Dear all,

I have a mosfet h-bridge feeding the phase of an electric motor.

For measuring the current we were using so far a LEM LTSR 15-NP but now the ADC will be in the high power side and therefore the isolation given by this component is not an advantage anymore.

I am trying to get an accuracy of around 0.15 A for a ±15 A range. (0.5 % of full range)

Is this kind of device still a good option or a sense resistor could give better results? (or other way)

I think that probably a precision resistor (0.1% tolerance for example), plus the amplification and offset (to get a proportional shifted voltage between 0 and 5 V to be read by the ADC) could give better accuracy than the LTSR (0.7% accuracy plus 0.1% non-linearity), but probably the common mode voltage applied to the resistor could make a precise measurement difficult.

The switching frequency is 50 kHz, and the voltage supply for the bridge 170 V.

Thank you very much in advance for any advice!
Ricardo

 

I guess you have to ask yourself if the increased accuracy is worth the effort of making a new measurement system. Do you really need .1%? Is the high power side really that critical? If the method you used worked before, it might be the best choice. That said, either method would yield good results.

 

I suppose a resistor as sensing element has to be cheaper than that tranformer-based device. If it's worth doing the design work to save the money, it should be pretty straightforward.

If it were my project, I'd have used Hall-effect sensors like this:
http://www.allegromicro.com/en/Prod...ifty-Amp-Integrated-Conductor-Sensor-ICs.aspx

But I do see that they only claim "typical output error of 1%", so maybe that wouldn't have been good enough.

 

While 1% error might sound like alot, remember that using a series resistor to sense current will add error to your system. For example, say you're measuring 10A on a 10V system, using a .01ohm resistor. For the same circuit without the current sense resistor, the current should be 10.1 amps. This is a 1% error, even if using a very precise resistor. Add to that errors in your amplifier, VOS, IOS, etc...