Is is a proper way to solve the problem of high common-mode voltage?

[johnsold]

Henrik,

If the switches are implemented as power MOSFETs, there are cases where current flows through the body diodes of an "off"switch.  Currents can circulate in the load without going to ground.  Even with ideal switches, consider the case where T2 and T4 are on while T1 and T3 are off. The current flowing in the inductive load will continue to flow until the energy is dissipated in the shunt resistor but no current would flow in the grounded shunt resistor.

In this type of circuit you really do need to measure the current at the load.

Lynn

[xcuresme]

Exactly!

[Henrik_Volkers]

douze points Johnsold 🙂

@ xcurseme

What is the spec you want to meet?

Bandwith, range and uncertaincy?

LEM sensor , or how about an isolation amplifier?

[xcuresme]

Hi, Henrik.

The most important thing for me now is uncertaincy which should be better than 0.5%. Bandwith is also critical but it's not the focus of current version. The bandwith of current version is about 1kHz.

The current sensor based on shunt instead of sensing current is my aim.

The isolation amplifier is not proper to be a pre-amplifier in precise measurement for its large offet voltage and offset drift.

There is a perfect current sensor developed by Isabellenhuette, a German company, whose performance is amazing!

Thank you!

[johnsold]

I am glad that you have identified a suitable sensor.

I would suggest that your comment about bandwidth shoud be reconsidered.  With the rapid switching of the loads, a bandwidth of 1 kHz may not provide current measurement with an uncertainty < 0.5%. Unless the inductance is high enough that the load time constant is >> 1 ms and the parasitic capacitance is quite low, transient currents may introduce significant errors.

Lynn

[xcuresme]

Hi, Lynn.

The current sensor I mentioned above, IMC-A-R0001, is the only one I have found in the market which is based on shunt, has the capability to measure large current under high and swinging common-mode voltage and  achieves high resolution and wide bandwith. It gives me confidence that a current sensor based on shunt can have excellent performance as well. Then I determine to develope a current sensor based on shunt.

I am always puzzled about the relationship between the uncertainty and bandwith. So your comment is very helpful for me though I can't understand well. And here I would like to illustrate my idea about them clearly.

In fact the waveform of the current is square wave whose rising time (tr) is about 350us. The uncertainty requirement focuses the flat section of the waveform. So the requirement of the bandwith is that it should be as wide as possible so that it will cost less time to settle within the uncertainty level during rising or falling. The bandwith of my sensor is limited by the RFI filter before the inputs of IA whose differential bandwith is about 2kHz. According to the relationship between tr and f3db, tr*f3db=0.35, the bandwith of my sensor is 2 times of the f3db required. So I think it may be enough.

Thank you.

[Henrik_Volkers]

So if you have a nice shunt , take a look at the ISO124 , add another IA for an initial gain, a small transformer for an isolated supply ( or a DC/DC block ) , DONE  🙂

(edit: Take care about parasitic couplings... so still a challenge, if it would be that easy, everybody could do it 😉 )

I assist Lynn, design the bandwidth as high as possible (10kHz should be easy, 40k-50kHz is better) even if you think you don't need it now, forget 1kHz for 0.5% , in even simpler cases I have seen differences of more than 2% in motor energy label tests between traced test labs and the main cause was a 4kHz to 8kHz sample rate at the energy meters.

http://www.zes.com/  have some nice appnotes on current / energy metering  with switched power supplies

[xcuresme]

Thank you for your share, Henrik.

I know I will be faced up with many unexpected challenges. Exciting~

The bandwith of my sensor is limited by the input fillter. I will reconsider it.

BR 

[xcuresme]

Er...I find the appnote (#109 in this page) can't be downloaded.

Do you have such problem?

[Henrik_Volkers]

---

@xcuresme wrote:

Er...I find the appnote (#109 in this page) can't be downloaded.

Do you have such problem?

---

No