I'm trying to use a Data acquisition unit (Microcontroller maybe) to measure a voltage difference between two cell of a battery (the battery is about 400V). The Data acquisition Unit and battery are connected in ground (this is call ground coupled, right?), I'm thinking of using a differential Amp- AD629 (which can measure ±270V) to measure the voltage differences between two cells. How should I connect it? Is there a high voltage virtual ground chip I can use?

datasheet for AD629
http://www.analog.com/static/imported-files/data_sheets/AD629.pdf

I'll probably use a ±15V DC power supply to power it.

Is there a differential Amp that can measure ±400V and up?

 

I have used the Analog Devices AD202 isolation amplifier as a shunt isolator--the shunt floated on a 700V bus. This is the type of application for which it is intended and would be a good solution for your problem. Analog Devices has a family of these products. The AD202 has a 2kV common mode voltage rating.
http://www.analog.com/en/specialty-amplifiers/isolation-amplifiers/ad202/products/product.html

Using an op amp for this application is like playing with fire because of the energy available in the event of a fault (IC failure). Properly fused, it would be safe, but may have reliability issues. Generally, it is not acceptable to operate that close to a device's voltage rating.

 

So, all I need is to set a few resister to adjust gain. Put in the two measuring voltage at pin1(in+) and pin3 (in-), then I can connect it to a microcontroller, to an ADC pin (pin38>Vout Hi) and ground (pin37>Vout Lo), then I'm set right? (I'm sorry for the extra care and confirmation because this sounds too easy! =D)

The other part is accuracy- it says 0.025% max, so the accuracy and resolution is 0.5V? (2000*0.025%), or the resolution could be higher?


Thank you so much for replying jimkeith! I really appreciate your help!

 

If you look at the data sheet you will see that the GAIN accuracy is \(\pm\)0.5% typical, \(\pm\)4% maximum.

Are you planning on measuring more than one cell in the battery?

 

After thinking about this, I changed my mind--proper application of a differential amp as an attenuator may be simpler and more accurate--check out the thumbnail.

The overall common mode input voltage = device common mode input voltage * gain e.g. 27V * 20 = 540V.

Device common mode input voltage range is generally about Vcc-3V and Vee + 3V. In this case, by making the positive and negative supplies unequal, the device common mode input voltage range is shifted up to adapt it to your situation.

So the overall common mode input voltage at the input resistors easily exceeds your battery voltage. Now the 3V differential voltage is divided by a factor of 20 by the attenuating amplifier, but it is restored by the 2nd stage.

I'll leave it to you to calculate the accuracy. You may use 0.1% resistors

p.s. enable your email function--I was going to email to tell you to check the thread

The isolation amplifier suggested does have an accuracy limitation as indicated by crutschow.

 

A word of explanation--my suggestion to initially use an isolation amplifier was based upon my erroneous thinking that the signal level (cell voltage) was too low to use an attenuating differential amplifier effectively--it would be to isolate a 100mV shunt signal--but the cell voltage is a factor of 30 to 60 times greater, so this makes it feasible.

Another variation would be to use a high voltage op amp e.g 70V rail to rail--this would reduce the attenuating factor by a factor of 2. But this would require a relatively high voltage e.g +65V power source just for the op amp--this would be an oddity.

 

If you look at the data sheet you will see that the GAIN accuracy is \(\pm\)0.5% typical, \(\pm\)4% maximum.

Are you planning on measuring more than one cell in the battery?

At max, I'll only measure 2 cells at the same time, and that's about it. The maximum differential voltage should never exceed 7V.

Thanks for answering me at the thanksgiving day BTW!

 

After thinking about this, I changed my mind--proper application of a differential amp as an attenuator may be simpler and more accurate--check out the thumbnail.

The overall common mode input voltage = device common mode input voltage * gain e.g. 27V * 20 = 540V.

Device common mode input voltage range is generally about Vcc-3V and Vee + 3V. In this case, by making the positive and negative supplies unequal, the device common mode input voltage range is shifted up to adapt it to your situation.

So the overall common mode input voltage at the input resistors easily exceeds your battery voltage. Now the 3V differential voltage is divided by a factor of 20 by the attenuating amplifier, but it is restored by the 2nd stage.

I'll leave it to you to calculate the accuracy. You may use 0.1% resistors
.....................

The worst-case common-mode rejection using 0.1% resistors is -60dB which would give a 400mV offset at the second stage output for a 400V common-mode signal. This can be nulled with a pot adjustment into the second stage.

 

If I were to use it to measure as low as 0.8V per cell, will I be getting about a 1% accuracy of finding the differential voltage from the two stage amplifier?
So the isolation amplifier is not that accurate?
I was just thinking that If it works, that would be good because I don't need to build a circuit almost with the isolation circuit.