolated voltage to supply AD8228 as shown in Figure 2. It should also be noted that the ground of the AD8228 is different from that of the H-bridge. So the common-mode voltage referred to AD8228 is 0 in this case!

Figure 2

But I’m not very sure whether AD8228 can work precisely and safety in such arrangement. Could anyone give me some advice?

PS: I know there are some types of instrumentation amplifiers, current sense amplifiers and isolation amplifiers can handle the common-mode voltage higher than 160V. But they all seem not to meet my requirement of precision.

       Thanks in advance!

  • 0
    •  Analog Employees 
    on May 26, 2013 10:25 PM over 7 years ago

    Hi xcuresme,

    Based on experience, having an isolated source can lead to problems with your measurement. I've seen it myself in the lab where I used a DC source which was grounded separately from my system and it yielded loss of precision on the data. I understand the concern with the common mode voltage, but configuring the system in this manner would probably not be optimal.

    You mentioned that you've already looked at some other parts. Have you had the chance to check out the AD8479? It is one of our newest parts which can handle up to 600V of common-mode voltage. If the precision on this part is not acceptable to you, you could work around the common-mode problem by doing low-side current sensing instead. In this configuration, the common mode can move around without damaging the AD8228. The current sense resistor would be placed in series with T3 and T4, in this case.

    Do you think this would be a viable option for you?

    Thanks,

    Kris

  • Hi, Kris.

    Thank you very much for your recommendation of AD8479 and low-level current sensing. However the focus of my current work is to find out whether this configuration can work precisely.

    And according to your reply, such configuration can't work as precisely as specification says, right? How much the error it causes?

    By the way, I am now using the evaluation board of AD8228. I would like to ask whether soldering the wires directly to the vector pins (testpoints) is the most optional way to connect the board?

    Best Regards

    xcuresme

  • 0
    •  Analog Employees 
    on May 28, 2013 12:12 AM over 7 years ago

    Hi xcuresme,

    I couldn't get a hold of the data as the experiment was done a few months back, so we just tried it out in the lab today. We used the AD8221, just to test whether the isolation would be a problem. We came up with about 400mV of error for a 5V input. That's a pretty big number, and I am wary that this error can be tolerated in your system. Again, I've usually come across the use of in amps in low-side current sensing, but I would be interested in how your results will come out. I still advise you to be careful about that 160V common mode though.

    For your question on the evaluation board, yes, I would recommend that you solder the wires to your vector pins. It would be preferred for you to use twisted pairs for your wires to avoid noise coupling into your circuit.

    I hope this helps!

    Kris

  • Hi, Kris.

    I have to say that your reply is very very helpful for me. The result of 10% error is totally  beyond my expectation and can't be tolerated at all.

    Well, your focus about how to measure the output of amp is insightful. And it also puzzles me for a long time. Now I'm using high voltage differential probe to get the result, based on the compare between differential probe, floating channel oscilloscope, DMM and DAQ.

    Because I'm going to do a similar experiment Thursday, I would to know  what your experiment is like, for example the circuits, the equipment and so on.

    Besides would you mind to show me a picture of a wire soldered with the vector pin? I should be very careful about the experiment, so I should make sure everything is correct.

    Best Regards

    xcuresme

  • 0
    •  Analog Employees 
    on May 28, 2013 10:48 PM over 7 years ago

    Hi xcuresme,

    This is a sample of a board that I worked on before. These twisted wires were tapped to the OUT+ and OUT- of the amplifier that I used. You'll want to keep the wires as short as possible.

    For the experimental configuration that you were asking for, I used the AD822x evaluation board with the AD8221 in a gain of 1 configuration. I used dual supplies of +/-15V using an Agilent 3631A. Measurement of the output was done with a Keithley 2400, and the input was supplied from another programmable power supply B&K 9130.

    Was this the information you wanted?

    Thanks!

    Kris