Can floating a signal be a proper way to solve the problem of high common-mode voltage?

[xcuresme]

Arrangement of the Amplifier and H-bridge

    First I want to introduce the whole experiment circuit.

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

 

Figure 1

    The connectivity information of amplifier is shown in Figure 2. The amplifier is an instrumentation amplifier ( IA). It should be noted that the ground of amplifier is different from that of the H-bridge because the IA cannot handle the common mode voltage of 160V. And that’s why I should examine the performance of the IA especially during the transition of switches.

 

Figure 2

Arrangement of PCI-6251

    As we can see from Figure 1 when T2 and T3 are closed, the common –mode voltage of Vsense is as high as 160V!!! But the working voltage of PCI-6251 is only 11V.

    So I plan to change the Vsense from a grounded signal to a floating signal. That means the DC voltage source of 160V is provided by an isolated AC/DC power supply as shown in Figure 3. In this way the “ground” of the Vsense is different from the ground of DAQ. So the problem of high common-mode voltage "disappear".

 

Figure 3

Puzzle

    Could anyone tell me whether this arrangement can work safety and precisely?

PS

    1) Because I want to examine the performance of amplifier under a wide and rapid variation in common mode voltage, to ground one end of the shunt won’t be proper.

    2) Someone recommends me to use DMM or oscilloscope for their capability of handling high common-mode voltage. But both of them can’t meet my requirement because the sample rate of DMM is too slow and the oscilloscope isn’t precise enough.

    Thanks in advance for any assistance!

[Henrik_Volkers]

Well, is the controller of the H-Bridge floating ?

How about the capacitive coupling in the tranformer?

Before connecting a DAQ: If the output is really floating, you can ground one pin of your shunt. Measure the current in this ground wire.

I would add some protection: Using the IA and put 100K on the input limiting the current in fault case . Take a look at the datasheet of the IA and check the current rating of the internal protection diodes, otherwise add your own, and check the input capacity since you build a RC input filter.

And check the Ti and Linear webpages for current sense amplifier....

[johnsold]

xcuresme,

Your switching rate is ~40 kHz. I have used the Allegro Microsystems, Inc. ASC756 Hall effect current sensors in exactly that kind of circuit (except MOSFETS rather than IGBTs) and switching at 100 to 130 kHz.  These come in 50 or 100 A versions with 500 VDC isolation.  Allegro makes other devices with other ranges.  The current leads go directly in series with the load and the output circuit runs on a +5 V and ground system which is compatible  with the DAQ devices.

Lynn

[xcuresme]

Hi, Henrik.

Yes, the drivers of the IGBT on the H-bridge are floating whose power supplys are isolated DC/DC. 

The capacitive coupling in the tranformer is one thing I'm not very sure. What damage will it cause and how to protect it?

In such arrangement I think the output is floating but I can't be sure for my lack of experiece. Is there any measure can make sure the safety?

As you can see from Arrangement of the Amplifier and H-bridge and PS, to ground one end of the shunt seems not to meet my requriement of examing the amplifier's performance with wide and swinging common-mode voltage.

Implementing protecting resistors will deteriorate the precision. 

I have checked the information of current sense amplifier before. They neither limit the applications nor can't meet the requirement of precision.

Thanks.

xcuresme

[xcuresme]

Hi, Lynn.

Thank you for your advice of using other type of current sensor.

Unfortunately, what I am most concerned about now is the small voltage on the shunt instead of the current. According to the reply these days, it seems to me that my plan is not so reliable. Sigh~

xcuresme