Electrical – Isolation Amplifier Gain Problem


I'm having a little trouble with an isolation amplifier. I'm using an HCPL-7800 isolation amplifier, in the circuit format provided by the manufacturer as a test circuit, and feeding it low voltage DC (0-200mV input). The gain of the amplifier should be flat at about 8.029 nominal, but at low input voltage (1mV) I'm seeing a gain of 25, and at my max input voltage (180mV) I'm seeing a gain of 7.9 (linear drop). I've almost exhausted every resource on this. I've laid out the board according to the manufacturer's recommendations, used the proper soldering temperatures, input resistance, etc. I've even been in correspondence with Avago, who claims it's properly set up, but they themselves have no idea why it isn't working correctly, and their demo board is a near copy of my circuit (different resistors, different input range). They have however given up on this issue and stopped responding. Does anyone have any insight as to what would cause this phenomenon? Thanks All![Test Circuit]1

Best Answer

This isolation amplifier is really designed for a very specific application- measurement of motor phase current on a VFD using a shunt. That's why it has a very low full scale voltage of +/-200mV (negating the need for an op-amp in that application), and why you need a ~55:1 attenuator for your application.

You may be able to use it for your ~+/-10V application but you will need to account for the input offset voltage and input offset voltage drift. The input-referred (to your divider) effect of the +/-3mV maximum offset voltage over temperature is about +/-170mV, so less than 2%, which is adequate for the intended application.

You can trim off the offset voltage at the output since the part is quite linear, however you cannot easily compensate for the drift with temperature, which is maximum +/-10uV/°C or about +/-550uV/°C at the input to your attenuator. So if you null it at 25°C and the temperature changes to 50°C that could represent as much as about 14mV change at the input to the attenuator.

That's still pretty good, and this is a pretty inexpensive isolation amplifier (due to the high volume application) so this is not a bad way to go if it meets your requirements, but each one would have to be individually nulled. This can be done digitally if the amplifier output is digitized by an MCU or with a trimpot with a suitable reference voltage and circuit. There are obvious advantages and disadvantages each way (for example, data stored in EEPROM has to be written there and it could be lost at some point in the life of the product).

Remember gain is delta-Vout/delta-Vin so you can't get accurate gain measurements by assuming Vout(0) = 0. It will be as much as +/-24mV differential voltage at the output.

Related Topic