INA381: 48V Common Mode Input

Steven,

Thanks for choosing TI! You are absolutely correct that you can use the topologies presented in SBOA198 to extend the Vcm of INA381 beyond its 26V maximum, but this does come with some challenges. The resistors placed in line on the sense line as you say will exhibit gain error on the device, as the gain is determined via internal resistance, and this resistance is placed in series with the internal resistor, fundamentally changing that gain. We publish a formula on page 24 of the datasheet that allows you to calculate the gain error you will generate based on the size resistor you choose to place in the sense line path. One way to alleviate this (although it comes with its own challenges in the area of mass production) is that you can perform a 2 point calibration on the device once the resistors are placed to calibrate the device to the "new" gain formed by the internal and external resistors. 

Hi, thanks for coming back. 

Are you able to confirm the internal resistors. I'm guessing it's 5k and 500k on the A3. Do you know the differential impedance? Is it 1250//5000?

I'm thinking I could use a divider of 2k and 1k to gnd. Think this gives me a large gain error but once confirmed with testing the new value can be calculated. That's why I'm thinking the A3 is a good option. 

I only need around 1A range on the 48v line and want the comparator for a overcurrent limit using a P-FET to turn off the load. 

Steven,

You are correct. Table 4 shows the value of Rint as 5kΩ and the corresponding feedback resistor should be 500kΩ to establish the correct gain. Regarding differential input impedance, I will reach out to our design team and see what we have that we can share. 

One issue that often arises with an approach such as this though is that for our processes, these resistors are laser trimmed to each other, rather than an absolute value. This means that from device to device, while centered around the values listed above, the actual resistances can fluctuate up to +/-20% of that value. This is why calibration is typically recommended for such as approach. 

Hi, I assume also that this will have a large impact on the comparator output I wish to use as over current cut off. If the values are not accurate to a value then each one will have a different trip point. 

I could correct current readings in software with calibration but not set point for comparator. Hmm this need re-evaluation

Steven, 

I agree, the internal fluctuations make it difficult to implement this kind of design when the design needs to scale. 

Can you share what current levels you are working with here, and what the application is? I may be able to recommend some additional options here but would like to know about your care-abouts here. 

Hi, I am trying to measure between 0.05 (0.1) to 1A on a 48V line and mainly using it as a latching over current. My biggest requirements are cost and availability to be honest. I will need around half a million chips in the next few years. 

Currently I'm setting the limit to be around 1.5A which +-20% should be fine.

It would be great to confirm the differential input impedance and also if the tolerance really is 20% on the actual value or hopefully tighter.  

I'm looking at using the A4 now as it is actually in stock currently and the gain error is around 70% which still scales well using a 50mR shunt. 

The INA200 has the common mode range but is about 7 times the price and isn't really in stock so completely ruled out.